Re: New discussion list.

Re: New discussion list.

[Robert Dewar]

<<For my own part, I think moving the discussion list to gnu.org is a
reasonable step.  It gives us a chance to try out gnu.org's quality
as a host, since unreliability is not a disaster (as it would be for,
say, CVS), and it's relatively easy to reverse if necessary.
>>

seems reasonable to me too.
I think the worry about the privacy policy is a red herring.

Re: New discussion list.

[Christopher Faylor]

On Wed, Nov 22, 2000 at 09:49:11PM -0500, Robert Dewar wrote:
><<For my own part, I think moving the discussion list to gnu.org is a
>reasonable step.  It gives us a chance to try out gnu.org's quality
>as a host, since unreliability is not a disaster (as it would be for,
>say, CVS), and it's relatively easy to reverse if necessary.
>>>
>
>seems reasonable to me too.
>I think the worry about the privacy policy is a red herring.

Perhaps it is.  How about this for a non-red-herring:

"I don't wanna."

I'm a volunteer who works on gdb.  I am one of the people who maintain
the mailing lists on this system.  I try very hard to be responsive to
people who make requests here.  I don't tend to be responsive to orders,
however, even when they are couched as ineffectively as this recent
email thread.

Personally, I do not understand the reasons for moving the mailing list.
I won't even pretend to ask what the technical reasons for the move are
because I know that there aren't any.

What you seem to want to do is move the mailing lists from one system
that is manned by people who believe in free software and are (or were)
proud to be associated with the FSF to another system with similar
people.  Technically speaking, the end result is a zero-sum game (unless
reports of unreliability of fencepost.gnu.org are true).  The effort
involved in moving is not "relatively easy", however, as has been
implied.

You are apparently expecting me, or some other volunteer sysadmin, to
stop whatever I'm doing and perform your bidding when we don't even,
officially, have any idea who you are.  Sure, I've know Stan Shebs.  The
last I heard he wasn't in a position to issue orders regarding gdb
mailing lists, however.

I have no, official, idea why Rober Dewar and Todd Whitesel are suddenly
making proclamations about the future of gdb and its mailing lists,
though.  What a botched job.

You guys realize that you are discussing this in a semi-public mailing
list and that many people probably have absolutely no idea who you are
and why you're making decisions about gdb, right?

Is this really your first public unveiling?  A cc'ed message with vague
mumblings about moving a mailing list?

AFAICT, you are representing the GDB Steering Committee (or is it the
GDB Maintainers Committee?) Apparently this committe has has been formed
as a mechanism for issuing edicts about gdb without the benefit of
public discussion.

Is this what we can expect for future pronouncements from the steering
committee?  Will people who are actually in the trenches doing something
be expected to drop everything when they get a Cc'ed email message with
hazy musings from Stan Shebs?

You probably ought to take a stroll over to the gcc mailing list where
the concept of steering committee openness has recently been discussed.
The GCC steering committee apparently values openness and seems to
be able to communicate its desires effectively.  I hope that you will
be able to learn from their example.

The only thing I know for sure right now, however, from this limited
exposure, is that you seem to be ineffective, email-challenged
communicators.  That is neither a propitious nor an auspicious debut for
your little cabal, IMO.  Of course, maybe a cabal doesn't need to worry
about public image.

So, can you take your discussion to your secret gdbgroins mailing list
and stop discussing things here?  Once you've reached a clear decision,
maybe you can ask Andrew Cagney to put it into words for you.  Hopefully
he will be able to provide slightly more rationale for whatever you are
pronouncing.  For the record, an self-congratulatory agreements with
each other that your clever plan is "a reasonable step" is neither a
rationale nor a rallying cry for getting people to help you.

cgf

Re: New discussion list.

[Stan Shebs]

Christopher Faylor wrote:
> 
> I'm a volunteer who works on gdb.  I am one of the people who maintain
> the mailing lists on this system.  I try very hard to be responsive to
> people who make requests here.  I don't tend to be responsive to orders,
> however, even when they are couched as ineffectively as this recent
> email thread.

Whoa, a bit of tense reaction here!  Perhaps the "orders" seem to be
couched ineffectively because they're not actually orders?  I purposely
cc'ed overseers on this discussion, so as to get input from relevant
and knowledgeable people, not to disseminate demands.

> Personally, I do not understand the reasons for moving the mailing list.
> I won't even pretend to ask what the technical reasons for the move are
> because I know that there aren't any.

That's right.  As you know, GNU is not a purely technical project;
there is a political aspect to it as well.  I've had occasion to
think about this many times over the years, starting with when I
ported GCC to the Mac while the FSF was officially boycotting Apple,
which meant that I was entirely on my own, and pissed upon if I
even dared to publicly mention what I was doing.  Even so, I think
the GNU project is right to take political stances; there are a
great many ways for corporations to take advantage of and to co-opt
free software projects, and now that "open source" is fashionable,
there are a lot more people thinking about ways to exploit us.  While
I don't always agree with RMS' political decisions, I do believe
that it is valid to include political issues in decisionmaking for GNU.

> I have no, official, idea why Rober Dewar and Todd Whitesel are suddenly
> making proclamations about the future of gdb and its mailing lists,
> though.  What a botched job.
> 
> You guys realize that you are discussing this in a semi-public mailing
> list and that many people probably have absolutely no idea who you are
> and why you're making decisions about gdb, right?

Well, we have a bit of chicken-and-egg problem, in that we want the
committee's announcement to include a public contact address, and
the consensus was to have it be the regular GDB address.  RMS pointed
out the problem with having the Red Hat address embedded in his
official announcement, but the usual way to discuss the handling of
mailing lists is via the semi-public overseers.

> Is this really your first public unveiling?  A cc'ed message with vague
> mumblings about moving a mailing list?
> 
> AFAICT, you are representing the GDB Steering Committee (or is it the
> GDB Maintainers Committee?) Apparently this committe has has been formed
> as a mechanism for issuing edicts about gdb without the benefit of
> public discussion.

As you may remember, it's been discussed in public before, indeed
last summer I put out a call for people interested in participating.

> Is this what we can expect for future pronouncements from the steering
> committee?  Will people who are actually in the trenches doing something
> be expected to drop everything when they get a Cc'ed email message with
> hazy musings from Stan Shebs?

Well, they would at least be expected not to flame - semi-publicly! -
about it.  The argument for hosting stuff at Red Hat is that the
service is good and reliable, but if attempts to include more Red Hat
people in a discussion are met with vitriol and insults, that is
actually a strong argument against hosting things there.

> You probably ought to take a stroll over to the gcc mailing list where
> the concept of steering committee openness has recently been discussed.
> The GCC steering committee apparently values openness and seems to
> be able to communicate its desires effectively.  I hope that you will
> be able to learn from their example.

As it happens, I've been on the GCC list for nearly a year now, and
on a number of occasions we get an announcement beginning with "The
committee has decided ..." and it's pretty clear there has been an
extensive private discussion leading up to it.  I'm OK with that;
I know that I personally don't want to have every bit of my
thoughts and writing out in public, and I don't think I'm unusual
in that.  Part of the reason is that sometimes I come up with
boneheaded ideas, and I want to be able to bring them up without
getting toasted by a hostile audience.  Indeed, had I not come to
expect a generally rational reception from overseers, I would not
have cc'ed that list.  I guess I was wrong.

Stan

Re: New discussion list.

[Christopher Faylor]

[Ok.  I've calmed down now.]
On Wed, Nov 22, 2000 at 09:00:14PM -0800, Stan Shebs wrote:
>Christopher Faylor wrote:
>> I'm a volunteer who works on gdb.  I am one of the people who maintain
>> the mailing lists on this system.  I try very hard to be responsive to
>> people who make requests here.  I don't tend to be responsive to orders,
>> however, even when they are couched as ineffectively as this recent
>> email thread.
>
>Whoa, a bit of tense reaction here!  Perhaps the "orders" seem to be
>couched ineffectively because they're not actually orders?  I purposely
>cc'ed overseers on this discussion, so as to get input from relevant
>and knowledgeable people, not to disseminate demands.

I don't think that this was clear from your email.  AFAICT, you were
letting us in on a private conversation with some FSF employee,
discussing the best ways for moving the mailing list with no background
or rationale for the move.

It was effectively like walking into my house wearing a hard hat and
carrying a clipboard, discussing the best ways to level the place on
your cell phone and then pausing to ask me "Do you have any ideas?"

>>Personally, I do not understand the reasons for moving the mailing
>>list.  I won't even pretend to ask what the technical reasons for the
>>move are because I know that there aren't any.
>
>That's right.  As you know, GNU is not a purely technical project;
>there is a political aspect to it as well.  I've had occasion to think
>about this many times over the years, starting with when I ported GCC
>to the Mac while the FSF was officially boycotting Apple, which meant
>that I was entirely on my own, and pissed upon if I even dared to
>publicly mention what I was doing.

Then perhaps you can understand my strong reaction to essentially being
asked to do something that I very much do not agree with.

>Even so, I think the GNU project is right to take political stances;
>there are a great many ways for corporations to take advantage of and
>to co-opt free software projects, and now that "open source" is
>fashionable, there are a lot more people thinking about ways to exploit
>us.  While I don't always agree with RMS' political decisions, I do
>believe that it is valid to include political issues in decisionmaking
>for GNU.

What is the political stance?  Red Hat is evil?  Red Hat is claiming to
own the FSF?  What?  I am not privy to the private discussions that you
must have had with RMS.  Maybe other people here are more aware of
this.

I am aware that there is an anti-Red Hat sentiment.  Why not make it
perfectly clear so that we can all understand it.

It is a good idea to move the mailing lists, because... 

Red Hat is claiming ownership of gdb?

Red Hat released an unsupported version of gcc?

Red Hat claims to have invented free software?

Red Hat is a corporation that sometimes does stupid things and must
therefore be punished?

I guess it is probably all of the above and that means that people will
now effectively be "punished" (pissed upon?) as you were punished for
working on Apple-related software.  I can understand that there must be
some real moral indignation there somewhere for this kind of decision to
take place but I certainly don't have to agree with it or like it.

>> You guys realize that you are discussing this in a semi-public mailing
>> list and that many people probably have absolutely no idea who you are
>> and why you're making decisions about gdb, right?
>
>Well, we have a bit of chicken-and-egg problem, in that we want the
>committee's announcement to include a public contact address, and
>the consensus was to have it be the regular GDB address.  RMS pointed
>out the problem with having the Red Hat address embedded in his
>official announcement, but the usual way to discuss the handling of
>mailing lists is via the semi-public overseers.

So why not provide a slight amount of background?  Do you think that
adding overseers@sources.redhat.com to an ongoing discussion with
someone is an effective technique for accomplishing your goals?  Maybe
I'm wrong but I don't think that everyone who reads this list was aware
of the existence of this committee or your role in it.

I was just barely aware of what was going on, myself until recently.  I
thought you were still talking about what to call yourselves and whether
people who had responsibility for gdb source code were helpers,
developers, contributors, or maintainers.

>As you may remember, it's been discussed in public before, indeed last
>summer I put out a call for people interested in participating.

Sure I remember.  I'm very familiar with the intent.  Was that somehow clear
from any of the posts here?  Was the 'gdbheads' email address supposed to
adequately convey the gravity of your position?

>> Is this what we can expect for future pronouncements from the steering
>> committee?  Will people who are actually in the trenches doing something
>> be expected to drop everything when they get a Cc'ed email message with
>> hazy musings from Stan Shebs?
>
>Well, they would at least be expected not to flame - semi-publicly! -
>about it.

Hmm.  I don't get this.  Are you saying that if I hadn't used the term
"hazy musings" you would be inducting me into the steering committee?
Damn.  So close.

>The argument for hosting stuff at Red Hat is that the service is good
>and reliable, but if attempts to include more Red Hat people in a
>discussion are met with vitriol and insults, that is actually a strong
>argument against hosting things there.

Ah.  So you are labelling me as a "Red Hat employee" and therefore using
me as YA reason to label Red Hat as inappropriate for hosting any FSF
software.  Hopefully no more Red Hat employees will enter the fray
slinging vitriol because that will offer still more reasons why it is an
excellent reason to move all services from sources.redhat.com.  I mean,
if people are obviously irritated while still providing good technical
services then that's a good political reason to move.  We can't
associate with a company that has people with stong opinions, can we?

FWIW, I'm an individual.  I am very very far from being either a
corporate shill and even farther from being a corporate spokesman.  I
have, in fact, tried hard to do whatever it takes to clean up the Red
Hat site to avoid the appearance of Red Hat owning anything that truly
belongs to the FSF.  I've even participated in a draft of a document
that will be sent to Red Hat internally detailing exactly how wrong it
is to crow about any kind of "ownership" of anything like gcc or gdb.

I am also a programmer who detests having to be uprooted and potentially
enjoined into making changes that make no technical sense.

>> You probably ought to take a stroll over to the gcc mailing list where
>> the concept of steering committee openness has recently been discussed.
>> The GCC steering committee apparently values openness and seems to
>> be able to communicate its desires effectively.  I hope that you will
>> be able to learn from their example.
>
>As it happens, I've been on the GCC list for nearly a year now, and
>on a number of occasions we get an announcement beginning with "The
>committee has decided ..." and it's pretty clear there has been an
>extensive private discussion leading up to it.  I'm OK with that;
>I know that I personally don't want to have every bit of my
>thoughts and writing out in public, and I don't think I'm unusual
>in that.  Part of the reason is that sometimes I come up with
>boneheaded ideas, and I want to be able to bring them up without
>getting toasted by a hostile audience.  Indeed, had I not come to
>expect a generally rational reception from overseers, I would not
>have cc'ed that list.  I guess I was wrong.

Perhaps if you had provided some background and not discussed this as if
it was all a done deal, carried out in a back room, with no hope for
discussion, I would not have reacted so strongly.  I knew that the
potential for moving the gdb list has been discussed for a year or more.
I was nonetheless surprised to see your email here which seemed to be
discussing it as a fait accompli.

I'm sure that Andrew would have notified us internally of this decision
but I suspect that it was made while he was enjoying the benefits of
being in another time zone.

However, on rereading my email, I said some things that were over the
line and I apologize for that.

Anyway, I know that I am not going to convince anyone of anything.  I
can only represent this view: As a programmer I do not want to be
bothered by having to change all of my bookmarks, my mail filters, and,
eventually, I suppose, my CVS repository.

As a sysadmin, I dread the thought of moving the mailing lists, the web
page links, the mail archives, the ftp archives, the cgi scripts,
and the CVS repository, and I am discouraged to think that the effort
that I have been putting into getting a gdb test results page up and
running is all for naught.

I could accept any of these if I understood the rationale.  The best that
I can figure from your statements above is that Red Hat is being viewed
as having an agenda that does not jive with the FSF and rather than
allowing us to try to address that, you are taking all of your marbles
and going home.

That is certainly within the right of the FSF.  I just wish that someone
could offer an explanation that would satisfy me, but I know that there
is small chance of that.

If this decision does come to pass, I'll participate in the move.  I
consider that my job on sources.redhat.com and I'm not going to shirk it
because I disagree with an FSF decision.  I should have made that clear
in my first message.

And, I again apologize for any harshness in my previous message.

cgf

Re: New discussion list. (APOLOGY)

[Christopher Faylor]

Several people have pointed out to me very politely in email that some
of my remarks here were over the top.  On rereading my email, I am
forced to agree.  Way too much flamage and not enough substance...

I apologize to Stan and anyone else who was offended and I am grateful
that everyone resisted the urge to reply in kind.

I will try to do better in the future.  :-)

cgf

Re: New discussion list.

[Jason Molenda]

Just to interject one point - Chris was correct in noting that
there is real work involved in a clean list move.  You need a lot
of background knowledge to know what you're doing, and even then
you've got to check over a host of loose ends to do it right.  It
is not something to be ventured lightly, it is not something you
would do to see how it goes, and maybe then just switch it back if
it isn't everything you'd hoped for.

Before anyone comes back to me saying, "Hey, my neighbor Joe says
he can move a mailing list in three minutes using just two shell
commands, a rubber band, and his pet rabbit," there's a world of
difference between a clean switchover where the users are pampered
all along the way and one that causes everyone a lot of grief.  It's
easy to do a minimal job, it's real work to do a great one.  This is
true of most things on sourceware - if you really look at the details,
you'll see that the infrastructure on sourceware is better thought
out than on nearly any other site on the net.


If a GDB steering committee really, actually exists (last I heard
they were busily taking a couple of years deciding whether they
were going to exist and whether anyone was going to actually do
anything besides complain about the Cygnus^H^H^H^H^HRed Hat hegemony),
and if that steering committee says that the list is going to move
to gnu.org, then I guess that's their business.  It'll be unfortunate
if inferior support and services are provided to gdb net developers
in the name of having everything end with a "gnu.org" domain name.

If the real priority is to purge the dreaded 'redhat.com' address,
has anyone considered mimicing the GCC location style (gdb.gnu.org)?
I don't know if RH would even be interested in providing those
services (believe me, it's a nontrivial amount of work to get it
all right), but if _I_ were on the GDB steering commitee and _I_
was trying to promote GDB development, and I had some great
infrastructure services being provided to me for free, I'd try to
avail myself of those services as much as possible.

Jason

Re: New discussion list.

[Andrew Cagney]

Stan Shebs wrote:

> Well, we have a bit of chicken-and-egg problem, in that we want the
> committee's announcement to include a public contact address, and
> the consensus was to have it be the regular GDB address.

I don't know if that is exactly correct.  My last unresponded posting
was:

> I don't think these two tasks [announcement and mailing list change]
> should be tied together.  If the mailing
> list name is of concern, then perhaps just leave it out.

Remind me next time to spell out the consequences.

	Andrew

Re: New discussion list.

[Stan Shebs]

Christopher Faylor wrote:
> 
> [...]  AFAICT, you were
> letting us in on a private conversation with some FSF employee,
> discussing the best ways for moving the mailing list with no background
> or rationale for the move.
> 
> It was effectively like walking into my house wearing a hard hat and
> carrying a clipboard, discussing the best ways to level the place on
> your cell phone and then pausing to ask me "Do you have any ideas?"

Sorry, I thought everybody would remember that this issue had
come up before.  Indeed, I remember it being discussed internally
when I was at Cygnus.

> What is the political stance?  Red Hat is evil?  Red Hat is claiming to
> own the FSF?  What?  I am not privy to the private discussions that you
> must have had with RMS.  Maybe other people here are more aware of
> this.

I don't think the basic stance is any secret - it's important that
the GNU project be seen as independent, and not controlled by any
corporate entity.  RMS has been consistently very clear on this.
It's nothing particular against Red Hat, it just affects RH more
because of its important role in GNU development.

> I am aware that there is an anti-Red Hat sentiment.  Why not make it
> perfectly clear so that we can all understand it.

There can hardly be any anti-Red Hat sentiment from me, for one
thing I still own stock in it!  There are GDB committee members
whose companies compete with RH though, probably hard for them
not to think a negative thought about RH from time to time. :-)
 
> It is a good idea to move the mailing lists, because...
> [ evil deeds :-) ]

At the risk of playing psychoanalyst, I think you're seeing enemies
that don't exist.  The goal of having gnu.org host GNU development
has existed for several years, long before the events you're
referring to.  You could make an argument that gnu.org was not a
good place to host development, and indeed I myself have made that
argument in the past, and RMS' commitment has been that he will
make sure that it is a good place.  This issue came up in GDB
committee discussions, and when I was at Cygnus, I sent out
committee meeting reports saying that this was the long-term goal,
but that it was contingent upon gnu.org being adequate to the task.

> So why not provide a slight amount of background?  Do you think that
> adding overseers@sources.redhat.com to an ongoing discussion with
> someone is an effective technique for accomplishing your goals?  Maybe
> I'm wrong but I don't think that everyone who reads this list was aware
> of the existence of this committee or your role in it.

How quickly they forget... :-) I know that this was an active topic of
discussion when I was at Cygnus, both verbally and in email.  (Hmmm,
hard to believe a year has gone by already!)

> I was just barely aware of what was going on, myself until recently.  I
> thought you were still talking about what to call yourselves and whether
> people who had responsibility for gdb source code were helpers,
> developers, contributors, or maintainers.

Heh-heh.

> We can't
> associate with a company that has people with stong opinions, can we?

I don't mind strong opinions, indeed I expect those from the
wise overseers, but I was a little taken aback by the more personal
stuff ("gdbgroin"?).

> FWIW, I'm an individual.  I am very very far from being either a
> corporate shill and even farther from being a corporate spokesman.  I
> have, in fact, tried hard to do whatever it takes to clean up the Red
> Hat site to avoid the appearance of Red Hat owning anything that truly
> belongs to the FSF.  I've even participated in a draft of a document
> that will be sent to Red Hat internally detailing exactly how wrong it
> is to crow about any kind of "ownership" of anything like gcc or gdb.

That's good!  The trick is to convince the rest of the steering
committee though, who have so far been more enthusiastic about
moving to gnu.org than staying on sources.redhat.com.

> Perhaps if you had provided some background and not discussed this as if
> it was all a done deal, carried out in a back room, with no hope for
> discussion, I would not have reacted so strongly.  I knew that the
> potential for moving the gdb list has been discussed for a year or more.
> I was nonetheless surprised to see your email here which seemed to be
> discussing it as a fait accompli.

The discussion leading up to my cc did not feature Andrew or Jim
saying "no way, over our inanimate corpses", or else I never got
that message. :-)

> I'm sure that Andrew would have notified us internally of this decision
> but I suspect that it was made while he was enjoying the benefits of
> being in another time zone.

The dating on the messages I have shows that Andrew's last message
on the subject showed up in my mbox 17 minutes before your first
message, and he was talking about a possible privacy issue, not
whether the move was desirable in the first place, so I don't think
it was too much of a leap to believe that Red Hat folks were up to
date on things.

> However, on rereading my email, I said some things that were over the
> line and I apologize for that.

Apology accepted.

> Anyway, I know that I am not going to convince anyone of anything.  I
> can only represent this view: As a programmer I do not want to be
> bothered by having to change all of my bookmarks, my mail filters, and,
> eventually, I suppose, my CVS repository.
> 
> As a sysadmin, I dread the thought of moving the mailing lists, the web
> page links, the mail archives, the ftp archives, the cgi scripts,
> and the CVS repository, and I am discouraged to think that the effort
> that I have been putting into getting a gdb test results page up and
> running is all for naught.

Now *this* is all good stuff to bring up.  The FSF position is that
the perception of independence is worth the trouble.  Personally,
I think it's possible to achieve that perception without moving
everything physically - sourceforge seems like a pretty successful
example.  But in committee discussion, persons in favor in staying
on sources.redhat.com have been lukewarm, while persons in favor
of moving to gnu.org have been pretty determined about that.  In the
year-and-some that the proto-committee has been in existence, no
one has yet made a strong and compelling argument for why and how to
keep GDB stuff on sources.redhat.com, and yet preserve the independence
associated with gnu.org.  Perhaps you need to be on the committee??

Stan

Re: New discussion list.

[Stan Shebs]

Andrew Cagney wrote:
> 
> Stan Shebs wrote:
> 
> > Well, we have a bit of chicken-and-egg problem, in that we want the
> > committee's announcement to include a public contact address, and
> > the consensus was to have it be the regular GDB address.
> 
> I don't know if that is exactly correct.  My last unresponded posting
> was:
> 
> > I don't think these two tasks [announcement and mailing list change]
> > should be tied together.  If the mailing
> > list name is of concern, then perhaps just leave it out.

Argh, you're right, I did see that message, but my brain
spasmed.  Sorry sorry sorry.

Yes, it's more important to get the committee announcement out
in a timely fashion, so we don't have to endure any more jokes
about being a lame committee wannabe.

Stan

Re: New discussion list.

[Andrew Cagney]

Jason Molenda wrote:

> If the real priority is to purge the dreaded 'redhat.com' address,
> has anyone considered mimicing the GCC location style (gdb.gnu.org)?
> I don't know if RH would even be interested in providing those
> services (believe me, it's a nontrivial amount of work to get it
> all right), but if _I_ were on the GDB steering commitee and _I_
> was trying to promote GDB development, and I had some great
> infrastructure services being provided to me for free, I'd try to
> avail myself of those services as much as possible.


[As an individual that happens to be on the GDB Committee and previously
happend to co-ordinate GDB development]

I've on a number of occasions put this position forward and consistently
had it rejected by Richard Stallman.  I don't believe that I've evern
been given a clear technical argument against http://gdb.gnu.org .

I've also gone to the effort of finding out how hard such a move would
be vis:

http://sources.redhat.com/ml/overseers/2000-q3/msg00010.html


[As, I guess a representative of Red Hat]

Red Hat don't have any problems with gdb.gnu.org.

	Andrew

Re: New discussion list.

[Andrew Cagney]

Stan Shebs wrote:

> Now *this* is all good stuff to bring up.  The FSF position is that
> the perception of independence is worth the trouble.  Personally,
> I think it's possible to achieve that perception without moving
> everything physically - sourceforge seems like a pretty successful
> example.  But in committee discussion, persons in favor in staying
> on sources.redhat.com have been lukewarm, while persons in favor
> of moving to gnu.org have been pretty determined about that.  In the
> year-and-some that the proto-committee has been in existence, no
> one has yet made a strong and compelling argument for why and how to
> keep GDB stuff on sources.redhat.com, and yet preserve the independence
> associated with gnu.org.  Perhaps you need to be on the committee??

Stan,

I don't think that is a fair representation of the discussion on this
matter.  Only two people were actually participating in the debate.

For what it is worth, my last posting on the matter is below.  Would you
have prefered it if I'd used my native Australian vernacular? :-)

> > [ ... move the GDB mailing list now ... ]
> 
> To move the mailing lists (there are many) and all the associated
> services (mail archives, bug tracking DB, web pages, man pages,
> testresults, ...) provided by the Red Hat (nee Cygnus) machine to the
> machines gnu.org/mail.gnu.org/* isn't a trivial problem.  The logistics
> of such a move will require significant time and other resources.  My
> preference is for those resources to be spent addressing what I consider
> to be other more important issues, such as GDB 5.1 and the GCC 3.0 v3
> G++ ABI.
> 
> Consequently, I again strongly recommend the more incremental approach
> adopted by the GCC group.  I've been handed an estimate of ``less than a
> day'' to establish the basics (including a working gdb@gdb.gnu.org
> et.al. e-mail addresses and the http://gdb.gnu.org URL).
> 
> However, regardless of the decision, I'd prefer to see the clear
> resolution of this matter by the GDB committee.
> 
> Finally, I'd also prefer it if the announcement of the committee wasn't
> delayed by this issue.  If the mailing list's domain name really is such
> a problem then either specify the existing ``bug-gdb@gnu.org'' or omit
> the mailing list entirely.

There were two further postings following this.

	o	[among other things]
		A request that I co-operate with
		the FSF person that will work on this.

	o	An announcement that a new list
		gdb@gnu.org had been created.

		(I wasn't contacted).

There never was a clear resolution of this matter by the committee.

	Andrew

Re: New discussion list.

[Stan Shebs]

Andrew Cagney wrote:
> 
> Stan Shebs wrote:
> 
> > Now *this* is all good stuff to bring up.  The FSF position is that
> > the perception of independence is worth the trouble.  Personally,
> > I think it's possible to achieve that perception without moving
> > everything physically - sourceforge seems like a pretty successful
> > example.  But in committee discussion, persons in favor in staying
> > on sources.redhat.com have been lukewarm, while persons in favor
> > of moving to gnu.org have been pretty determined about that.  In the
> > year-and-some that the proto-committee has been in existence, no
> > one has yet made a strong and compelling argument for why and how to
> > keep GDB stuff on sources.redhat.com, and yet preserve the independence
> > associated with gnu.org.  Perhaps you need to be on the committee??
> 
> Stan,
> 
> I don't think that is a fair representation of the discussion on this
> matter.  Only two people were actually participating in the debate.

I was referring to the aggregate of the discussion that has happened
since our first meeting in May 1999.  Since the committee has to work
by consensus, all it would take is one person saying "No, I can't
accept the move to gnu.org".  As far as I remember, no one has ever
made that statement, while at least one person has said "I can't accept
the status quo".  If you (meaning any committee member) really don't
want to move anything to gnu.org, you need to say it in so many words.

> There never was a clear resolution of this matter by the committee.

Clearly we need more discussion, and given that many of the USAians
are incommunicado this weekend, we shouldn't be trying to take any
action until we hear from more people.  We've also gotten some
valuable technical input from Jason and Chris (thanks!), and I hope
the committee will take heed of that when deciding what to do and
when to do it.  I also want to thank the overseers list for putting
up with this intrusion, apologize for the distraction, and undertake
to make the next communication more useful.

Stan

Moving Projects

[Gerald Pfeifer]

On Thu, 23 Nov 2000, Andrew Cagney wrote:
> [As an individual that happens to be on the GDB Committee and previously
> happend to co-ordinate GDB development]

As an individual that happens to be on the GCC Committee and overseers,
and has some of experience with the current setup, the following analysis
that I've written some time ago and updated now might prove useful.

This is neither pro nor con doing such a move, just a(n incomplete)
list of issues that have to be considered and resolved *before* moving
gcc.gnu.org or GDB or some similiar project to a different machine.

MAILING LIST PERFORMANCE: gcc.gnu.org (with the current setup) performs
  extremely well, delivering messages within seconds. One has to verify
  that the new machines will sustain this under heavy load as well and
  not bring us back to the state where we had delays of several hours up
  to even a day.

SPAM: The gcc.gnu.org machine and all mailing lists as well as accounts
  there have been well secured against spam and many developers rely on
  that, while the gnu.org machines do not filter spamming servers.

  (For example one of our more well-known contributors had his regular
  GNU account closed because of the vast amounts of spam he received
  there. We do not want that to happen.)

BACKUP: Are the new machines properly backuped? How often? How?

GNATS DATABASES: Do these exist on the new machines? Are all interfaces
  (gnatsweb, directly, mail2gnats) available? Can the old databases be
  moved?

MAILING LIST ARCHIVE: This is especially tricky to handle correctly
  such that we do not lose messages.

  Specifically, the move should not be disruptive in that we do not want
  to split between old and new archives.

  Also mbox files of these archives should be kept and continued to
  be generated.

SEARCH ENGINE: Does it already exist on the new machine?

  Can the current reconfiguration be reused? We have some special settings.

  Who will take care of it regularily? Such a search engine needs a bit
  of hand-holding now and then.

ACCOUNT DATA: Obviously tons of account data will have to be moved.
  We are talking about one-hundred accounts here. This needs a bit of
  cooperation and planning, but shouldn't be hard per se.

RSYNC: Does the new server support rsync? This important to distribute
  the repository.

MOST SERVICES ARE RUNNING ASYNCHRONOUSLY: This means that it is critical
  to move services in the right order and with some short-time measures.

Overall, I feel that moving is by far less trivial than many might
believe, and I'm sure the list above is not complete. This does not
mean that such a move can or should not be done in principle, it just
means that we/you need some volunteers that are willing to do that.

Keep or just cloning the current box/setup, and perhaps relocating it
under a new name like sources.gnu.org with aliases gcc.gnu.org, gdb.gnu.org
might make sense.

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/

Re: New discussion list.

[Andrew Cagney]

Stan Shebs wrote:

> Now *this* is all good stuff to bring up.  The FSF position is that
> the perception of independence is worth the trouble.  Personally,
> I think it's possible to achieve that perception without moving
> everything physically - sourceforge seems like a pretty successful
> example.  But in committee discussion, persons in favor in staying
> on sources.redhat.com have been lukewarm, while persons in favor
> of moving to gnu.org have been pretty determined about that.  In the
> year-and-some that the proto-committee has been in existence, no
> one has yet made a strong and compelling argument for why and how to
> keep GDB stuff on sources.redhat.com, and yet preserve the independence
> associated with gnu.org.  Perhaps you need to be on the committee??

Stan,

I don't think that is a fair representation of the discussion on this
matter.  Only two people were actually participating in the debate.

For what it is worth, my last posting on the matter is below.  Would you
have prefered it if I'd used my native Australian vernacular? :-)

> > [ ... move the GDB mailing list now ... ]
> 
> To move the mailing lists (there are many) and all the associated
> services (mail archives, bug tracking DB, web pages, man pages,
> testresults, ...) provided by the Red Hat (nee Cygnus) machine to the
> machines gnu.org/mail.gnu.org/* isn't a trivial problem.  The logistics
> of such a move will require significant time and other resources.  My
> preference is for those resources to be spent addressing what I consider
> to be other more important issues, such as GDB 5.1 and the GCC 3.0 v3
> G++ ABI.
> 
> Consequently, I again strongly recommend the more incremental approach
> adopted by the GCC group.  I've been handed an estimate of ``less than a
> day'' to establish the basics (including a working gdb@gdb.gnu.org
> et.al. e-mail addresses and the http://gdb.gnu.org URL).
> 
> However, regardless of the decision, I'd prefer to see the clear
> resolution of this matter by the GDB committee.
> 
> Finally, I'd also prefer it if the announcement of the committee wasn't
> delayed by this issue.  If the mailing list's domain name really is such
> a problem then either specify the existing ``bug-gdb@gnu.org'' or omit
> the mailing list entirely.

There were two further postings following this.

	o	[among other things]
		A request that I co-operate with
		the FSF person that will work on this.

	o	An announcement that a new list
		gdb@gnu.org had been created.

		(I wasn't contacted).

There never was a clear resolution of this matter by the committee.

	Andrew

Re: New discussion list.

[Andrew Cagney]

Jason Molenda wrote:

> If the real priority is to purge the dreaded 'redhat.com' address,
> has anyone considered mimicing the GCC location style (gdb.gnu.org)?
> I don't know if RH would even be interested in providing those
> services (believe me, it's a nontrivial amount of work to get it
> all right), but if _I_ were on the GDB steering commitee and _I_
> was trying to promote GDB development, and I had some great
> infrastructure services being provided to me for free, I'd try to
> avail myself of those services as much as possible.


[As an individual that happens to be on the GDB Committee and previously
happend to co-ordinate GDB development]

I've on a number of occasions put this position forward and consistently
had it rejected by Richard Stallman.  I don't believe that I've evern
been given a clear technical argument against http://gdb.gnu.org .

I've also gone to the effort of finding out how hard such a move would
be vis:

http://sources.redhat.com/ml/overseers/2000-q3/msg00010.html


[As, I guess a representative of Red Hat]

Red Hat don't have any problems with gdb.gnu.org.

	Andrew

Re: New discussion list.

[Christopher Faylor]

[Ok.  I've calmed down now.]
On Wed, Nov 22, 2000 at 09:00:14PM -0800, Stan Shebs wrote:
>Christopher Faylor wrote:
>> I'm a volunteer who works on gdb.  I am one of the people who maintain
>> the mailing lists on this system.  I try very hard to be responsive to
>> people who make requests here.  I don't tend to be responsive to orders,
>> however, even when they are couched as ineffectively as this recent
>> email thread.
>
>Whoa, a bit of tense reaction here!  Perhaps the "orders" seem to be
>couched ineffectively because they're not actually orders?  I purposely
>cc'ed overseers on this discussion, so as to get input from relevant
>and knowledgeable people, not to disseminate demands.

I don't think that this was clear from your email.  AFAICT, you were
letting us in on a private conversation with some FSF employee,
discussing the best ways for moving the mailing list with no background
or rationale for the move.

It was effectively like walking into my house wearing a hard hat and
carrying a clipboard, discussing the best ways to level the place on
your cell phone and then pausing to ask me "Do you have any ideas?"

>>Personally, I do not understand the reasons for moving the mailing
>>list.  I won't even pretend to ask what the technical reasons for the
>>move are because I know that there aren't any.
>
>That's right.  As you know, GNU is not a purely technical project;
>there is a political aspect to it as well.  I've had occasion to think
>about this many times over the years, starting with when I ported GCC
>to the Mac while the FSF was officially boycotting Apple, which meant
>that I was entirely on my own, and pissed upon if I even dared to
>publicly mention what I was doing.

Then perhaps you can understand my strong reaction to essentially being
asked to do something that I very much do not agree with.

>Even so, I think the GNU project is right to take political stances;
>there are a great many ways for corporations to take advantage of and
>to co-opt free software projects, and now that "open source" is
>fashionable, there are a lot more people thinking about ways to exploit
>us.  While I don't always agree with RMS' political decisions, I do
>believe that it is valid to include political issues in decisionmaking
>for GNU.

What is the political stance?  Red Hat is evil?  Red Hat is claiming to
own the FSF?  What?  I am not privy to the private discussions that you
must have had with RMS.  Maybe other people here are more aware of
this.

I am aware that there is an anti-Red Hat sentiment.  Why not make it
perfectly clear so that we can all understand it.

It is a good idea to move the mailing lists, because... 

Red Hat is claiming ownership of gdb?

Red Hat released an unsupported version of gcc?

Red Hat claims to have invented free software?

Red Hat is a corporation that sometimes does stupid things and must
therefore be punished?

I guess it is probably all of the above and that means that people will
now effectively be "punished" (pissed upon?) as you were punished for
working on Apple-related software.  I can understand that there must be
some real moral indignation there somewhere for this kind of decision to
take place but I certainly don't have to agree with it or like it.

>> You guys realize that you are discussing this in a semi-public mailing
>> list and that many people probably have absolutely no idea who you are
>> and why you're making decisions about gdb, right?
>
>Well, we have a bit of chicken-and-egg problem, in that we want the
>committee's announcement to include a public contact address, and
>the consensus was to have it be the regular GDB address.  RMS pointed
>out the problem with having the Red Hat address embedded in his
>official announcement, but the usual way to discuss the handling of
>mailing lists is via the semi-public overseers.

So why not provide a slight amount of background?  Do you think that
adding overseers@sources.redhat.com to an ongoing discussion with
someone is an effective technique for accomplishing your goals?  Maybe
I'm wrong but I don't think that everyone who reads this list was aware
of the existence of this committee or your role in it.

I was just barely aware of what was going on, myself until recently.  I
thought you were still talking about what to call yourselves and whether
people who had responsibility for gdb source code were helpers,
developers, contributors, or maintainers.

>As you may remember, it's been discussed in public before, indeed last
>summer I put out a call for people interested in participating.

Sure I remember.  I'm very familiar with the intent.  Was that somehow clear
from any of the posts here?  Was the 'gdbheads' email address supposed to
adequately convey the gravity of your position?

>> Is this what we can expect for future pronouncements from the steering
>> committee?  Will people who are actually in the trenches doing something
>> be expected to drop everything when they get a Cc'ed email message with
>> hazy musings from Stan Shebs?
>
>Well, they would at least be expected not to flame - semi-publicly! -
>about it.

Hmm.  I don't get this.  Are you saying that if I hadn't used the term
"hazy musings" you would be inducting me into the steering committee?
Damn.  So close.

>The argument for hosting stuff at Red Hat is that the service is good
>and reliable, but if attempts to include more Red Hat people in a
>discussion are met with vitriol and insults, that is actually a strong
>argument against hosting things there.

Ah.  So you are labelling me as a "Red Hat employee" and therefore using
me as YA reason to label Red Hat as inappropriate for hosting any FSF
software.  Hopefully no more Red Hat employees will enter the fray
slinging vitriol because that will offer still more reasons why it is an
excellent reason to move all services from sources.redhat.com.  I mean,
if people are obviously irritated while still providing good technical
services then that's a good political reason to move.  We can't
associate with a company that has people with stong opinions, can we?

FWIW, I'm an individual.  I am very very far from being either a
corporate shill and even farther from being a corporate spokesman.  I
have, in fact, tried hard to do whatever it takes to clean up the Red
Hat site to avoid the appearance of Red Hat owning anything that truly
belongs to the FSF.  I've even participated in a draft of a document
that will be sent to Red Hat internally detailing exactly how wrong it
is to crow about any kind of "ownership" of anything like gcc or gdb.

I am also a programmer who detests having to be uprooted and potentially
enjoined into making changes that make no technical sense.

>> You probably ought to take a stroll over to the gcc mailing list where
>> the concept of steering committee openness has recently been discussed.
>> The GCC steering committee apparently values openness and seems to
>> be able to communicate its desires effectively.  I hope that you will
>> be able to learn from their example.
>
>As it happens, I've been on the GCC list for nearly a year now, and
>on a number of occasions we get an announcement beginning with "The
>committee has decided ..." and it's pretty clear there has been an
>extensive private discussion leading up to it.  I'm OK with that;
>I know that I personally don't want to have every bit of my
>thoughts and writing out in public, and I don't think I'm unusual
>in that.  Part of the reason is that sometimes I come up with
>boneheaded ideas, and I want to be able to bring them up without
>getting toasted by a hostile audience.  Indeed, had I not come to
>expect a generally rational reception from overseers, I would not
>have cc'ed that list.  I guess I was wrong.

Perhaps if you had provided some background and not discussed this as if
it was all a done deal, carried out in a back room, with no hope for
discussion, I would not have reacted so strongly.  I knew that the
potential for moving the gdb list has been discussed for a year or more.
I was nonetheless surprised to see your email here which seemed to be
discussing it as a fait accompli.

I'm sure that Andrew would have notified us internally of this decision
but I suspect that it was made while he was enjoying the benefits of
being in another time zone.

However, on rereading my email, I said some things that were over the
line and I apologize for that.

Anyway, I know that I am not going to convince anyone of anything.  I
can only represent this view: As a programmer I do not want to be
bothered by having to change all of my bookmarks, my mail filters, and,
eventually, I suppose, my CVS repository.

As a sysadmin, I dread the thought of moving the mailing lists, the web
page links, the mail archives, the ftp archives, the cgi scripts,
and the CVS repository, and I am discouraged to think that the effort
that I have been putting into getting a gdb test results page up and
running is all for naught.

I could accept any of these if I understood the rationale.  The best that
I can figure from your statements above is that Red Hat is being viewed
as having an agenda that does not jive with the FSF and rather than
allowing us to try to address that, you are taking all of your marbles
and going home.

That is certainly within the right of the FSF.  I just wish that someone
could offer an explanation that would satisfy me, but I know that there
is small chance of that.

If this decision does come to pass, I'll participate in the move.  I
consider that my job on sources.redhat.com and I'm not going to shirk it
because I disagree with an FSF decision.  I should have made that clear
in my first message.

And, I again apologize for any harshness in my previous message.

cgf

Re: New discussion list.

[Andrew Cagney]

Stan Shebs wrote:

> Well, we have a bit of chicken-and-egg problem, in that we want the
> committee's announcement to include a public contact address, and
> the consensus was to have it be the regular GDB address.

I don't know if that is exactly correct.  My last unresponded posting
was:

> I don't think these two tasks [announcement and mailing list change]
> should be tied together.  If the mailing
> list name is of concern, then perhaps just leave it out.

Remind me next time to spell out the consequences.

	Andrew

Re: New discussion list.

[Stan Shebs]

Andrew Cagney wrote:
> 
> Stan Shebs wrote:
> 
> > Now *this* is all good stuff to bring up.  The FSF position is that
> > the perception of independence is worth the trouble.  Personally,
> > I think it's possible to achieve that perception without moving
> > everything physically - sourceforge seems like a pretty successful
> > example.  But in committee discussion, persons in favor in staying
> > on sources.redhat.com have been lukewarm, while persons in favor
> > of moving to gnu.org have been pretty determined about that.  In the
> > year-and-some that the proto-committee has been in existence, no
> > one has yet made a strong and compelling argument for why and how to
> > keep GDB stuff on sources.redhat.com, and yet preserve the independence
> > associated with gnu.org.  Perhaps you need to be on the committee??
> 
> Stan,
> 
> I don't think that is a fair representation of the discussion on this
> matter.  Only two people were actually participating in the debate.

I was referring to the aggregate of the discussion that has happened
since our first meeting in May 1999.  Since the committee has to work
by consensus, all it would take is one person saying "No, I can't
accept the move to gnu.org".  As far as I remember, no one has ever
made that statement, while at least one person has said "I can't accept
the status quo".  If you (meaning any committee member) really don't
want to move anything to gnu.org, you need to say it in so many words.

> There never was a clear resolution of this matter by the committee.

Clearly we need more discussion, and given that many of the USAians
are incommunicado this weekend, we shouldn't be trying to take any
action until we hear from more people.  We've also gotten some
valuable technical input from Jason and Chris (thanks!), and I hope
the committee will take heed of that when deciding what to do and
when to do it.  I also want to thank the overseers list for putting
up with this intrusion, apologize for the distraction, and undertake
to make the next communication more useful.

Stan

Re: New discussion list.

[Jason Molenda]

Just to interject one point - Chris was correct in noting that
there is real work involved in a clean list move.  You need a lot
of background knowledge to know what you're doing, and even then
you've got to check over a host of loose ends to do it right.  It
is not something to be ventured lightly, it is not something you
would do to see how it goes, and maybe then just switch it back if
it isn't everything you'd hoped for.

Before anyone comes back to me saying, "Hey, my neighbor Joe says
he can move a mailing list in three minutes using just two shell
commands, a rubber band, and his pet rabbit," there's a world of
difference between a clean switchover where the users are pampered
all along the way and one that causes everyone a lot of grief.  It's
easy to do a minimal job, it's real work to do a great one.  This is
true of most things on sourceware - if you really look at the details,
you'll see that the infrastructure on sourceware is better thought
out than on nearly any other site on the net.


If a GDB steering committee really, actually exists (last I heard
they were busily taking a couple of years deciding whether they
were going to exist and whether anyone was going to actually do
anything besides complain about the Cygnus^H^H^H^H^HRed Hat hegemony),
and if that steering committee says that the list is going to move
to gnu.org, then I guess that's their business.  It'll be unfortunate
if inferior support and services are provided to gdb net developers
in the name of having everything end with a "gnu.org" domain name.

If the real priority is to purge the dreaded 'redhat.com' address,
has anyone considered mimicing the GCC location style (gdb.gnu.org)?
I don't know if RH would even be interested in providing those
services (believe me, it's a nontrivial amount of work to get it
all right), but if _I_ were on the GDB steering commitee and _I_
was trying to promote GDB development, and I had some great
infrastructure services being provided to me for free, I'd try to
avail myself of those services as much as possible.

Jason

Re: New discussion list.

[Stan Shebs]

Andrew Cagney wrote:
> 
> Stan Shebs wrote:
> 
> > Well, we have a bit of chicken-and-egg problem, in that we want the
> > committee's announcement to include a public contact address, and
> > the consensus was to have it be the regular GDB address.
> 
> I don't know if that is exactly correct.  My last unresponded posting
> was:
> 
> > I don't think these two tasks [announcement and mailing list change]
> > should be tied together.  If the mailing
> > list name is of concern, then perhaps just leave it out.

Argh, you're right, I did see that message, but my brain
spasmed.  Sorry sorry sorry.

Yes, it's more important to get the committee announcement out
in a timely fashion, so we don't have to endure any more jokes
about being a lame committee wannabe.

Stan

Re: New discussion list.

[Stan Shebs]

Christopher Faylor wrote:
> 
> [...]  AFAICT, you were
> letting us in on a private conversation with some FSF employee,
> discussing the best ways for moving the mailing list with no background
> or rationale for the move.
> 
> It was effectively like walking into my house wearing a hard hat and
> carrying a clipboard, discussing the best ways to level the place on
> your cell phone and then pausing to ask me "Do you have any ideas?"

Sorry, I thought everybody would remember that this issue had
come up before.  Indeed, I remember it being discussed internally
when I was at Cygnus.

> What is the political stance?  Red Hat is evil?  Red Hat is claiming to
> own the FSF?  What?  I am not privy to the private discussions that you
> must have had with RMS.  Maybe other people here are more aware of
> this.

I don't think the basic stance is any secret - it's important that
the GNU project be seen as independent, and not controlled by any
corporate entity.  RMS has been consistently very clear on this.
It's nothing particular against Red Hat, it just affects RH more
because of its important role in GNU development.

> I am aware that there is an anti-Red Hat sentiment.  Why not make it
> perfectly clear so that we can all understand it.

There can hardly be any anti-Red Hat sentiment from me, for one
thing I still own stock in it!  There are GDB committee members
whose companies compete with RH though, probably hard for them
not to think a negative thought about RH from time to time. :-)
 
> It is a good idea to move the mailing lists, because...
> [ evil deeds :-) ]

At the risk of playing psychoanalyst, I think you're seeing enemies
that don't exist.  The goal of having gnu.org host GNU development
has existed for several years, long before the events you're
referring to.  You could make an argument that gnu.org was not a
good place to host development, and indeed I myself have made that
argument in the past, and RMS' commitment has been that he will
make sure that it is a good place.  This issue came up in GDB
committee discussions, and when I was at Cygnus, I sent out
committee meeting reports saying that this was the long-term goal,
but that it was contingent upon gnu.org being adequate to the task.

> So why not provide a slight amount of background?  Do you think that
> adding overseers@sources.redhat.com to an ongoing discussion with
> someone is an effective technique for accomplishing your goals?  Maybe
> I'm wrong but I don't think that everyone who reads this list was aware
> of the existence of this committee or your role in it.

How quickly they forget... :-) I know that this was an active topic of
discussion when I was at Cygnus, both verbally and in email.  (Hmmm,
hard to believe a year has gone by already!)

> I was just barely aware of what was going on, myself until recently.  I
> thought you were still talking about what to call yourselves and whether
> people who had responsibility for gdb source code were helpers,
> developers, contributors, or maintainers.

Heh-heh.

> We can't
> associate with a company that has people with stong opinions, can we?

I don't mind strong opinions, indeed I expect those from the
wise overseers, but I was a little taken aback by the more personal
stuff ("gdbgroin"?).

> FWIW, I'm an individual.  I am very very far from being either a
> corporate shill and even farther from being a corporate spokesman.  I
> have, in fact, tried hard to do whatever it takes to clean up the Red
> Hat site to avoid the appearance of Red Hat owning anything that truly
> belongs to the FSF.  I've even participated in a draft of a document
> that will be sent to Red Hat internally detailing exactly how wrong it
> is to crow about any kind of "ownership" of anything like gcc or gdb.

That's good!  The trick is to convince the rest of the steering
committee though, who have so far been more enthusiastic about
moving to gnu.org than staying on sources.redhat.com.

> Perhaps if you had provided some background and not discussed this as if
> it was all a done deal, carried out in a back room, with no hope for
> discussion, I would not have reacted so strongly.  I knew that the
> potential for moving the gdb list has been discussed for a year or more.
> I was nonetheless surprised to see your email here which seemed to be
> discussing it as a fait accompli.

The discussion leading up to my cc did not feature Andrew or Jim
saying "no way, over our inanimate corpses", or else I never got
that message. :-)

> I'm sure that Andrew would have notified us internally of this decision
> but I suspect that it was made while he was enjoying the benefits of
> being in another time zone.

The dating on the messages I have shows that Andrew's last message
on the subject showed up in my mbox 17 minutes before your first
message, and he was talking about a possible privacy issue, not
whether the move was desirable in the first place, so I don't think
it was too much of a leap to believe that Red Hat folks were up to
date on things.

> However, on rereading my email, I said some things that were over the
> line and I apologize for that.

Apology accepted.

> Anyway, I know that I am not going to convince anyone of anything.  I
> can only represent this view: As a programmer I do not want to be
> bothered by having to change all of my bookmarks, my mail filters, and,
> eventually, I suppose, my CVS repository.
> 
> As a sysadmin, I dread the thought of moving the mailing lists, the web
> page links, the mail archives, the ftp archives, the cgi scripts,
> and the CVS repository, and I am discouraged to think that the effort
> that I have been putting into getting a gdb test results page up and
> running is all for naught.

Now *this* is all good stuff to bring up.  The FSF position is that
the perception of independence is worth the trouble.  Personally,
I think it's possible to achieve that perception without moving
everything physically - sourceforge seems like a pretty successful
example.  But in committee discussion, persons in favor in staying
on sources.redhat.com have been lukewarm, while persons in favor
of moving to gnu.org have been pretty determined about that.  In the
year-and-some that the proto-committee has been in existence, no
one has yet made a strong and compelling argument for why and how to
keep GDB stuff on sources.redhat.com, and yet preserve the independence
associated with gnu.org.  Perhaps you need to be on the committee??

Stan

Moving Projects

[Gerald Pfeifer]

On Thu, 23 Nov 2000, Andrew Cagney wrote:
> [As an individual that happens to be on the GDB Committee and previously
> happend to co-ordinate GDB development]

As an individual that happens to be on the GCC Committee and overseers,
and has some of experience with the current setup, the following analysis
that I've written some time ago and updated now might prove useful.

This is neither pro nor con doing such a move, just a(n incomplete)
list of issues that have to be considered and resolved *before* moving
gcc.gnu.org or GDB or some similiar project to a different machine.

MAILING LIST PERFORMANCE: gcc.gnu.org (with the current setup) performs
  extremely well, delivering messages within seconds. One has to verify
  that the new machines will sustain this under heavy load as well and
  not bring us back to the state where we had delays of several hours up
  to even a day.

SPAM: The gcc.gnu.org machine and all mailing lists as well as accounts
  there have been well secured against spam and many developers rely on
  that, while the gnu.org machines do not filter spamming servers.

  (For example one of our more well-known contributors had his regular
  GNU account closed because of the vast amounts of spam he received
  there. We do not want that to happen.)

BACKUP: Are the new machines properly backuped? How often? How?

GNATS DATABASES: Do these exist on the new machines? Are all interfaces
  (gnatsweb, directly, mail2gnats) available? Can the old databases be
  moved?

MAILING LIST ARCHIVE: This is especially tricky to handle correctly
  such that we do not lose messages.

  Specifically, the move should not be disruptive in that we do not want
  to split between old and new archives.

  Also mbox files of these archives should be kept and continued to
  be generated.

SEARCH ENGINE: Does it already exist on the new machine?

  Can the current reconfiguration be reused? We have some special settings.

  Who will take care of it regularily? Such a search engine needs a bit
  of hand-holding now and then.

ACCOUNT DATA: Obviously tons of account data will have to be moved.
  We are talking about one-hundred accounts here. This needs a bit of
  cooperation and planning, but shouldn't be hard per se.

RSYNC: Does the new server support rsync? This important to distribute
  the repository.

MOST SERVICES ARE RUNNING ASYNCHRONOUSLY: This means that it is critical
  to move services in the right order and with some short-time measures.

Overall, I feel that moving is by far less trivial than many might
believe, and I'm sure the list above is not complete. This does not
mean that such a move can or should not be done in principle, it just
means that we/you need some volunteers that are willing to do that.

Keep or just cloning the current box/setup, and perhaps relocating it
under a new name like sources.gnu.org with aliases gcc.gnu.org, gdb.gnu.org
might make sense.

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/

Re: New discussion list. (APOLOGY)

[Christopher Faylor]

Several people have pointed out to me very politely in email that some
of my remarks here were over the top.  On rereading my email, I am
forced to agree.  Way too much flamage and not enough substance...

I apologize to Stan and anyone else who was offended and I am grateful
that everyone resisted the urge to reply in kind.

I will try to do better in the future.  :-)

cgf

Re: New discussion list.

[Christopher Faylor]

On Wed, Nov 22, 2000 at 09:49:11PM -0500, Robert Dewar wrote:
><<For my own part, I think moving the discussion list to gnu.org is a
>reasonable step.  It gives us a chance to try out gnu.org's quality
>as a host, since unreliability is not a disaster (as it would be for,
>say, CVS), and it's relatively easy to reverse if necessary.
>>>
>
>seems reasonable to me too.
>I think the worry about the privacy policy is a red herring.

Perhaps it is.  How about this for a non-red-herring:

"I don't wanna."

I'm a volunteer who works on gdb.  I am one of the people who maintain
the mailing lists on this system.  I try very hard to be responsive to
people who make requests here.  I don't tend to be responsive to orders,
however, even when they are couched as ineffectively as this recent
email thread.

Personally, I do not understand the reasons for moving the mailing list.
I won't even pretend to ask what the technical reasons for the move are
because I know that there aren't any.

What you seem to want to do is move the mailing lists from one system
that is manned by people who believe in free software and are (or were)
proud to be associated with the FSF to another system with similar
people.  Technically speaking, the end result is a zero-sum game (unless
reports of unreliability of fencepost.gnu.org are true).  The effort
involved in moving is not "relatively easy", however, as has been
implied.

You are apparently expecting me, or some other volunteer sysadmin, to
stop whatever I'm doing and perform your bidding when we don't even,
officially, have any idea who you are.  Sure, I've know Stan Shebs.  The
last I heard he wasn't in a position to issue orders regarding gdb
mailing lists, however.

I have no, official, idea why Rober Dewar and Todd Whitesel are suddenly
making proclamations about the future of gdb and its mailing lists,
though.  What a botched job.

You guys realize that you are discussing this in a semi-public mailing
list and that many people probably have absolutely no idea who you are
and why you're making decisions about gdb, right?

Is this really your first public unveiling?  A cc'ed message with vague
mumblings about moving a mailing list?

AFAICT, you are representing the GDB Steering Committee (or is it the
GDB Maintainers Committee?) Apparently this committe has has been formed
as a mechanism for issuing edicts about gdb without the benefit of
public discussion.

Is this what we can expect for future pronouncements from the steering
committee?  Will people who are actually in the trenches doing something
be expected to drop everything when they get a Cc'ed email message with
hazy musings from Stan Shebs?

You probably ought to take a stroll over to the gcc mailing list where
the concept of steering committee openness has recently been discussed.
The GCC steering committee apparently values openness and seems to
be able to communicate its desires effectively.  I hope that you will
be able to learn from their example.

The only thing I know for sure right now, however, from this limited
exposure, is that you seem to be ineffective, email-challenged
communicators.  That is neither a propitious nor an auspicious debut for
your little cabal, IMO.  Of course, maybe a cabal doesn't need to worry
about public image.

So, can you take your discussion to your secret gdbgroins mailing list
and stop discussing things here?  Once you've reached a clear decision,
maybe you can ask Andrew Cagney to put it into words for you.  Hopefully
he will be able to provide slightly more rationale for whatever you are
pronouncing.  For the record, an self-congratulatory agreements with
each other that your clever plan is "a reasonable step" is neither a
rationale nor a rallying cry for getting people to help you.

cgf

Re: New discussion list.

[Robert Dewar]

<<For my own part, I think moving the discussion list to gnu.org is a
reasonable step.  It gives us a chance to try out gnu.org's quality
as a host, since unreliability is not a disaster (as it would be for,
say, CVS), and it's relatively easy to reverse if necessary.
>>

seems reasonable to me too.
I think the worry about the privacy policy is a red herring.

Re: New discussion list.

[Stan Shebs]

Christopher Faylor wrote:
> 
> I'm a volunteer who works on gdb.  I am one of the people who maintain
> the mailing lists on this system.  I try very hard to be responsive to
> people who make requests here.  I don't tend to be responsive to orders,
> however, even when they are couched as ineffectively as this recent
> email thread.

Whoa, a bit of tense reaction here!  Perhaps the "orders" seem to be
couched ineffectively because they're not actually orders?  I purposely
cc'ed overseers on this discussion, so as to get input from relevant
and knowledgeable people, not to disseminate demands.

> Personally, I do not understand the reasons for moving the mailing list.
> I won't even pretend to ask what the technical reasons for the move are
> because I know that there aren't any.

That's right.  As you know, GNU is not a purely technical project;
there is a political aspect to it as well.  I've had occasion to
think about this many times over the years, starting with when I
ported GCC to the Mac while the FSF was officially boycotting Apple,
which meant that I was entirely on my own, and pissed upon if I
even dared to publicly mention what I was doing.  Even so, I think
the GNU project is right to take political stances; there are a
great many ways for corporations to take advantage of and to co-opt
free software projects, and now that "open source" is fashionable,
there are a lot more people thinking about ways to exploit us.  While
I don't always agree with RMS' political decisions, I do believe
that it is valid to include political issues in decisionmaking for GNU.

> I have no, official, idea why Rober Dewar and Todd Whitesel are suddenly
> making proclamations about the future of gdb and its mailing lists,
> though.  What a botched job.
> 
> You guys realize that you are discussing this in a semi-public mailing
> list and that many people probably have absolutely no idea who you are
> and why you're making decisions about gdb, right?

Well, we have a bit of chicken-and-egg problem, in that we want the
committee's announcement to include a public contact address, and
the consensus was to have it be the regular GDB address.  RMS pointed
out the problem with having the Red Hat address embedded in his
official announcement, but the usual way to discuss the handling of
mailing lists is via the semi-public overseers.

> Is this really your first public unveiling?  A cc'ed message with vague
> mumblings about moving a mailing list?
> 
> AFAICT, you are representing the GDB Steering Committee (or is it the
> GDB Maintainers Committee?) Apparently this committe has has been formed
> as a mechanism for issuing edicts about gdb without the benefit of
> public discussion.

As you may remember, it's been discussed in public before, indeed
last summer I put out a call for people interested in participating.

> Is this what we can expect for future pronouncements from the steering
> committee?  Will people who are actually in the trenches doing something
> be expected to drop everything when they get a Cc'ed email message with
> hazy musings from Stan Shebs?

Well, they would at least be expected not to flame - semi-publicly! -
about it.  The argument for hosting stuff at Red Hat is that the
service is good and reliable, but if attempts to include more Red Hat
people in a discussion are met with vitriol and insults, that is
actually a strong argument against hosting things there.

> You probably ought to take a stroll over to the gcc mailing list where
> the concept of steering committee openness has recently been discussed.
> The GCC steering committee apparently values openness and seems to
> be able to communicate its desires effectively.  I hope that you will
> be able to learn from their example.

As it happens, I've been on the GCC list for nearly a year now, and
on a number of occasions we get an announcement beginning with "The
committee has decided ..." and it's pretty clear there has been an
extensive private discussion leading up to it.  I'm OK with that;
I know that I personally don't want to have every bit of my
thoughts and writing out in public, and I don't think I'm unusual
in that.  Part of the reason is that sometimes I come up with
boneheaded ideas, and I want to be able to bring them up without
getting toasted by a hostile audience.  Indeed, had I not come to
expect a generally rational reception from overseers, I would not
have cc'ed that list.  I guess I was wrong.

Stan

Re: New discussion list.

[Robert Dewar]

<<Do you think that because ACT derive benefit from holding the "master"
repository to GNU ADA (all changes must go through ACT and into their
repository first), that this is similarly bad? In many respects that is a
more extreme situation than with GDB. Do you remember the arguments used to
>>>

Just to note here, we have no requirement that "all changes" must go
through ACT. The idea of setting up a publoic tree is precisely so
that
the process can be opened up.

Re: New discussion list.

[Todd Whitesel]

> >I agree with RMS' goal here, but I also don't think it should be rushed
> >just so the announcement can pretend that no transition happened.
> 
> And, what is that goal?  If you are going to discuss something in a
> semi-public mailing list then provide details.  Otherwise please drop
> overseers@sources.redhat.com from your list.

Sorry, I am just group-replying. I shouldn't be the first email you've seen
about this, unless things arrived out of order (which does happen).

Rather than risk misrepresenting the discussion however, I will leave it to
others to explain what's going on here.

Todd Whitesel
toddpw @ best.com

Re: New discussion list.

[Alexandre Oliva]

On Nov 28, 2000, Richard Stallman <rms@gnu.org> wrote:

>     The rumours I keep hearing are directed at the current systems.  The
>     most recent being that the autoconf mailing list stopping dead for a few
>     hours (~9:20UCT and 16:53UCT) on Nov 22.

> I can investigate why this happened, but it looks like a fairly small
> problem that was fixed reasonably fast.

Actually, messages only started to show up on Wednesday, but they had
been posted on Monday morning, French time.

> You seem to be trying desperately to find fault with our servers.

Well, if you think so, here's one more ``desperate try''.  The CVS
server is amazingly slow.  A single commit in a small file in autoconf
or libtool takes forever (ask Akim), whereas committing even large
files in gcc.gnu.org works reasonably fast (even though I'm on the
other end of a very slow network connection).  I can't understand why
it's that slow.  Maybe it's on the same network link as Sourceforge?
I also observe very long commit times when checking in stuff in
amanda.sourceforge.net.

> That is not the right way to treat other people who work on the GNU
> Project, so please don't do it any more.

Do you think forcing people to use a machine on a slower network
connection to do their volunteer work is the right way to treat them?

We do have a gcc.gnu.org machine with a reasonably fast network link.
Well, Richard Kenner feels otherwise, but there's no indication that
moving to subversions.gnu.org would improve anything for him or
anybody else.

Moving GDB would just make it harder to have the unified CVS tree lots
of people have been asking for.  I'd rather keep everything in the
current machine, maybe rename it to dev.gnu.org (keeping gcc.gnu.org
as an alias and using virtual WWW hosting so that people opening
gcc.gnu.org would get to the right web-page), and create the
long-awaited unified tree in it.  *That* would be a step in the right
direction, IMO.

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me

Re: New discussion list.

[Stan Shebs]

So the theory of this list (as I understand it) is to be able to
run generic GDB discussion.  If so, we now need to move list
members over en masse, set up forwarding, and ideally copy over
the old archives and moosh them into the mailman format.

Presumably we can do all the setup and test it a bit before
turning off gdb@sources?

Stan

Phillip Rulon wrote:
> 
> Greetings,
>    There is a new mailing list, gdb@gnu.org, available for gdb
> discussion.  The list administrator is currently me, if there
> is someone else interested in doing it let me know.
> 
> The list is run under GNU/mailman.
> 
> List-Help:        < mailto:gdb-request@gnu.org?subject=help >
> List-Post:        < mailto:gdb@gnu.org >
> List-Subscribe:   < http://mail.gnu.org/mailman/listinfo/gdb >,
>                   < mailto:gdb-request@gnu.org?subject=subscribe >
> List-Unsubscribe: < http://mail.gnu.org/mailman/listinfo/gdb >,
>                   < mailto:gdb-request@gnu.org?subject=unsubscribe >
> List-Archive:     < http://mail.gnu.org/pipermail/gdb/ >
> 
> No one from gdbheads has been added thus far.  Questions to
> pjr@gnu.org
> 
> Thanks,
> pjr

Re: New discussion list.

[Christopher Faylor]

On Wed, Nov 22, 2000 at 07:37:34PM -0800, Todd Whitesel wrote:
>> :-)  Yes, there's another dozen people on the steering committee who
>> *should* be weighing in on this, hint hint guys...
>
>Ok.
>
>I agree with RMS' goal here, but I also don't think it should be rushed
>just so the announcement can pretend that no transition happened.

And, what is that goal?  If you are going to discuss something in a
semi-public mailing list then provide details.  Otherwise please drop
overseers@sources.redhat.com from your list.

>Couldn't the announcement name the current list and mention that there
>will be a transition in the near future?
>
>It could be argued that then the transition will never happen, but if
>we can agree on the announcement then that commits us to eventually
>doing it, and might even draw a few volunteers out of the woodwork.

Volunteers with root access to sources.redhat.com and knowledge of the
mailing lists there?  Hmm.

cgf

Re: New discussion list.

[Jonathan Larmour]

Robert Dewar wrote:
> 
> <<Do you think that because ACT derive benefit from holding the "master"
> repository to GNU ADA (all changes must go through ACT and into their
> repository first), that this is similarly bad? In many respects that is a
> more extreme situation than with GDB. Do you remember the arguments used to
> >>>
> 
> Just to note here, we have no requirement that "all changes" must go
> through ACT. The idea of setting up a publoic tree is precisely so
> that the process can be opened up.

Sorry yes, I have reread the thread on gcc, and I think I suffered the same
confusion as some others probably did (mostly from lack of background).
I've now worked out your intent! What I said above was bogus.

Jifl
-- 
Red Hat, 35 Cambridge Place, Cambridge, UK. CB2 1NS  Tel: +44 (1223) 728762
"Plan to be spontaneous tomorrow."  ||  These opinions are all my own fault

Re: New discussion list.

[Jeffrey A Law]

  In message < 200011252219.PAA21460@wijiji.santafe.edu >you write:
  >     - and most importantly, unless something recently has changed, gnu.org
  >     machines are not renowned for being reliable.
  > 
  > That reputation is based on things that happened on other machines (no
  > longer used) which were kept at an inconvenient location where they
  > were often not physically accessible to anyone.  However, people keep
  > repeating the old information that "the GNU servers are not reliable".
  > It looks like this will continue to propagate, regardless of what
  > changes, until more people have actual experience with the current
  > situation.
Err, you seem to have forgotten my recent problems accessing FSF machines
via kerberos (within the last month).

From what I can tell, when the FSF converted to kerberos a set of machines
as publicized as the right machines to connect to.  Then one day, they all
went off the air.

Kerberos rlogin to the new server (fencepost) didn't work, while kerberos
telnet did work.  Rather odd and extremely inconvenient since I almost never
use kerberos telnet.

  > Experience with the machine gcc.gnu.org shows that the actual nature
  > of the machine can be seen by users in various ways.  These servers
  > support many different protocols.
And we've stated that we'll fix them.  Jeez.  We know the worst one right
now is ftp service isn't virtualized.  We're going to fix it.


jeff

Re: New discussion list.

[Richard Stallman]

    - and most importantly, unless something recently has changed, gnu.org
    machines are not renowned for being reliable.

That reputation is based on things that happened on other machines (no
longer used) which were kept at an inconvenient location where they
were often not physically accessible to anyone.  However, people keep
repeating the old information that "the GNU servers are not reliable".
It looks like this will continue to propagate, regardless of what
changes, until more people have actual experience with the current
situation.

Moving the GDB discussion list, repository, and web pages will show
people that our new servers are reliable.

    But gdb.gnu.org would fully give the impression of being a GNU machine.

Experience with the machine gcc.gnu.org shows that the actual nature
of the machine can be seen by users in various ways.  These servers
support many different protocols.

Re: New discussion list.

[Robert Dewar]

<<Be careful: do not confuse an e-mail address with a role. I, and no doubt
most others, are speaking as individuals. I don't care what my employer
thinks about what I write here.
>>

You misunderstood what I was saying, I was not saying that you were speaking
*for* Red Hat, but you were certainly speaking *about* Red Hat, and as I say,
it is a good idea to keep such discussions calm and non-personal!

Re: New discussion list.

[Robert Dewar]

<<There can hardly be any anti-Red Hat sentiment from me, for one
thing I still own stock in it!  There are GDB committee members
whose companies compete with RH though, probably hard for them
not to think a negative thought about RH from time to time. :-)
>>

Speaking as a "GDB committee member whose company [might] complete with
RH". I don't have negative thoughts about RH. I always welcome competition
providing it is fair and above board, something which I think we achieve
in our community better than a lot of other companies. The odd thing
here is that in the open development model for Free Software, we all
depend on our competitors to help us. That's a bit odd, and would
puzzle the Ford's and Rockefeller's of the world perhaps, but it works
out amazing well. Part of what we are doing here is working ahead of
time to avoid any problems that come from this unusual cooperation.
RMS can help us in this goal, by providing a guiding influence that
is definitely NOT associated with any of the companies involved,
and I find his encouragement to be sensitive to appearences of
excessive connection of stuff to individual companies to be 
helpful!

Given Red Hat's central and critical contributions, it is also very
important to make sure that Red Hat is happy with how things proceed
(that's just a special case of it being important that everyone be
happy with the setup). So the discussion is definitely legitimate,
but it is helpful to keep it calm and non-personal (especially where
the persons are giant companies :-)

Robert Dewar

Re: New discussion list.

[Andrew Cagney]

Richard Stallman wrote:
> 
>     The rumours I keep hearing are directed at the current systems.  The
>     most recent being that the autoconf mailing list stopping dead for a few
>     hours (~9:20UCT and 16:53UCT) on Nov 22.
> 
> I can investigate why this happened, but it looks like a fairly small
> problem that was fixed reasonably fast.

I'm sorry but 7 hours to have a problem like this resolved is not
reasonably fast.

> You seem to be trying desperately to find fault with our servers.
> That is not the right way to treat other people who work on the GNU
> Project, so please don't do it any more.

I'm not trying to be disrespectful towards the people who maintain the
GNU servers.  They do a very good job with the limited resources they
have available.

In the case of the above I'm assuming that the problem was indeed
resolved promptly and professionally once the required resources were
available.

	Andrew

Re: New discussion list.

[Todd Whitesel]

> :-)  Yes, there's another dozen people on the steering committee who
> *should* be weighing in on this, hint hint guys...

Ok.

I agree with RMS' goal here, but I also don't think it should be rushed
just so the announcement can pretend that no transition happened.

Couldn't the announcement name the current list and mention that there
will be a transition in the near future?

It could be argued that then the transition will never happen, but if we
can agree on the announcement then that commits us to eventually doing it,
and might even draw a few volunteers out of the woodwork.

Todd Whitesel
toddpw @ best.com

Re: New discussion list.

[Richard Stallman]

    Err, you seem to have forgotten my recent problems accessing FSF machines
    via kerberos (within the last month).

    >From what I can tell, when the FSF converted to kerberos a set of machines
    as publicized as the right machines to connect to.  Then one day, they all
    went off the air.

That was a couple of months ago, when we turned off the old
inaccessible mail and file servers, and turned on the new one
(fencepost) that we can maintain better.  We sent an announcement to
gnu-prog, a list that all GNU maintainers should be on, but you were
not on the list.  Once you contacted us, everything was straightened
out quickly.

Re: New discussion list.

[Andrew Cagney]

Richard Stallman wrote:
> 
>     - and most importantly, unless something recently has changed, gnu.org
>     machines are not renowned for being reliable.
> 
> That reputation is based on things that happened on other machines (no
> longer used) which were kept at an inconvenient location where they
> were often not physically accessible to anyone.  However, people keep
> repeating the old information that "the GNU servers are not reliable".
> It looks like this will continue to propagate, regardless of what
> changes, until more people have actual experience with the current
> situation.

The rumours I keep hearing are directed at the current systems.  The
most recent being that the autoconf mailing list stopping dead for a few
hours (~9:20UCT and 16:53UCT) on Nov 22.  Grubbing around and it would
appear that this was indeed the case.

	Andrew

http://mail.gnu.org/pipermail/autoconf-patches/2000-November/006339.html
http://sources.redhat.com/ml/autoconf-patches/2000-11/msg00223.html

Re: New discussion list.

[Stan Shebs]

Jeffrey A Law wrote:
> 
>   > Experience with the machine gcc.gnu.org shows that the actual nature
>   > of the machine can be seen by users in various ways.  These servers
>   > support many different protocols.
> And we've stated that we'll fix them.  Jeez.  We know the worst one right
> now is ftp service isn't virtualized.  We're going to fix it.

One of the things that strikes about this discussion is that there
are two separate issues that have gotten tied together.  One is
the hosting and name mapping, and the other is the question of
who maintains the hosts.

For instance, suppose one were to transplant all the existing software
bits on sources over to a gnu.org machine, and give all the people on
sources the same degree of access at gnu.org.  Of course there would
still be some temporary disruption, but is this a satisfactory long-term
situation?  If not, why not?

Conversely, what is the disadvantage for the GNU project of having
some of its machines be seen to be run by Red Hat if one looks hard
enough?  Many different corporations donate to GNU, either in time
or money, and yet we don't worry that GNU somehow becomes beholden
to them as a result.  Red Hat is basically offering, for certain
projects, better hosting services than GNU can normally provide,
and they're also offering to give up corporate visibility if any
is discovered to leak through.  Is there something about the physical
hosting of gnu.org that is important, and if so, what is it?

Stan

Re: New discussion list.

[Richard Stallman]

    The rumours I keep hearing are directed at the current systems.  The
    most recent being that the autoconf mailing list stopping dead for a few
    hours (~9:20UCT and 16:53UCT) on Nov 22.

I can investigate why this happened, but it looks like a fairly small
problem that was fixed reasonably fast.

You seem to be trying desperately to find fault with our servers.
That is not the right way to treat other people who work on the GNU
Project, so please don't do it any more.

Re: New discussion list.

[Jonathan Larmour]

Richard Stallman wrote:
> 
>     - and most importantly, unless something recently has changed, gnu.org
>     machines are not renowned for being reliable.
> 
> That reputation is based on things that happened on other machines (no
> longer used) which were kept at an inconvenient location where they
> were often not physically accessible to anyone.  However, people keep
> repeating the old information that "the GNU servers are not reliable".
> It looks like this will continue to propagate, regardless of what
> changes, until more people have actual experience with the current
> situation.

I chose my words carefully, and in other mails solicited people's opinions
on whether it was still the case. Thank you for clarifying.

>     But gdb.gnu.org would fully give the impression of being a GNU machine.
> 
> Experience with the machine gcc.gnu.org shows that the actual nature
> of the machine can be seen by users in various ways.  These servers
> support many different protocols.

I know of no user-visible way, other than deliberately looking to find the
identity of the machine, e.g. via traceroute etc, and that is not relevant
for your average user. Since the gcc folks are trying hard to clear out
such references, then anything user-visible is really a problem that needs
fixing. What are you thinking of?

Jifl

Re: New discussion list.

[Christopher Faylor]

On Wed, Nov 22, 2000 at 04:16:08PM -0800, Stan Shebs wrote:
>For my own part, I think moving the discussion list to gnu.org is a
>reasonable step.  It gives us a chance to try out gnu.org's quality
>as a host, since unreliability is not a disaster (as it would be for,
>say, CVS), and it's relatively easy to reverse if necessary.

"relatively easy" == "I don't have to do it, I just have to ask someone
to do it".

>Interesting point - I had always considered Cygnus and Red Hat to be
>managing the list on behalf of the FSF, so in a sense the FSF "owns"
>the information.  But it's not something that's well defined.  (Quick,
>to the Florida courts! :-) )

If many are under the impression that the mailing lists are managed by
the FSF, doesn't that sort of invalidate the whole reason for moving the
lists?

Not that the slightest hint of a rationale has surfaced in this discussion,
of course.

cgf

Re: New discussion list.

[Andrew Cagney]

Stan Shebs wrote:
> 
> So the theory of this list (as I understand it) is to be able to
> run generic GDB discussion.  If so, we now need to move list
> members over en masse, set up forwarding, and ideally copy over
> the old archives and moosh them into the mailman format.
> 
> Presumably we can do all the setup and test it a bit before
> turning off gdb@sources?

Firstly I've not seen a clear decision from the GDB Maintainer regarding
the question gdb@gnu.org VS gdb@gdb.gnu.org.  Perhaphs I should ask for
a re-count.

Secondly, I have real reservations over simply handing over the mailing
list gdb@sources.redhat.com.  Refering to sources.redhat.com's privicy
statement: ``We'd never use any of the mailing list subscription
information for anything other than the normal distribution of e-mail to
that list.''.  I'm honestly sorry about this.  In my opinion, people
will need to be asked to make the move.

With regard to archives, they are all available for download so that
isn't a problem.

	Andrew

Re: New discussion list.

[Jeffrey A Law]

  In message < 3A22976A.46AAC43B@shebs.cnchost.com >you write:
  > Jeffrey A Law wrote:
  > One of the things that strikes about this discussion is that there
  > are two separate issues that have gotten tied together.  One is
  > the hosting and name mapping, and the other is the question of
  > who maintains the hosts.
Right.

  > For instance, suppose one were to transplant all the existing software
  > bits on sources over to a gnu.org machine, and give all the people on
  > sources the same degree of access at gnu.org.  Of course there would
  > still be some temporary disruption, but is this a satisfactory long-term
  > situation?  If not, why not?
It would be acceptable as far as I'm concerned -- as long as the FSF folks
continue to make progress on the reliability of their machines.  Contrary
to RMS's assertions there have been significant outages and problems over
the last few months, though things are better than they were a year+ ago.

Not to say that Red Hat hasn't had outages (we have), but I believe our
track record for providing reliable service over the last 3 years speaks
for itself.

  > Conversely, what is the disadvantage for the GNU project of having
  > some of its machines be seen to be run by Red Hat if one looks hard
  > enough?  Many different corporations donate to GNU, either in time
  > or money, and yet we don't worry that GNU somehow becomes beholden
  > to them as a result.  Red Hat is basically offering, for certain
  > projects, better hosting services than GNU can normally provide,
  > and they're also offering to give up corporate visibility if any
  > is discovered to leak through.  Is there something about the physical
  > hosting of gnu.org that is important, and if so, what is it?
Furthermore, how does the answer to the physical hosting question change
if the machine isn't physically at one of Red Hat's sites?  Red Hat
is planning to physically move the machine to a co-location site to
greatly increase network bandwidth.

jeff

Re: New discussion list.

[Richard Stallman]

      Mooshing may not be necessary, we can html-ize the archive
    and provide a link from the new archive.

It is ok to continue to refer to the old mail archive where it is now;
that is one thing which is not crucial to move.  However, it is also
ok to move the archive.  (Mail archives are one of the things that
we don't try to always have on gnu.org machines.)

Re: New discussion list.

[Jason Molenda]

On Wed, Nov 22, 2000 at 12:19:06PM -0800, Stan Shebs wrote:
> So the theory of this list (as I understand it) is to be able to
> run generic GDB discussion.  If so, we now need to move list
> members over en masse, set up forwarding, and ideally copy over
> the old archives and moosh them into the mailman format.
> 
> Presumably we can do all the setup and test it a bit before
> turning off gdb@sources?



Whoever does the move, this is an identical repeat of the guile list
move in September.  My note outlining the process is here:

	http://sources.redhat.com/ml/overseers/2000-q3/msg00300.html

Follow the thread for other notes on the subject.  You should be able
to refer to the current guile list setup for tips on exactly how a
moved list should look in the end.


Jason

Re: New discussion list.

[Phillip Rulon]

   Envelope-to: pjr@gnu.org
   From: Todd Whitesel <toddpw@best.com>
   Date: Wed, 22 Nov 2000 19:37:34 -0800 (PST)
   Cc: ac131313@cygnus.com, pjr@gnu.org, gdbheads@gnu.org,
	   overseers@sources.redhat.com
   Content-Type: text/plain; charset=US-ASCII

   > :-)  Yes, there's another dozen people on the steering committee who
   > *should* be weighing in on this, hint hint guys...

   Ok.

   I agree with RMS' goal here, but I also don't think it should be
   rushed just so the announcement can pretend that no transition
   happened.

   Couldn't the announcement name the current list and mention that
   there will be a transition in the near future?

I think it is probably a good idea to boot the new setup with the new
list.  We can, of course, import the old list at the same time.  Just
my opinion.

   It could be argued that then the transition will never happen, but
   if we can agree on the announcement then that commits us to
   eventually doing it, and might even draw a few volunteers out of
   the woodwork.

I think it's safe to say that the transition will happen, I've been
asked to make it so.  I don't have much to say about the announcement
but the list is kind of a done deal.

pjr

Re: New discussion list.

[Christopher Faylor]

On Sat, Nov 25, 2000 at 10:33:45PM +0000, Jonathan Larmour wrote:
>Richard Stallman wrote:
>> Experience with the machine gcc.gnu.org shows that the actual nature
>> of the machine can be seen by users in various ways.  These servers
>> support many different protocols.
>
>I know of no user-visible way, other than deliberately looking to find the
>identity of the machine, e.g. via traceroute etc, and that is not relevant
>for your average user. Since the gcc folks are trying hard to clear out
>such references, then anything user-visible is really a problem that needs
>fixing. What are you thinking of?

The Received headers in mail messages will display the
sources.redhat.com address.  I don't think think there is any around
this since I believe that they are generated on the machine receiving
the email.

I think we may still have problems with GNATS displaying
"sources.redhat.com" in some places, too, but we're working on that.

One thing that we are discussing is changing the name of
sources.redhat.com to something that is vendor neutral.  The trick, of
course, is coming up with a domain name that is catchy and mnemonic.
That's almost as hard as the technical aspects of changing the name.

The reality of sources.redhat.com is that Red Hat corporate really only
takes note of its existence when they notice the bill for the T1 line.
Since this site started out as a Cygnus system, it doesn't seem to have
reached the awareness of many people in the home office.

Oddly enough, however, whenever the concept of this system is explained
there is never any problem with justifying its ongoing expense.  I
wasn't involved in its creation, but I believe that this system was
intended purely to be a donation to the free software community.  That's
how I always portray it, anyway, and no one has ever had any problems
with that.

So, I don't believe that changing the domain name will be a big deal.
IMO, we'll probably be changing the domain name of this system
regardless of the outcome of any move to gnu.org.  I have personal
reasons for wanting to do this.  A move to a new domain name should stop
the flow of email to sourcemaster regarding problems (real and
imaginary) with Red Hat's GNU/Linux releases.  Not having to explain
that we have nothing to do with this every other day will make a switch
to a new domain name a blessing for me.  :-)

cgf

Re: New discussion list.

[Phillip Rulon]

   Envelope-to: pjr@gnu.org
   Sender: ac131313@localhost.cygnus.com
   Date: Thu, 23 Nov 2000 14:21:49 +1100
   From: Andrew Cagney <ac131313@cygnus.com>
   X-Accept-Language: en
   Cc: Phillip Rulon <pjr@gnu.org>, gdbheads@gnu.org,
	   overseers@sources.redhat.com
   Content-Type: text/plain; charset=us-ascii

   Stan Shebs wrote:

   > Interesting point - I had always considered Cygnus and Red Hat to be
   > managing the list on behalf of the FSF, so in a sense the FSF "owns"
   > the information.  But it's not something that's well defined.  (Quick,
   > to the Florida courts! :-) )

   That is the view that Red Hat liked to have.  However, according to
   http://mail.gnu.org/mailman/listinfo/bug-gdb for bug-gdb@gnu.org ``There
   are no other GNU mailing lists or gnUSENET newsgroups for GDB'' (well
   until gdb@gnu.org was created).

Our first bug, thanks, I'll fix it.

pjr

Re: New discussion list.

[Andrew Cagney]

Stan Shebs wrote:

> Interesting point - I had always considered Cygnus and Red Hat to be
> managing the list on behalf of the FSF, so in a sense the FSF "owns"
> the information.  But it's not something that's well defined.  (Quick,
> to the Florida courts! :-) )

That is the view that Red Hat liked to have.  However, according to
http://mail.gnu.org/mailman/listinfo/bug-gdb for bug-gdb@gnu.org ``There
are no other GNU mailing lists or gnUSENET newsgroups for GDB'' (well
until gdb@gnu.org was created).

> I think it should suffice to inform list members that it will be moving,
> give them a week or two to unsubscribe if they're really worried about
> the FSF selling their address to spammers. :-)  I suspect that many
> will be surprised to learn that their addresses weren't already being
> managed by the FSF...

I'm sure of that as well.  I think there are some very efficient ways of
getting people to transfer across with out actually transfering the
list.  If the FSF don't consider the list theirs then a simple mail out
declaring the existing list dead with a URL for how to subscribe to the
new list would be the obvious answer.  If the FSF do actually consider
the list theirs then giving everyone the oportunity to opt out and then
transfer things would be better.

	Andrew

Re: New discussion list.

[Jonathan Larmour]

Robert Dewar wrote:
> 
> <<Given where the majority of GDB work comes from, I can't help but think
> that it throws obstacles in the way of those who contribute the most,
> *while not removing any from anyone else*.
> >>
> 
> To me that's a worrisome argument, it is close to arguing that Red Hat
> should get favored treatment *because* it is "where the majority of GDBwork comes from"
> 
> I understand what you are saying, but you should see that it is precisely
> this argument that suggests avoiding any hint of favored treatment.

And I understand what you are saying :-), but my point is that if moving
the list would make things equally good for everyone that would be
sensible. But an argument that suggests things should be equally *bad* for
everyone out of principle is not very practical nor constructive. It would
sound like dogma.

Do you think that because ACT derive benefit from holding the "master"
repository to GNU ADA (all changes must go through ACT and into their
repository first), that this is similarly bad? In many respects that is a
more extreme situation than with GDB. Do you remember the arguments used to
justify that situation? [ I don't want to repeat the thread here by
explaining to everyone and reiterating here what was said - treat it as a
rhetorical question :) ]

> Certainly we do not want to throw obstacles in any one's way. Can you
> explain clearly *why* moving the mailing list would have that effect.

- Moving management of lists away from many GDB maintainer's/contributor's
influence (one notable person involved with GDB, for example, has root
access; several have shell access). This is related to...

- response time to problems. The ability to do things yourself, or tell
someone close by what to do will always be better than e-mailing a remote
volunteer sysadmin with no involvement or personal interest in GDB.

- integration with the rest of the infrastructure (think about things like
integration betwen GNATS, mailing list archives (especially gdb-patches),
CVS, and automatic test infrastructure, for example);

- related to that type of thing, customizability - sources, is very liberal
with what you can do, and has an infrastructure good enough to handle it...
it may be an opinion, but I cannot visualize the same flexibility on a
gnu.org machine. I'm happy to be wrong, so please tell me that they really
are this flexible.

- and most importantly, unless something recently has changed, gnu.org
machines are not renowned for being reliable. If the SC really advocates
moving somewhere known to be less reliable (if it truly is so!) then GDB
users would rightly feel disenfranchised, and the SC would lose respect
before it's had a chance to prove its value.

All but one of these points (integration) could be irrelevant if gnu.org
runs a very different regime from what I believe it does. But integration
is a bit of a stumbling block that could only be resolved by running most
services on the same machine, wherever it is located.

But gdb.gnu.org would fully give the impression of being a GNU machine. Who
cares where it really is?

I probably won't post any more as I already have several questions needing
answering before I could really 100% advocate one way or the other - I'm
far from set against the idea; but I have many doubts that want pacifying
first. And the opposition to gdb.gnu.org is also a mystery.

Jifl

Re: New discussion list.

[Phillip Rulon]

   Envelope-to: pjr@gnu.org
   Date: Wed, 22 Nov 2000 12:19:06 -0800
   From: Stan Shebs <shebs@apple.com>
   X-Accept-Language: en
   CC: gdbheads@gnu.org, overseers@sources.redhat.com
   Content-Type: text/plain; charset=us-ascii

   So the theory of this list (as I understand it) is to be able to
   run generic GDB discussion.

I'm open to theory ideas, I just joined at the request of RMS.  I'm
in the interesting position of running the list while not being a
recognized gdb-hacker.  I know enough about the system setup at GNU
to make the list useful but I'll defer to this list for policy.

   If so, we now need to move list members over en masse, set up
   forwarding, and ideally copy over the old archives and moosh them
   into the mailman format.

All doable.  Mooshing may not be necessary, we can html-ize the archive
and provide a link from the new archive.

   Presumably we can do all the setup and test it a bit before
   turning off gdb@sources?

I've done some testing, please do exercize it as much as you wish.

pjr

   Phillip Rulon wrote:
   > 
   > Greetings,
   >    There is a new mailing list, gdb@gnu.org, available for gdb
   > discussion.  The list administrator is currently me, if there
   > is someone else interested in doing it let me know.
   > 
   > The list is run under GNU/mailman.
   > 
   > List-Help:        < mailto:gdb-request@gnu.org?subject=help >
   > List-Post:        < mailto:gdb@gnu.org >
   > List-Subscribe:   < http://mail.gnu.org/mailman/listinfo/gdb >,
   >                   < mailto:gdb-request@gnu.org?subject=subscribe >
   > List-Unsubscribe: < http://mail.gnu.org/mailman/listinfo/gdb >,
   >                   < mailto:gdb-request@gnu.org?subject=unsubscribe >
   > List-Archive:     < http://mail.gnu.org/pipermail/gdb/ >
   > 
   > No one from gdbheads has been added thus far.  Questions to
   > pjr@gnu.org
   > 
   > Thanks,
   > pjr

Re: New discussion list.

[Robert Dewar]

<<Given where the majority of GDB work comes from, I can't help but think
that it throws obstacles in the way of those who contribute the most,
*while not removing any from anyone else*.
>>

To me that's a worrisome argument, it is close to arguing that Red Hat
should get favored treatment *because* it is "where the majority of GDBwork comes from"

I understand what you are saying, but you should see that it is precisely
this argument that suggests avoiding any hint of favored treatment.

Certainly we do not want to throw obstacles in any one's way. Can you
explain clearly *why* moving the mailing list would have that effect. 

Re: New discussion list.

[Robert Dewar]

>I am aware that there is an anti-Red Hat sentiment.  Why not make it
>perfectly clear so that we can all understand it.

I think that's not a fair characterization.

>Red Hat is claiming ownership of gdb?
>Red Hat released an unsupported version of gcc?
>Red Hat claims to have invented free software?
>Red Hat is a corporation that sometimes does stupid things and must
>therefore be punished?

>I guess it is probably all of the above

I think that's inaccurate, it is none of the above. Rather, we want to
avoid any possible impression of undue influence by a single company.
It's more like

Red Hat is very big and successful, and we want to avoid the impression
that they own and control the software in question. In our discussions
in the GDB steering committee (more accurately in meetings of its
precursor group), we all agreed that we wanted to work to avoid 
creating such an impression not just for Red Hat, but for any
company. 

Now of course everyone recognizes that the contributions of Red Hat/Cygnus
are huge and invaluable, and also everyone recognizes that we have to work
out ways to achieve the goal of avoiding incorrect impressions without
compromising the ability of RH/C (or any other company or individual) to
work smoothly.

Moving the list seems to be a decision that can help in this goal to many
of us. Yes, there is some initial effort in the move, but it does not seem
that anyone is going to be seriously inconvenienced, and there is some
small but significant gain in the move.

We all have to be sensitive to the fact that gcc/gdb/gnu-linux... are now
center stage and of great commercial importance to many companies. That 
means that it takes more work to smoothly work together.

At one meeting of the GDB group, one of the folks from Cygnus noted that
they were "suspicious" of Ada Core Technologies. What they meant was that
they regarded ACT as a potential competitor in some areas, and, although
we have not really competed head on to any great extent so far, yes, it
is possible that the potential for competition exists, and of course at
one level, competition is a good thing which can help everyone. However
it can also complicate things, because cooperation between competitors
has to be managed carefully. In particular, at the time, I had played
the role of convening the GDB meetings, but, sensitive to Red Hat
concerns, I suggested that Stan Shebs take the role of convenor (at
that time he was not associated with any company specifically). Later
it turned out that Stan joined Apple, which per se, would have been
no problem [remember we are all acting as individuals looking out for
the interests of the GNU project on such a committee, NOT as
representatives of our respective companies). However, not surprisingly
the move left Stan super busy, and so we agreed for me to be renamed
as convenor. I think that will be fine, it is not as though the convenor
has some hugely influential power in a group like this :-) More a matter
of just trying to move things along occasionally, and as I noted above,
I am acting as Robert Dewar, not as a representative of Ada Core 
Technologies. I am of course sensitive to possible concerns that come
from my position as CEO of ACT, and will try to ensure that this is
simply not an issue that bothers any one. It is after all inevitable
at this stage that a lot of people who are interested in GNU software
are people working for companies for which GNU software is commercially
critical. We cannot get away from that.

So, speaking as someone relatively new to this list, I would ask people
to try to stay as calm as possible in these discussions, and to understand
that no one is being "pissed on", or punished. No one is being angry in
any way at anyone else. No one is trying to gain unfair advantage. No one
thinks that Red Hat is in any way doing something wrong. It is just that
we want to work on avoiding the *appearence* of any one company having
special connections.

Yes, moving the list, or other similar changes we make in the future may
cause some extra work, and may particularly cause Red Hat some extra work.
We can't avoid that completely, but we certainly want to understand any
case in which this extra work is impractical or excessive, or where a
proposed change would cause Red Hat (or other company or individual)
real grief.

I hope this makes things a bit clearer. Perhaps others can explain the
situation better than me.

Robert Dewar

Re: New discussion list.

[Stan Shebs]

Andrew Cagney wrote:
> 
> Stan Shebs wrote:
> >
> > So the theory of this list (as I understand it) is to be able to
> > run generic GDB discussion.  If so, we now need to move list
> > members over en masse, set up forwarding, and ideally copy over
> > the old archives and moosh them into the mailman format.
> >
> > Presumably we can do all the setup and test it a bit before
> > turning off gdb@sources?
> 
> Firstly I've not seen a clear decision from the GDB Maintainer regarding
> the question gdb@gnu.org VS gdb@gdb.gnu.org.  Perhaphs I should ask for
> a re-count.

:-)  Yes, there's another dozen people on the steering committee who
*should* be weighing in on this, hint hint guys...

For my own part, I think moving the discussion list to gnu.org is a
reasonable step.  It gives us a chance to try out gnu.org's quality
as a host, since unreliability is not a disaster (as it would be for,
say, CVS), and it's relatively easy to reverse if necessary.

> Secondly, I have real reservations over simply handing over the mailing
> list gdb@sources.redhat.com.  Refering to sources.redhat.com's privicy
> statement: ``We'd never use any of the mailing list subscription
> information for anything other than the normal distribution of e-mail to
> that list.''.  I'm honestly sorry about this.  In my opinion, people
> will need to be asked to make the move.

Interesting point - I had always considered Cygnus and Red Hat to be
managing the list on behalf of the FSF, so in a sense the FSF "owns"
the information.  But it's not something that's well defined.  (Quick,
to the Florida courts! :-) )

I think it should suffice to inform list members that it will be moving,
give them a week or two to unsubscribe if they're really worried about
the FSF selling their address to spammers. :-)  I suspect that many
will be surprised to learn that their addresses weren't already being
managed by the FSF...

Stan

Re: New discussion list.

[Jonathan Larmour]

[ Consider me just an individual who occasionally contributes to gdb as a
result of working with GDB, not on it, i.e. I'm not a Red Hat GDB team
member. ]

Robert Dewar wrote:
> Red Hat is very big and successful, and we want to avoid the impression
> that they own and control the software in question. In our discussions
> in the GDB steering committee (more accurately in meetings of its
> precursor group), we all agreed that we wanted to work to avoid
> creating such an impression not just for Red Hat, but for any
> company.

So why is gdb.gnu.org unacceptable for this purpose? Why do the lists have
to physically moved rather than virtually moved? No-one has yet answered
that in anything I've seen here or elsewhere.

> Now of course everyone recognizes that the contributions of Red Hat/Cygnus
> are huge and invaluable, and also everyone recognizes that we have to work
> out ways to achieve the goal of avoiding incorrect impressions without
> compromising the ability of RH/C (or any other company or individual) to
> work smoothly.

Given where the majority of GDB work comes from, I can't help but think
that it throws obstacles in the way of those who contribute the most,
*while not removing any from anyone else*.

> Moving the list seems to be a decision that can help in this goal to many
> of us. Yes, there is some initial effort in the move, but it does not seem
> that anyone is going to be seriously inconvenienced, and there is some
> small but significant gain in the move.

I think reliability of gnu.org has always been a sore point. If this is no
longer true I sit corrected; but if so, is political dogma more important
than having something that works?

That's all I really care about - what we have now works, and I have doubts
that a gnu.org replacement would be as fast or reliable as sources. Putting
it another way: personally I would feel happier with a move to sourceforge
than gnu.org. Can someone dissuade me of my doubts about gnu.org without
merely sounding optimistic about the future?

Elsewhere you write:
> So the discussion is definitely legitimate,
> but it is helpful to keep it calm and non-personal (especially where
> the persons are giant companies :-)

Be careful: do not confuse an e-mail address with a role. I, and no doubt
most others, are speaking as individuals. I don't care what my employer
thinks about what I write here.

Jifl

Re: New discussion list.

[Alexandre Oliva]

On Nov 28, 2000, Richard Stallman <rms@gnu.org> wrote:

>     The rumours I keep hearing are directed at the current systems.  The
>     most recent being that the autoconf mailing list stopping dead for a few
>     hours (~9:20UCT and 16:53UCT) on Nov 22.

> I can investigate why this happened, but it looks like a fairly small
> problem that was fixed reasonably fast.

Actually, messages only started to show up on Wednesday, but they had
been posted on Monday morning, French time.

> You seem to be trying desperately to find fault with our servers.

Well, if you think so, here's one more ``desperate try''.  The CVS
server is amazingly slow.  A single commit in a small file in autoconf
or libtool takes forever (ask Akim), whereas committing even large
files in gcc.gnu.org works reasonably fast (even though I'm on the
other end of a very slow network connection).  I can't understand why
it's that slow.  Maybe it's on the same network link as Sourceforge?
I also observe very long commit times when checking in stuff in
amanda.sourceforge.net.

> That is not the right way to treat other people who work on the GNU
> Project, so please don't do it any more.

Do you think forcing people to use a machine on a slower network
connection to do their volunteer work is the right way to treat them?

We do have a gcc.gnu.org machine with a reasonably fast network link.
Well, Richard Kenner feels otherwise, but there's no indication that
moving to subversions.gnu.org would improve anything for him or
anybody else.

Moving GDB would just make it harder to have the unified CVS tree lots
of people have been asking for.  I'd rather keep everything in the
current machine, maybe rename it to dev.gnu.org (keeping gcc.gnu.org
as an alias and using virtual WWW hosting so that people opening
gcc.gnu.org would get to the right web-page), and create the
long-awaited unified tree in it.  *That* would be a step in the right
direction, IMO.

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me

Re: New discussion list.

[Andrew Cagney]

Richard Stallman wrote:
> 
>     The rumours I keep hearing are directed at the current systems.  The
>     most recent being that the autoconf mailing list stopping dead for a few
>     hours (~9:20UCT and 16:53UCT) on Nov 22.
> 
> I can investigate why this happened, but it looks like a fairly small
> problem that was fixed reasonably fast.

I'm sorry but 7 hours to have a problem like this resolved is not
reasonably fast.

> You seem to be trying desperately to find fault with our servers.
> That is not the right way to treat other people who work on the GNU
> Project, so please don't do it any more.

I'm not trying to be disrespectful towards the people who maintain the
GNU servers.  They do a very good job with the limited resources they
have available.

In the case of the above I'm assuming that the problem was indeed
resolved promptly and professionally once the required resources were
available.

	Andrew

Re: New discussion list.

[Richard Stallman]

    Err, you seem to have forgotten my recent problems accessing FSF machines
    via kerberos (within the last month).

    >From what I can tell, when the FSF converted to kerberos a set of machines
    as publicized as the right machines to connect to.  Then one day, they all
    went off the air.

That was a couple of months ago, when we turned off the old
inaccessible mail and file servers, and turned on the new one
(fencepost) that we can maintain better.  We sent an announcement to
gnu-prog, a list that all GNU maintainers should be on, but you were
not on the list.  Once you contacted us, everything was straightened
out quickly.

Re: New discussion list.

[Richard Stallman]

    The rumours I keep hearing are directed at the current systems.  The
    most recent being that the autoconf mailing list stopping dead for a few
    hours (~9:20UCT and 16:53UCT) on Nov 22.

I can investigate why this happened, but it looks like a fairly small
problem that was fixed reasonably fast.

You seem to be trying desperately to find fault with our servers.
That is not the right way to treat other people who work on the GNU
Project, so please don't do it any more.

Re: New discussion list.

[Jeffrey A Law]

  In message < 3A22976A.46AAC43B@shebs.cnchost.com >you write:
  > Jeffrey A Law wrote:
  > One of the things that strikes about this discussion is that there
  > are two separate issues that have gotten tied together.  One is
  > the hosting and name mapping, and the other is the question of
  > who maintains the hosts.
Right.

  > For instance, suppose one were to transplant all the existing software
  > bits on sources over to a gnu.org machine, and give all the people on
  > sources the same degree of access at gnu.org.  Of course there would
  > still be some temporary disruption, but is this a satisfactory long-term
  > situation?  If not, why not?
It would be acceptable as far as I'm concerned -- as long as the FSF folks
continue to make progress on the reliability of their machines.  Contrary
to RMS's assertions there have been significant outages and problems over
the last few months, though things are better than they were a year+ ago.

Not to say that Red Hat hasn't had outages (we have), but I believe our
track record for providing reliable service over the last 3 years speaks
for itself.

  > Conversely, what is the disadvantage for the GNU project of having
  > some of its machines be seen to be run by Red Hat if one looks hard
  > enough?  Many different corporations donate to GNU, either in time
  > or money, and yet we don't worry that GNU somehow becomes beholden
  > to them as a result.  Red Hat is basically offering, for certain
  > projects, better hosting services than GNU can normally provide,
  > and they're also offering to give up corporate visibility if any
  > is discovered to leak through.  Is there something about the physical
  > hosting of gnu.org that is important, and if so, what is it?
Furthermore, how does the answer to the physical hosting question change
if the machine isn't physically at one of Red Hat's sites?  Red Hat
is planning to physically move the machine to a co-location site to
greatly increase network bandwidth.

jeff

Re: New discussion list.

[Stan Shebs]

Jeffrey A Law wrote:
> 
>   > Experience with the machine gcc.gnu.org shows that the actual nature
>   > of the machine can be seen by users in various ways.  These servers
>   > support many different protocols.
> And we've stated that we'll fix them.  Jeez.  We know the worst one right
> now is ftp service isn't virtualized.  We're going to fix it.

One of the things that strikes about this discussion is that there
are two separate issues that have gotten tied together.  One is
the hosting and name mapping, and the other is the question of
who maintains the hosts.

For instance, suppose one were to transplant all the existing software
bits on sources over to a gnu.org machine, and give all the people on
sources the same degree of access at gnu.org.  Of course there would
still be some temporary disruption, but is this a satisfactory long-term
situation?  If not, why not?

Conversely, what is the disadvantage for the GNU project of having
some of its machines be seen to be run by Red Hat if one looks hard
enough?  Many different corporations donate to GNU, either in time
or money, and yet we don't worry that GNU somehow becomes beholden
to them as a result.  Red Hat is basically offering, for certain
projects, better hosting services than GNU can normally provide,
and they're also offering to give up corporate visibility if any
is discovered to leak through.  Is there something about the physical
hosting of gnu.org that is important, and if so, what is it?

Stan

Re: New discussion list.

[Jeffrey A Law]

  In message < 200011252219.PAA21460@wijiji.santafe.edu >you write:
  >     - and most importantly, unless something recently has changed, gnu.org
  >     machines are not renowned for being reliable.
  > 
  > That reputation is based on things that happened on other machines (no
  > longer used) which were kept at an inconvenient location where they
  > were often not physically accessible to anyone.  However, people keep
  > repeating the old information that "the GNU servers are not reliable".
  > It looks like this will continue to propagate, regardless of what
  > changes, until more people have actual experience with the current
  > situation.
Err, you seem to have forgotten my recent problems accessing FSF machines
via kerberos (within the last month).

From what I can tell, when the FSF converted to kerberos a set of machines
as publicized as the right machines to connect to.  Then one day, they all
went off the air.

Kerberos rlogin to the new server (fencepost) didn't work, while kerberos
telnet did work.  Rather odd and extremely inconvenient since I almost never
use kerberos telnet.

  > Experience with the machine gcc.gnu.org shows that the actual nature
  > of the machine can be seen by users in various ways.  These servers
  > support many different protocols.
And we've stated that we'll fix them.  Jeez.  We know the worst one right
now is ftp service isn't virtualized.  We're going to fix it.


jeff

Re: New discussion list.

[Andrew Cagney]

Richard Stallman wrote:
> 
>     - and most importantly, unless something recently has changed, gnu.org
>     machines are not renowned for being reliable.
> 
> That reputation is based on things that happened on other machines (no
> longer used) which were kept at an inconvenient location where they
> were often not physically accessible to anyone.  However, people keep
> repeating the old information that "the GNU servers are not reliable".
> It looks like this will continue to propagate, regardless of what
> changes, until more people have actual experience with the current
> situation.

The rumours I keep hearing are directed at the current systems.  The
most recent being that the autoconf mailing list stopping dead for a few
hours (~9:20UCT and 16:53UCT) on Nov 22.  Grubbing around and it would
appear that this was indeed the case.

	Andrew

http://mail.gnu.org/pipermail/autoconf-patches/2000-November/006339.html
http://sources.redhat.com/ml/autoconf-patches/2000-11/msg00223.html

Re: New discussion list.

[Christopher Faylor]

On Sat, Nov 25, 2000 at 10:33:45PM +0000, Jonathan Larmour wrote:
>Richard Stallman wrote:
>> Experience with the machine gcc.gnu.org shows that the actual nature
>> of the machine can be seen by users in various ways.  These servers
>> support many different protocols.
>
>I know of no user-visible way, other than deliberately looking to find the
>identity of the machine, e.g. via traceroute etc, and that is not relevant
>for your average user. Since the gcc folks are trying hard to clear out
>such references, then anything user-visible is really a problem that needs
>fixing. What are you thinking of?

The Received headers in mail messages will display the
sources.redhat.com address.  I don't think think there is any around
this since I believe that they are generated on the machine receiving
the email.

I think we may still have problems with GNATS displaying
"sources.redhat.com" in some places, too, but we're working on that.

One thing that we are discussing is changing the name of
sources.redhat.com to something that is vendor neutral.  The trick, of
course, is coming up with a domain name that is catchy and mnemonic.
That's almost as hard as the technical aspects of changing the name.

The reality of sources.redhat.com is that Red Hat corporate really only
takes note of its existence when they notice the bill for the T1 line.
Since this site started out as a Cygnus system, it doesn't seem to have
reached the awareness of many people in the home office.

Oddly enough, however, whenever the concept of this system is explained
there is never any problem with justifying its ongoing expense.  I
wasn't involved in its creation, but I believe that this system was
intended purely to be a donation to the free software community.  That's
how I always portray it, anyway, and no one has ever had any problems
with that.

So, I don't believe that changing the domain name will be a big deal.
IMO, we'll probably be changing the domain name of this system
regardless of the outcome of any move to gnu.org.  I have personal
reasons for wanting to do this.  A move to a new domain name should stop
the flow of email to sourcemaster regarding problems (real and
imaginary) with Red Hat's GNU/Linux releases.  Not having to explain
that we have nothing to do with this every other day will make a switch
to a new domain name a blessing for me.  :-)

cgf

Re: New discussion list.

[Jonathan Larmour]

Richard Stallman wrote:
> 
>     - and most importantly, unless something recently has changed, gnu.org
>     machines are not renowned for being reliable.
> 
> That reputation is based on things that happened on other machines (no
> longer used) which were kept at an inconvenient location where they
> were often not physically accessible to anyone.  However, people keep
> repeating the old information that "the GNU servers are not reliable".
> It looks like this will continue to propagate, regardless of what
> changes, until more people have actual experience with the current
> situation.

I chose my words carefully, and in other mails solicited people's opinions
on whether it was still the case. Thank you for clarifying.

>     But gdb.gnu.org would fully give the impression of being a GNU machine.
> 
> Experience with the machine gcc.gnu.org shows that the actual nature
> of the machine can be seen by users in various ways.  These servers
> support many different protocols.

I know of no user-visible way, other than deliberately looking to find the
identity of the machine, e.g. via traceroute etc, and that is not relevant
for your average user. Since the gcc folks are trying hard to clear out
such references, then anything user-visible is really a problem that needs
fixing. What are you thinking of?

Jifl

Re: New discussion list.

[Richard Stallman]

    - and most importantly, unless something recently has changed, gnu.org
    machines are not renowned for being reliable.

That reputation is based on things that happened on other machines (no
longer used) which were kept at an inconvenient location where they
were often not physically accessible to anyone.  However, people keep
repeating the old information that "the GNU servers are not reliable".
It looks like this will continue to propagate, regardless of what
changes, until more people have actual experience with the current
situation.

Moving the GDB discussion list, repository, and web pages will show
people that our new servers are reliable.

    But gdb.gnu.org would fully give the impression of being a GNU machine.

Experience with the machine gcc.gnu.org shows that the actual nature
of the machine can be seen by users in various ways.  These servers
support many different protocols.

Re: New discussion list.

[Jonathan Larmour]

Robert Dewar wrote:
> 
> <<Do you think that because ACT derive benefit from holding the "master"
> repository to GNU ADA (all changes must go through ACT and into their
> repository first), that this is similarly bad? In many respects that is a
> more extreme situation than with GDB. Do you remember the arguments used to
> >>>
> 
> Just to note here, we have no requirement that "all changes" must go
> through ACT. The idea of setting up a publoic tree is precisely so
> that the process can be opened up.

Sorry yes, I have reread the thread on gcc, and I think I suffered the same
confusion as some others probably did (mostly from lack of background).
I've now worked out your intent! What I said above was bogus.

Jifl
-- 
Red Hat, 35 Cambridge Place, Cambridge, UK. CB2 1NS  Tel: +44 (1223) 728762
"Plan to be spontaneous tomorrow."  ||  These opinions are all my own fault

Re: New discussion list.

[Robert Dewar]

<<Do you think that because ACT derive benefit from holding the "master"
repository to GNU ADA (all changes must go through ACT and into their
repository first), that this is similarly bad? In many respects that is a
more extreme situation than with GDB. Do you remember the arguments used to
>>>

Just to note here, we have no requirement that "all changes" must go
through ACT. The idea of setting up a publoic tree is precisely so
that
the process can be opened up.

Re: New discussion list.

[Richard Stallman]

      Mooshing may not be necessary, we can html-ize the archive
    and provide a link from the new archive.

It is ok to continue to refer to the old mail archive where it is now;
that is one thing which is not crucial to move.  However, it is also
ok to move the archive.  (Mail archives are one of the things that
we don't try to always have on gnu.org machines.)

Re: New discussion list.

[Jonathan Larmour]

Robert Dewar wrote:
> 
> <<Given where the majority of GDB work comes from, I can't help but think
> that it throws obstacles in the way of those who contribute the most,
> *while not removing any from anyone else*.
> >>
> 
> To me that's a worrisome argument, it is close to arguing that Red Hat
> should get favored treatment *because* it is "where the majority of GDBwork comes from"
> 
> I understand what you are saying, but you should see that it is precisely
> this argument that suggests avoiding any hint of favored treatment.

And I understand what you are saying :-), but my point is that if moving
the list would make things equally good for everyone that would be
sensible. But an argument that suggests things should be equally *bad* for
everyone out of principle is not very practical nor constructive. It would
sound like dogma.

Do you think that because ACT derive benefit from holding the "master"
repository to GNU ADA (all changes must go through ACT and into their
repository first), that this is similarly bad? In many respects that is a
more extreme situation than with GDB. Do you remember the arguments used to
justify that situation? [ I don't want to repeat the thread here by
explaining to everyone and reiterating here what was said - treat it as a
rhetorical question :) ]

> Certainly we do not want to throw obstacles in any one's way. Can you
> explain clearly *why* moving the mailing list would have that effect.

- Moving management of lists away from many GDB maintainer's/contributor's
influence (one notable person involved with GDB, for example, has root
access; several have shell access). This is related to...

- response time to problems. The ability to do things yourself, or tell
someone close by what to do will always be better than e-mailing a remote
volunteer sysadmin with no involvement or personal interest in GDB.

- integration with the rest of the infrastructure (think about things like
integration betwen GNATS, mailing list archives (especially gdb-patches),
CVS, and automatic test infrastructure, for example);

- related to that type of thing, customizability - sources, is very liberal
with what you can do, and has an infrastructure good enough to handle it...
it may be an opinion, but I cannot visualize the same flexibility on a
gnu.org machine. I'm happy to be wrong, so please tell me that they really
are this flexible.

- and most importantly, unless something recently has changed, gnu.org
machines are not renowned for being reliable. If the SC really advocates
moving somewhere known to be less reliable (if it truly is so!) then GDB
users would rightly feel disenfranchised, and the SC would lose respect
before it's had a chance to prove its value.

All but one of these points (integration) could be irrelevant if gnu.org
runs a very different regime from what I believe it does. But integration
is a bit of a stumbling block that could only be resolved by running most
services on the same machine, wherever it is located.

But gdb.gnu.org would fully give the impression of being a GNU machine. Who
cares where it really is?

I probably won't post any more as I already have several questions needing
answering before I could really 100% advocate one way or the other - I'm
far from set against the idea; but I have many doubts that want pacifying
first. And the opposition to gdb.gnu.org is also a mystery.

Jifl

Re: New discussion list.

[Robert Dewar]

<<Given where the majority of GDB work comes from, I can't help but think
that it throws obstacles in the way of those who contribute the most,
*while not removing any from anyone else*.
>>

To me that's a worrisome argument, it is close to arguing that Red Hat
should get favored treatment *because* it is "where the majority of GDBwork comes from"

I understand what you are saying, but you should see that it is precisely
this argument that suggests avoiding any hint of favored treatment.

Certainly we do not want to throw obstacles in any one's way. Can you
explain clearly *why* moving the mailing list would have that effect. 

Re: New discussion list.

[Robert Dewar]

<<Be careful: do not confuse an e-mail address with a role. I, and no doubt
most others, are speaking as individuals. I don't care what my employer
thinks about what I write here.
>>

You misunderstood what I was saying, I was not saying that you were speaking
*for* Red Hat, but you were certainly speaking *about* Red Hat, and as I say,
it is a good idea to keep such discussions calm and non-personal!

Re: New discussion list.

[Jonathan Larmour]

[ Consider me just an individual who occasionally contributes to gdb as a
result of working with GDB, not on it, i.e. I'm not a Red Hat GDB team
member. ]

Robert Dewar wrote:
> Red Hat is very big and successful, and we want to avoid the impression
> that they own and control the software in question. In our discussions
> in the GDB steering committee (more accurately in meetings of its
> precursor group), we all agreed that we wanted to work to avoid
> creating such an impression not just for Red Hat, but for any
> company.

So why is gdb.gnu.org unacceptable for this purpose? Why do the lists have
to physically moved rather than virtually moved? No-one has yet answered
that in anything I've seen here or elsewhere.

> Now of course everyone recognizes that the contributions of Red Hat/Cygnus
> are huge and invaluable, and also everyone recognizes that we have to work
> out ways to achieve the goal of avoiding incorrect impressions without
> compromising the ability of RH/C (or any other company or individual) to
> work smoothly.

Given where the majority of GDB work comes from, I can't help but think
that it throws obstacles in the way of those who contribute the most,
*while not removing any from anyone else*.

> Moving the list seems to be a decision that can help in this goal to many
> of us. Yes, there is some initial effort in the move, but it does not seem
> that anyone is going to be seriously inconvenienced, and there is some
> small but significant gain in the move.

I think reliability of gnu.org has always been a sore point. If this is no
longer true I sit corrected; but if so, is political dogma more important
than having something that works?

That's all I really care about - what we have now works, and I have doubts
that a gnu.org replacement would be as fast or reliable as sources. Putting
it another way: personally I would feel happier with a move to sourceforge
than gnu.org. Can someone dissuade me of my doubts about gnu.org without
merely sounding optimistic about the future?

Elsewhere you write:
> So the discussion is definitely legitimate,
> but it is helpful to keep it calm and non-personal (especially where
> the persons are giant companies :-)

Be careful: do not confuse an e-mail address with a role. I, and no doubt
most others, are speaking as individuals. I don't care what my employer
thinks about what I write here.

Jifl

Re: New discussion list.

[Robert Dewar]

<<There can hardly be any anti-Red Hat sentiment from me, for one
thing I still own stock in it!  There are GDB committee members
whose companies compete with RH though, probably hard for them
not to think a negative thought about RH from time to time. :-)
>>

Speaking as a "GDB committee member whose company [might] complete with
RH". I don't have negative thoughts about RH. I always welcome competition
providing it is fair and above board, something which I think we achieve
in our community better than a lot of other companies. The odd thing
here is that in the open development model for Free Software, we all
depend on our competitors to help us. That's a bit odd, and would
puzzle the Ford's and Rockefeller's of the world perhaps, but it works
out amazing well. Part of what we are doing here is working ahead of
time to avoid any problems that come from this unusual cooperation.
RMS can help us in this goal, by providing a guiding influence that
is definitely NOT associated with any of the companies involved,
and I find his encouragement to be sensitive to appearences of
excessive connection of stuff to individual companies to be 
helpful!

Given Red Hat's central and critical contributions, it is also very
important to make sure that Red Hat is happy with how things proceed
(that's just a special case of it being important that everyone be
happy with the setup). So the discussion is definitely legitimate,
but it is helpful to keep it calm and non-personal (especially where
the persons are giant companies :-)

Robert Dewar

Re: New discussion list.

[Robert Dewar]

>I am aware that there is an anti-Red Hat sentiment.  Why not make it
>perfectly clear so that we can all understand it.

I think that's not a fair characterization.

>Red Hat is claiming ownership of gdb?
>Red Hat released an unsupported version of gcc?
>Red Hat claims to have invented free software?
>Red Hat is a corporation that sometimes does stupid things and must
>therefore be punished?

>I guess it is probably all of the above

I think that's inaccurate, it is none of the above. Rather, we want to
avoid any possible impression of undue influence by a single company.
It's more like

Red Hat is very big and successful, and we want to avoid the impression
that they own and control the software in question. In our discussions
in the GDB steering committee (more accurately in meetings of its
precursor group), we all agreed that we wanted to work to avoid 
creating such an impression not just for Red Hat, but for any
company. 

Now of course everyone recognizes that the contributions of Red Hat/Cygnus
are huge and invaluable, and also everyone recognizes that we have to work
out ways to achieve the goal of avoiding incorrect impressions without
compromising the ability of RH/C (or any other company or individual) to
work smoothly.

Moving the list seems to be a decision that can help in this goal to many
of us. Yes, there is some initial effort in the move, but it does not seem
that anyone is going to be seriously inconvenienced, and there is some
small but significant gain in the move.

We all have to be sensitive to the fact that gcc/gdb/gnu-linux... are now
center stage and of great commercial importance to many companies. That 
means that it takes more work to smoothly work together.

At one meeting of the GDB group, one of the folks from Cygnus noted that
they were "suspicious" of Ada Core Technologies. What they meant was that
they regarded ACT as a potential competitor in some areas, and, although
we have not really competed head on to any great extent so far, yes, it
is possible that the potential for competition exists, and of course at
one level, competition is a good thing which can help everyone. However
it can also complicate things, because cooperation between competitors
has to be managed carefully. In particular, at the time, I had played
the role of convening the GDB meetings, but, sensitive to Red Hat
concerns, I suggested that Stan Shebs take the role of convenor (at
that time he was not associated with any company specifically). Later
it turned out that Stan joined Apple, which per se, would have been
no problem [remember we are all acting as individuals looking out for
the interests of the GNU project on such a committee, NOT as
representatives of our respective companies). However, not surprisingly
the move left Stan super busy, and so we agreed for me to be renamed
as convenor. I think that will be fine, it is not as though the convenor
has some hugely influential power in a group like this :-) More a matter
of just trying to move things along occasionally, and as I noted above,
I am acting as Robert Dewar, not as a representative of Ada Core 
Technologies. I am of course sensitive to possible concerns that come
from my position as CEO of ACT, and will try to ensure that this is
simply not an issue that bothers any one. It is after all inevitable
at this stage that a lot of people who are interested in GNU software
are people working for companies for which GNU software is commercially
critical. We cannot get away from that.

So, speaking as someone relatively new to this list, I would ask people
to try to stay as calm as possible in these discussions, and to understand
that no one is being "pissed on", or punished. No one is being angry in
any way at anyone else. No one is trying to gain unfair advantage. No one
thinks that Red Hat is in any way doing something wrong. It is just that
we want to work on avoiding the *appearence* of any one company having
special connections.

Yes, moving the list, or other similar changes we make in the future may
cause some extra work, and may particularly cause Red Hat some extra work.
We can't avoid that completely, but we certainly want to understand any
case in which this extra work is impractical or excessive, or where a
proposed change would cause Red Hat (or other company or individual)
real grief.

I hope this makes things a bit clearer. Perhaps others can explain the
situation better than me.

Robert Dewar

Re: New discussion list.

[Todd Whitesel]

> >I agree with RMS' goal here, but I also don't think it should be rushed
> >just so the announcement can pretend that no transition happened.
> 
> And, what is that goal?  If you are going to discuss something in a
> semi-public mailing list then provide details.  Otherwise please drop
> overseers@sources.redhat.com from your list.

Sorry, I am just group-replying. I shouldn't be the first email you've seen
about this, unless things arrived out of order (which does happen).

Rather than risk misrepresenting the discussion however, I will leave it to
others to explain what's going on here.

Todd Whitesel
toddpw @ best.com

Re: New discussion list.

[Phillip Rulon]

   Envelope-to: pjr@gnu.org
   From: Todd Whitesel <toddpw@best.com>
   Date: Wed, 22 Nov 2000 19:37:34 -0800 (PST)
   Cc: ac131313@cygnus.com, pjr@gnu.org, gdbheads@gnu.org,
	   overseers@sources.redhat.com
   Content-Type: text/plain; charset=US-ASCII

   > :-)  Yes, there's another dozen people on the steering committee who
   > *should* be weighing in on this, hint hint guys...

   Ok.

   I agree with RMS' goal here, but I also don't think it should be
   rushed just so the announcement can pretend that no transition
   happened.

   Couldn't the announcement name the current list and mention that
   there will be a transition in the near future?

I think it is probably a good idea to boot the new setup with the new
list.  We can, of course, import the old list at the same time.  Just
my opinion.

   It could be argued that then the transition will never happen, but
   if we can agree on the announcement then that commits us to
   eventually doing it, and might even draw a few volunteers out of
   the woodwork.

I think it's safe to say that the transition will happen, I've been
asked to make it so.  I don't have much to say about the announcement
but the list is kind of a done deal.

pjr

Re: New discussion list.

[Phillip Rulon]

   Envelope-to: pjr@gnu.org
   Sender: ac131313@localhost.cygnus.com
   Date: Thu, 23 Nov 2000 14:21:49 +1100
   From: Andrew Cagney <ac131313@cygnus.com>
   X-Accept-Language: en
   Cc: Phillip Rulon <pjr@gnu.org>, gdbheads@gnu.org,
	   overseers@sources.redhat.com
   Content-Type: text/plain; charset=us-ascii

   Stan Shebs wrote:

   > Interesting point - I had always considered Cygnus and Red Hat to be
   > managing the list on behalf of the FSF, so in a sense the FSF "owns"
   > the information.  But it's not something that's well defined.  (Quick,
   > to the Florida courts! :-) )

   That is the view that Red Hat liked to have.  However, according to
   http://mail.gnu.org/mailman/listinfo/bug-gdb for bug-gdb@gnu.org ``There
   are no other GNU mailing lists or gnUSENET newsgroups for GDB'' (well
   until gdb@gnu.org was created).

Our first bug, thanks, I'll fix it.

pjr

Re: New discussion list.

[Christopher Faylor]

On Wed, Nov 22, 2000 at 07:37:34PM -0800, Todd Whitesel wrote:
>> :-)  Yes, there's another dozen people on the steering committee who
>> *should* be weighing in on this, hint hint guys...
>
>Ok.
>
>I agree with RMS' goal here, but I also don't think it should be rushed
>just so the announcement can pretend that no transition happened.

And, what is that goal?  If you are going to discuss something in a
semi-public mailing list then provide details.  Otherwise please drop
overseers@sources.redhat.com from your list.

>Couldn't the announcement name the current list and mention that there
>will be a transition in the near future?
>
>It could be argued that then the transition will never happen, but if
>we can agree on the announcement then that commits us to eventually
>doing it, and might even draw a few volunteers out of the woodwork.

Volunteers with root access to sources.redhat.com and knowledge of the
mailing lists there?  Hmm.

cgf

Re: New discussion list.

[Christopher Faylor]

On Wed, Nov 22, 2000 at 04:16:08PM -0800, Stan Shebs wrote:
>For my own part, I think moving the discussion list to gnu.org is a
>reasonable step.  It gives us a chance to try out gnu.org's quality
>as a host, since unreliability is not a disaster (as it would be for,
>say, CVS), and it's relatively easy to reverse if necessary.

"relatively easy" == "I don't have to do it, I just have to ask someone
to do it".

>Interesting point - I had always considered Cygnus and Red Hat to be
>managing the list on behalf of the FSF, so in a sense the FSF "owns"
>the information.  But it's not something that's well defined.  (Quick,
>to the Florida courts! :-) )

If many are under the impression that the mailing lists are managed by
the FSF, doesn't that sort of invalidate the whole reason for moving the
lists?

Not that the slightest hint of a rationale has surfaced in this discussion,
of course.

cgf

Re: New discussion list.

[Todd Whitesel]

> :-)  Yes, there's another dozen people on the steering committee who
> *should* be weighing in on this, hint hint guys...

Ok.

I agree with RMS' goal here, but I also don't think it should be rushed
just so the announcement can pretend that no transition happened.

Couldn't the announcement name the current list and mention that there
will be a transition in the near future?

It could be argued that then the transition will never happen, but if we
can agree on the announcement then that commits us to eventually doing it,
and might even draw a few volunteers out of the woodwork.

Todd Whitesel
toddpw @ best.com

Re: New discussion list.

[Andrew Cagney]

Stan Shebs wrote:

> Interesting point - I had always considered Cygnus and Red Hat to be
> managing the list on behalf of the FSF, so in a sense the FSF "owns"
> the information.  But it's not something that's well defined.  (Quick,
> to the Florida courts! :-) )

That is the view that Red Hat liked to have.  However, according to
http://mail.gnu.org/mailman/listinfo/bug-gdb for bug-gdb@gnu.org ``There
are no other GNU mailing lists or gnUSENET newsgroups for GDB'' (well
until gdb@gnu.org was created).

> I think it should suffice to inform list members that it will be moving,
> give them a week or two to unsubscribe if they're really worried about
> the FSF selling their address to spammers. :-)  I suspect that many
> will be surprised to learn that their addresses weren't already being
> managed by the FSF...

I'm sure of that as well.  I think there are some very efficient ways of
getting people to transfer across with out actually transfering the
list.  If the FSF don't consider the list theirs then a simple mail out
declaring the existing list dead with a URL for how to subscribe to the
new list would be the obvious answer.  If the FSF do actually consider
the list theirs then giving everyone the oportunity to opt out and then
transfer things would be better.

	Andrew

Re: New discussion list.

[Stan Shebs]

Andrew Cagney wrote:
> 
> Stan Shebs wrote:
> >
> > So the theory of this list (as I understand it) is to be able to
> > run generic GDB discussion.  If so, we now need to move list
> > members over en masse, set up forwarding, and ideally copy over
> > the old archives and moosh them into the mailman format.
> >
> > Presumably we can do all the setup and test it a bit before
> > turning off gdb@sources?
> 
> Firstly I've not seen a clear decision from the GDB Maintainer regarding
> the question gdb@gnu.org VS gdb@gdb.gnu.org.  Perhaphs I should ask for
> a re-count.

:-)  Yes, there's another dozen people on the steering committee who
*should* be weighing in on this, hint hint guys...

For my own part, I think moving the discussion list to gnu.org is a
reasonable step.  It gives us a chance to try out gnu.org's quality
as a host, since unreliability is not a disaster (as it would be for,
say, CVS), and it's relatively easy to reverse if necessary.

> Secondly, I have real reservations over simply handing over the mailing
> list gdb@sources.redhat.com.  Refering to sources.redhat.com's privicy
> statement: ``We'd never use any of the mailing list subscription
> information for anything other than the normal distribution of e-mail to
> that list.''.  I'm honestly sorry about this.  In my opinion, people
> will need to be asked to make the move.

Interesting point - I had always considered Cygnus and Red Hat to be
managing the list on behalf of the FSF, so in a sense the FSF "owns"
the information.  But it's not something that's well defined.  (Quick,
to the Florida courts! :-) )

I think it should suffice to inform list members that it will be moving,
give them a week or two to unsubscribe if they're really worried about
the FSF selling their address to spammers. :-)  I suspect that many
will be surprised to learn that their addresses weren't already being
managed by the FSF...

Stan

Re: New discussion list.

[Andrew Cagney]

Stan Shebs wrote:
> 
> So the theory of this list (as I understand it) is to be able to
> run generic GDB discussion.  If so, we now need to move list
> members over en masse, set up forwarding, and ideally copy over
> the old archives and moosh them into the mailman format.
> 
> Presumably we can do all the setup and test it a bit before
> turning off gdb@sources?

Firstly I've not seen a clear decision from the GDB Maintainer regarding
the question gdb@gnu.org VS gdb@gdb.gnu.org.  Perhaphs I should ask for
a re-count.

Secondly, I have real reservations over simply handing over the mailing
list gdb@sources.redhat.com.  Refering to sources.redhat.com's privicy
statement: ``We'd never use any of the mailing list subscription
information for anything other than the normal distribution of e-mail to
that list.''.  I'm honestly sorry about this.  In my opinion, people
will need to be asked to make the move.

With regard to archives, they are all available for download so that
isn't a problem.

	Andrew

Re: New discussion list.

[Jason Molenda]

On Wed, Nov 22, 2000 at 12:19:06PM -0800, Stan Shebs wrote:
> So the theory of this list (as I understand it) is to be able to
> run generic GDB discussion.  If so, we now need to move list
> members over en masse, set up forwarding, and ideally copy over
> the old archives and moosh them into the mailman format.
> 
> Presumably we can do all the setup and test it a bit before
> turning off gdb@sources?



Whoever does the move, this is an identical repeat of the guile list
move in September.  My note outlining the process is here:

	http://sources.redhat.com/ml/overseers/2000-q3/msg00300.html

Follow the thread for other notes on the subject.  You should be able
to refer to the current guile list setup for tips on exactly how a
moved list should look in the end.


Jason

Re: New discussion list.

[Phillip Rulon]

   Envelope-to: pjr@gnu.org
   Date: Wed, 22 Nov 2000 12:19:06 -0800
   From: Stan Shebs <shebs@apple.com>
   X-Accept-Language: en
   CC: gdbheads@gnu.org, overseers@sources.redhat.com
   Content-Type: text/plain; charset=us-ascii

   So the theory of this list (as I understand it) is to be able to
   run generic GDB discussion.

I'm open to theory ideas, I just joined at the request of RMS.  I'm
in the interesting position of running the list while not being a
recognized gdb-hacker.  I know enough about the system setup at GNU
to make the list useful but I'll defer to this list for policy.

   If so, we now need to move list members over en masse, set up
   forwarding, and ideally copy over the old archives and moosh them
   into the mailman format.

All doable.  Mooshing may not be necessary, we can html-ize the archive
and provide a link from the new archive.

   Presumably we can do all the setup and test it a bit before
   turning off gdb@sources?

I've done some testing, please do exercize it as much as you wish.

pjr

   Phillip Rulon wrote:
   > 
   > Greetings,
   >    There is a new mailing list, gdb@gnu.org, available for gdb
   > discussion.  The list administrator is currently me, if there
   > is someone else interested in doing it let me know.
   > 
   > The list is run under GNU/mailman.
   > 
   > List-Help:        < mailto:gdb-request@gnu.org?subject=help >
   > List-Post:        < mailto:gdb@gnu.org >
   > List-Subscribe:   < http://mail.gnu.org/mailman/listinfo/gdb >,
   >                   < mailto:gdb-request@gnu.org?subject=subscribe >
   > List-Unsubscribe: < http://mail.gnu.org/mailman/listinfo/gdb >,
   >                   < mailto:gdb-request@gnu.org?subject=unsubscribe >
   > List-Archive:     < http://mail.gnu.org/pipermail/gdb/ >
   > 
   > No one from gdbheads has been added thus far.  Questions to
   > pjr@gnu.org
   > 
   > Thanks,
   > pjr

Re: New discussion list.

[Stan Shebs]

So the theory of this list (as I understand it) is to be able to
run generic GDB discussion.  If so, we now need to move list
members over en masse, set up forwarding, and ideally copy over
the old archives and moosh them into the mailman format.

Presumably we can do all the setup and test it a bit before
turning off gdb@sources?

Stan

Phillip Rulon wrote:
> 
> Greetings,
>    There is a new mailing list, gdb@gnu.org, available for gdb
> discussion.  The list administrator is currently me, if there
> is someone else interested in doing it let me know.
> 
> The list is run under GNU/mailman.
> 
> List-Help:        < mailto:gdb-request@gnu.org?subject=help >
> List-Post:        < mailto:gdb@gnu.org >
> List-Subscribe:   < http://mail.gnu.org/mailman/listinfo/gdb >,
>                   < mailto:gdb-request@gnu.org?subject=subscribe >
> List-Unsubscribe: < http://mail.gnu.org/mailman/listinfo/gdb >,
>                   < mailto:gdb-request@gnu.org?subject=unsubscribe >
> List-Archive:     < http://mail.gnu.org/pipermail/gdb/ >
> 
> No one from gdbheads has been added thus far.  Questions to
> pjr@gnu.org
> 
> Thanks,
> pjr