cygwin-xfree

cygwin-xfree

[Chris Faylor]

I have recently offered CVS space to the the guy who has been working on
the port of XFree86 to Cygwin.

I thought he wanted it to upload the source code for a few drivers.

It turns out that he wants to upload pretty much all of the XFree86
source code since there were, apparently, a number of changes required
for Cygwin.

So, two question:

1) What does everyone think about adding another 150MB to the server.  There
   seems to be more than adequate disk space available for this but 150MB
   is still a pretty massive project.  I can just imagine what will happen
   to sourceware the first time someone checks this out.

2) I assume that if you all think that 1) is no problem then this directory
   should be winsup/xfree.  The winsup directory is not being mirrored by
   any of the CVS mirrors, right?

I'm have mixed feelings about this.  If we can't provide this, the developer
will go to Source Forge.  Maybe that's the best thing.  He's already split
the development that had started for cygwin so that there is just a generic
msvc/mingw version of everything as well as a cygwin-specific branch.  It
would probably be not too much work to set up on Source Forge and maybe
it would even give Cygwin a little publicity.

Any thoughts on this, anyone?

cgf

Re: cygwin-xfree

[Andrew Cagney]

Chris Faylor wrote:
> 
> I have recently offered CVS space to the the guy who has been working on
> the port of XFree86 to Cygwin.
> 
> I thought he wanted it to upload the source code for a few drivers.
> 
> It turns out that he wants to upload pretty much all of the XFree86
> source code since there were, apparently, a number of changes required
> for Cygwin.
> 
> So, two question:
> 
> 1) What does everyone think about adding another 150MB to the server.  There
>    seems to be more than adequate disk space available for this but 150MB
>    is still a pretty massive project.  I can just imagine what will happen
>    to sourceware the first time someone checks this out.
> 
> 2) I assume that if you all think that 1) is no problem then this directory
>    should be winsup/xfree.  The winsup directory is not being mirrored by
>    any of the CVS mirrors, right?
> 
> I'm have mixed feelings about this.  If we can't provide this, the developer
> will go to Source Forge.  Maybe that's the best thing.  He's already split
> the development that had started for cygwin so that there is just a generic
> msvc/mingw version of everything as well as a cygwin-specific branch.  It
> would probably be not too much work to set up on Source Forge and maybe
> it would even give Cygwin a little publicity.

Where is the public xfree86 CVS repository? Hmm:

http://www.xfree86.org/developer.html
> To become an XFree86 developer you first need to join The XFree86
> Project as a non-voting member. Access to the XFree86 development
> code is only available to project members, and being a member provides
> the necessary legal status to allow developers to access information
> available to the project from other sources. 

there isn't one.

	Andrew

Re: cygwin-xfree

[Chris Faylor]

On Thu, May 04, 2000 at 12:29:03PM +1000, Andrew Cagney wrote:
>Where is the public xfree86 CVS repository? Hmm:
>
> http://www.xfree86.org/developer.html
>> To become an XFree86 developer you first need to join The XFree86
>> Project as a non-voting member. Access to the XFree86 development
>> code is only available to project members, and being a member provides
>> the necessary legal status to allow developers to access information
>> available to the project from other sources. 
>
>there isn't one.

AFAIK, this isn't development code.  It should be released, publicly
available code.

I'm not 100% certain of that, though.

cgf

Re: cygwin-xfree

[Jason Molenda]

On Wed, May 03, 2000 at 10:18:05PM -0400, Chris Faylor wrote:

> 1) What does everyone think about adding another 150MB to the server.  There
>    seems to be more than adequate disk space available for this but 150MB
>    is still a pretty massive project.  I can just imagine what will happen
>    to sourceware the first time someone checks this out.


IMHO 150MB isn't very much - I way overspec'ed the CVS drive so
there is a lot of slack there.

(On the other hand, the web drive and the htdig drive are filling
at a rate faster than I had anticipated and won't last the ~18 month
lifetime I had intended for this system :-(


> 2) I assume that if you all think that 1) is no problem then this directory
>    should be winsup/xfree.  

It almost seems like something you'd want to put in a separate
repository.  I mean, it's _really_ Xfree86 with a few patches on
top (OK, a lot of patches).  How would you handle e.g. make or perl
(I'm trying to guess at programs that might have cygwin patches to
them) if someone wanted to do the same?


> The winsup directory is not being mirrored by any of the CVS mirrors, right?

If anyone is rsync'ing the entire src repository, they'll pick it
up, but it's not the end of the world for them to end up with
another ~150MB.  I would think about it more like "Where do I want
third party programs that we maintain patches to be checked in?"

Maybe a new repository, "cygwin-mods" or something, would be the cleanest
approach.  But cygwin-mods is a terrible name.


If he ever does get ready to start importing Xfree86, please make sure
he reads the CVS manual on tracking third party sources,
	http://www.loria.fr/~molli/cvs/doc/cvs_13.html#SEC98

or provide exact command lines for him.  It'd probably make sense
to import the most recent release of xfree86, then the development
version on which he's based his changes, and then check in his
changes on top of that.

That way if he dies in a tragic pingpong ball manufacturing plant accident,
people will be able to disentangle what he was up to.  It's really
easy to import sources incorrectly, and annoying to recover from that
at a later date.



> I'm have mixed feelings about this.  If we can't provide this, the developer
> will go to Source Forge.  

I don't have a strong opinion either, I did want to interject some
facts about the system and stuff, tho.



J

Re: cygwin-xfree

[Jim Kingdon]

> Where is the public xfree86 CVS repository? Hmm: . . .
> there isn't one.

Well, yes, and we (there is no we) would like them to open up.
Progress so far has been pretty slow although there seems to be
general claims that they are moving in the openness direction.  One
bright spot is that the Direct Rendering Infrastructure is at
http://dri.sourceforge.net/ - the Precision Insight folks are doing a
lot of X work, some of it with Red Hat funding.

As for what to tell the cygwin-xfree people, sounds good to me if they
take it to sourceforge.  Unless people see some big problem with that
(as it says on the main page, we are the Hosting Site To the World at
this time).

Re: cygwin-xfree

[Chris Faylor]

On Wed, May 03, 2000 at 07:44:27PM -0700, Jason Molenda wrote:
>> 2) I assume that if you all think that 1) is no problem then this directory
>>    should be winsup/xfree.  
>
>It almost seems like something you'd want to put in a separate
>repository.  I mean, it's _really_ Xfree86 with a few patches on
>top (OK, a lot of patches).  How would you handle e.g. make or perl
>(I'm trying to guess at programs that might have cygwin patches to
>them) if someone wanted to do the same?

Dunno.  I've been thinking that the winsup directory should be its
own little mini-build tree.  I didn't want to pollute the rest of
sourceware with stuff that was only intended for cygwin.

>> The winsup directory is not being mirrored by any of the CVS mirrors, right?
>
>If anyone is rsync'ing the entire src repository, they'll pick it
>up, but it's not the end of the world for them to end up with
>another ~150MB.  I would think about it more like "Where do I want
>third party programs that we maintain patches to be checked in?"
>
>Maybe a new repository, "cygwin-mods" or something, would be the cleanest
>approach.  But cygwin-mods is a terrible name.

Maybe cygwin-apps?

>If he ever does get ready to start importing Xfree86, please make sure
>he reads the CVS manual on tracking third party sources,
>	http://www.loria.fr/~molli/cvs/doc/cvs_13.html#SEC98
>
>or provide exact command lines for him.  It'd probably make sense
>to import the most recent release of xfree86, then the development
>version on which he's based his changes, and then check in his
>changes on top of that.

Yup.  I supplied the exact command lines for him.  And emphasized
repeatedly that he follow them exactly.  And he screwed them up...

>That way if he dies in a tragic pingpong ball manufacturing plant accident,
>people will be able to disentangle what he was up to.  It's really
>easy to import sources incorrectly, and annoying to recover from that
>at a later date.

I've repeatedly stressed this to him.  He spent all last night checking
everything in but somehow it never showed up on sourceware, or, at least
I can't find it.  He says that he understands why this happened.

>>I'm have mixed feelings about this.  If we can't provide this, the
>>developer will go to Source Forge.
>
>I don't have a strong opinion either, I did want to interject some
>facts about the system and stuff, tho.

Thanks, Jason.

cgf

Re: cygwin-xfree

[Tom Tromey]

Chris> Any thoughts on this, anyone?

My thought is that the very best approach is for us to be good
citizens and get this into the official X distribution.
Based on history that is a political nightmare, but we must definitely
try before we give up.

As to whether we should host it while we (a word which I sometimes use
to mean "someone else") do this, I don't know.  Hey, I just make user
accounts and occasionally complain.

Tom

Re: cygwin-xfree

[Chris Faylor]

On Wed, May 03, 2000 at 07:59:22PM -0700, Tom Tromey wrote:
>Chris> Any thoughts on this, anyone?
>
>My thought is that the very best approach is for us to be good
>citizens and get this into the official X distribution.
>Based on history that is a political nightmare, but we must definitely
>try before we give up.

Believe me they've tried.  Cygwin's GPL licensing restrictions are
apparently unacceptable to the xfree project.  That's a real shame.

>As to whether we should host it while we (a word which I sometimes use
>to mean "someone else") do this, I don't know.  Hey, I just make user
>accounts and occasionally complain.

Well, I just ask for user accounts and occasionally complain but I
suspect that this would have to be hosted for a long time.  I wouldn't
want to anger the XFree86 developers, though.

I hate politics.

cgf

Re: cygwin-xfree

[Tom Tromey]

>> Based on history that is a political nightmare, but we must definitely
>> try before we give up.

Chris> Believe me they've tried.

That's good enough for me.  If Jason says that sourceware can handle
the load, then I'm ok with hosting it.  Ideal would be if we could
point to somebody with final say.  When Jason was running sourceware
it was him.  Jim, are you the guy now?  I know it isn't me.

Tom

libstdc++-v3 things that still are not working correctly

[Benjamin Kosnik]

Please, somebody, help.


1) CVS web seems to be pointing at the old repository

2) html web pages are still not being updated, even with all the latest 
greatest helpful hints and everything. 

3) libstdc++-cvs mapped over to egcs/libstdc++-v3 checkins.

-benjamin

Re: cygwin-xfree

[Chris Faylor]

On Wed, May 03, 2000 at 08:09:38PM -0700, Tom Tromey wrote:
>>> Based on history that is a political nightmare, but we must definitely
>>> try before we give up.
>
>Chris> Believe me they've tried.
>
>That's good enough for me.  If Jason says that sourceware can handle
>the load, then I'm ok with hosting it.  Ideal would be if we could
>point to somebody with final say.  When Jason was running sourceware
>it was him.  Jim, are you the guy now?  I know it isn't me.

We used to have the kindly king Jason, who ruled over sourceware in the
Golden Age.

I sort of thought that, in the current Silver Age, there was a
triumverate of terror who ruled over all things sourceware.  These three
are trying to stamp out all memory of Jason with their sinister "red
hat" squad that is attempting to wipe out all memory of the beloved
monarch by renaming the sourceware site to something else.

I know that Jason claimed to have left office voluntarily but I've
always had my suspicions...

cgf

htdig, was Re: cygwin-xfree

[Frank Ch. Eigler]

crash wrote:

> [...]
> (On the other hand, the web drive and the htdig drive are filling
> at a rate faster than I had anticipated and won't last the ~18 month
> lifetime I had intended for this system :-(
> [...]

For the htdig databases, adding "compression_level: 9"
to htdig.conf should reduce the disk space usage by about 20%.
Also, the rundig scripts use htdig's "-i" option (causing the
databases to be regenerated from scratch), the "db.wordlist"
files may be deleted after each search, saving another 30%.

- FChE
-- 
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.0 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE5EO0hVZbdDOm/ZT0RAWchAJ90EhzYYEqMQzMx1dkHtIVSMKKhdwCeO3FP
MEJF2ABtyxvn7TY+1mz1tWM=
=gja3
-----END PGP SIGNATURE-----

Re: cygwin-xfree

[Bob Manson]

In message < 20000503232221.D13804@cygnus.com >, Chris Faylor writes:
>We used to have the kindly king Jason, who ruled over sourceware in the
>Golden Age.

He was no kindly king.  He was the Ruler of Evil!  Evil, I tell you!
I quaked in fear whenver he typed!  I used to tremble in my luxurious
penthouse condominium whenver he would threaten us with bodily harm.

Oh, Jason, wherefor art thou?
						Bob

Re: libstdc++-v3 things that still are not working correctly

[Jeffrey A Law]

  In message < Pine.SOL.3.91.1000503201215.12936B-100000@cse.cygnus.com >you writ
e:
  > 
  > Please, somebody, help.
  > 
  > 
  > 1) CVS web seems to be pointing at the old repository
That would be a problem with your download.html page.  I tried to fix it,
but the change apparently didn't show up on the web server.



  > 2) html web pages are still not being updated, even with all the latest 
  > greatest helpful hints and everything. 
Well, at least one problem was the loginfo file was wrong.  I still do
not know why it isn't working though.  A "+" in a regexp means one or more
copies of the previously matched pattern.  So "libstdc++-v3" didn't work
like people expected.  Changing it to libstdc..-v3 should help.

I ran into this when fixing #3 below.



  > 3) libstdc++-cvs mapped over to egcs/libstdc++-v3 checkins.
Fixed.
jeff

Re: cygwin-xfree

[Per Bothner]

Chris Faylor <cgf@cygnus.com> writes:

> Believe me they've tried.  Cygwin's GPL licensing restrictions are
> apparently unacceptable to the xfree project.  That's a real shame.

But that shouldn't really be an issue, should it, assuming I correctly
understand that the goal of this project is to create a *port* of X
that runs under Cygwin.  How is this different from creating a port of
X that runs under SunOS?  You just have to keep separate the
modifications to X (which remain under the X license) and the
modifications to Cygwin (which are under the GPL).  Presumably, he has
to write some new glue code - which should be considered part of (and
distributed with) either X or Cygwin, depending on what makes most
sense.

Now of course there is nothing that prevents us from releasing a
CD that contain *both* Cygwin and X - just like there is nothing
preventing Sun from releasing a CD containing both Solaris
including X.
-- 
	--Per Bothner
per@bothner.com   http://www.bothner.com/~per/

Re: cygwin-xfree

[Jim Kingdon]

> How is this different from creating a port of X that runs under SunOS?

At the risk of starting/continuing Yet Another Pointless Licensing
Flamewar....

The SunOS libc and other required libraries aren't under a viral
license (some proprietary compiler vendors, at least in the PC world,
tried this some years ago, but they didn't get away with it.  The most
amusing crash and burn being the TI-99/4 which wanted developers to
pay to develop products on that platform, sign NDA's, etc.  With the
predictable result - no applications and the eventual death of the
platform).

Now whether that has anything to do with the difficulty in getting the
cygwin XFree patches merged in, I don't know.  Over in Cristian's
group (don't remember who specifically builds the XFree RPMs for the
Red Hat Linux product but someone over there) they ship about 40
patches to XFree (last I looked), and licensing isn't an issue in that
context.

Re: cygwin-xfree

[Jim Kingdon]

> I sort of thought that, in the current Silver Age, there was a
> triumverate of terror who ruled over all things sourceware.

Yup, that's the way I see it.  C'mon, Tom, you can't inspire Terror
(that's with a capital "T") by saying "I only work here".  If you are
in doubt, just keep them guessing about what the decision was and who
made it.  That's always a good way to be an Evil Overlord.

More seriously, usually Jeff, Tom and I are able to agree on stuff
with some discussion.  For the most part things end of reducing to
"who will do the work" rather than "who decides stuff like technical
direction" since the latter is more the project leaders anyway.

As for cygwin-xfree, I'm still looking for objections to suggesting
they go to sourceforge.  One of the requirements for being on
kingdon-rules.redhat.com, er, sourceware.cygnus.com, is a sponsor
within Red Hat and in the XFree case we only seem to have one partial
reluctant one (Chris).  It's different than Cygwin itself as nearly as
I can tell.

As for angering XFree.  I could write essays on the subject (short
summary: don't let such concerns gridlock development) but the point
is, that's something that a project leader would need to be willing to
deal with.  I don't see it as a matter for the Sourceware Central.  At
least, not primarily.

> I know that Jason claimed to have left office voluntarily but I've
> always had my suspicions...

And suspicions they will remain.  We can't very well maintain our aura
of palace intrigue if we tell you what *really* happened, can we?

Re: cygwin-xfree

[Chris Faylor]

On Thu, May 04, 2000 at 07:52:02AM -0400, Jim Kingdon wrote:
>As for cygwin-xfree, I'm still looking for objections to suggesting
>they go to sourceforge.  One of the requirements for being on
>kingdon-rules.redhat.com, er, sourceware.cygnus.com, is a sponsor
>within Red Hat and in the XFree case we only seem to have one partial
>reluctant one (Chris).  It's different than Cygwin itself as nearly as
>I can tell.

I didn't mean to portray reluctance.  I *am* really interested in having
this on sourceware.  I just wanted to make sure that everyone realized
that I was not going to stomp my feet if the consensus was that
sourceware should not host this.  I probably went a little too much
in the other direction when I mentioned Source Forge.

cgf

Re: cygwin-xfree

[Jeffrey A Law]

  In message < 20000503221805.A13179@cygnus.com >you write:
  > 1) What does everyone think about adding another 150MB to the server.  Ther
  > e
  >    seems to be more than adequate disk space available for this but 150MB
  >    is still a pretty massive project.  I can just imagine what will happen
  >    to sourceware the first time someone checks this out.
I don't see a major problem with that.  We've got gobs of space for the
CVS stuff.  The only places we're even close to having to keep an eye on
are /ftp and /www which are running at 71% and 61% capacity respectively.

  > 2) I assume that if you all think that 1) is no problem then this directory
  >    should be winsup/xfree. 
Whatever seems reasonable to you since you're more familiar with this guy's
work that I am.

jeff

Re: cygwin-xfree

[Jason Molenda]

On Wed, May 03, 2000 at 09:12:14PM -0700, Bob Manson wrote:

> He was no kindly king.  He was the Ruler of Evil!  Evil, I tell you!
> I quaked in fear whenver he typed!  

It's true, it's all true.  Cygnus is actually a training ground
for Evil, and after half a dozen years, I had been well trained in
the Insufficient Light Side by my master, Stan Shebs, as were the
other disciples.  As I write this, we are all moving into the
Silicon Valley power structure and we will soon issue our demands
for <pinkie>one MILLION dollars</pinkie> or we'll bring SV to its
knees.

Evil is a lost art these days, and it behooves all of us to live
it every day.  I got a copy of that "ILOVEYOU" worm.  Its evilness
is so pathetic -- the author showed no pride in workmanship, no true
commitment to evil (Oooh, it overwrites your mp3 files, so destructive.)
Pathetic!



On a slightly serious note, it was the easiest path for me to take
to decide and implement (nearly) everything on the sourceware
system.  It was helpful for really rapid deployment.  I think I've
written about this in the past, but, in retrospect, it was a mistake
to not get more buy-in from Cygnus officially at some point.  Back
in the summer of '98 it would have been a huge mistake to get
marketing et al tied in to sourceware.  But the result of doing it
on the side, with managers looking the other way when sourceware
maintainers have to spend time on it, is not very effective.  We
should have (or present tense, should) convinced management to pony
up for a full time person and hooked it into the rest of the
company's organization.


Jason

Re: htdig, was Re: cygwin-xfree

[Hans-Peter Nilsson]

On Wed, 3 May 2000, Frank Ch. Eigler wrote:
> For the htdig databases, adding "compression_level: 9"
> to htdig.conf should reduce the disk space usage by about 20%.

Tests show that the db.docdb "before merging" went from 324231168 to
180889600 bytes, so I made that change for the gcc installation.  It
is running its normally scheduled update right now, but from scratch
today to save extra.  Doh, I just think I should have thought about that
long ago.

For the curious (but not curious enough to read up) this uses zlib to
individually compress each "document head"; the text that shows up for
each search hit.  FWIW, in 2.0 compression will be done at the DB level,
which would be the right thing.

If nothing broke (I'm cautious sometimes :-) we can move that conf item 
from gcc.conf to site.conf.

> Also, the rundig scripts use htdig's "-i" option (causing the
> databases to be regenerated from scratch),

No, since we don't use the basic rundig, as Jason told you.

> the "db.wordlist"
> files may be deleted after each search, saving another 30%.

No, it's needed when you do an "update" (not running from scratch).

Thanks!

brgds, H-P

Re: cygwin-xfree

[Jim Kingdon]

> in retrospect, it was a mistake to not get more buy-in from Cygnus
> officially at some point.  Back in the summer of '98 it would have
> been a huge mistake to get marketing et al tied in to sourceware.

Well, I'm sort of trying to involve the whole company more, e.g. with
sending the rename question to the company wide list.  And I have
talked with a few people in marketing (not decision makers in
marketing, but actually "decision maker in marketing" can be a bit of
an oxymoron just now - most notably because Tom Butta (chief marketing
officer) was replaced only a few weeks ago).

> We should have (or present tense, should) convinced management to pony
> up for a full time person and hooked it into the rest of the company's
> organization.

Part of the problem in the past was figuring out where in the
organization this kind of thing might hook in.

Although I guess the problem now is more that loss leaders are just
really hard to get funded (or maybe not, that's what I thought and
then Jim Blandy got funded for 6 months to go hack on version control,
which just goes to show that I don't really understand - or
something).

If anyone in Red Hat is interested in working on this problem, take a
look at my internal page on sourceware strategy (email me for the URL
on the intranet if you need it), and let me know your thoughts.

Re: cygwin-xfree

[Stan Shebs]

Jason Molenda wrote:
> 
> On Wed, May 03, 2000 at 09:12:14PM -0700, Bob Manson wrote:
> 
> > He was no kindly king.  He was the Ruler of Evil!  Evil, I tell you!
> > I quaked in fear whenver he typed!
> 
> It's true, it's all true.  Cygnus is actually a training ground
> for Evil,

Not true!  Don't you remember when we got the fan mail sent into
info@cygnus, telling us how great GNU was and how we Cygnites were
doing God's work?  Alas, it wasn't one of the paying customers...

> We
> should have (or present tense, should) convinced management to pony
> up for a full time person and hooked it into the rest of the
> company's organization.

For those who need to work on this, allow me to recommend the power
of the nag.  These days, managers are so interrupt-driven that the
one-time mail message gets lost in the noise.  Nagging serves the
dual purposes of regular reminder and effective increase in priority;
keeping the issue "paged in", as it were.  Once a week is good, daily
is usually too much, monthly is too sparse.  Be sure the nag includes
specific activities with dollars and dates, otherwise it degenerates
into whining, which is useless.

Stan

Re: cygwin-xfree

[Mark Galassi]

    Stan> Not true!  Don't you remember when we got the fan mail sent
    Stan> into info@cygnus, telling us how great GNU was and how we
    Stan> Cygnites were doing God's work?  Alas, it wasn't one of the
    Stan> paying customers...

Yes, but do you also remember the time we got a letter from a guy who
was really upset with us because of Cygwin32?  He said that it was
possible to run POSIX applications on NT with that, so his employer
insisted on using NT.  If Cygwin32 had not been around he would have
been able to use UNIX at work.

Re: cygwin-xfree

[Jeffrey A Law]

  In message < 20000503221805.A13179@cygnus.com >you write:
  > 1) What does everyone think about adding another 150MB to the server.  Ther
  > e
  >    seems to be more than adequate disk space available for this but 150MB
  >    is still a pretty massive project.  I can just imagine what will happen
  >    to sourceware the first time someone checks this out.
I don't see a major problem with that.  We've got gobs of space for the
CVS stuff.  The only places we're even close to having to keep an eye on
are /ftp and /www which are running at 71% and 61% capacity respectively.

  > 2) I assume that if you all think that 1) is no problem then this directory
  >    should be winsup/xfree. 
Whatever seems reasonable to you since you're more familiar with this guy's
work that I am.

jeff

Re: cygwin-xfree

[Tom Tromey]

>> Based on history that is a political nightmare, but we must definitely
>> try before we give up.

Chris> Believe me they've tried.

That's good enough for me.  If Jason says that sourceware can handle
the load, then I'm ok with hosting it.  Ideal would be if we could
point to somebody with final say.  When Jason was running sourceware
it was him.  Jim, are you the guy now?  I know it isn't me.

Tom

Re: cygwin-xfree

[Jim Kingdon]

> in retrospect, it was a mistake to not get more buy-in from Cygnus
> officially at some point.  Back in the summer of '98 it would have
> been a huge mistake to get marketing et al tied in to sourceware.

Well, I'm sort of trying to involve the whole company more, e.g. with
sending the rename question to the company wide list.  And I have
talked with a few people in marketing (not decision makers in
marketing, but actually "decision maker in marketing" can be a bit of
an oxymoron just now - most notably because Tom Butta (chief marketing
officer) was replaced only a few weeks ago).

> We should have (or present tense, should) convinced management to pony
> up for a full time person and hooked it into the rest of the company's
> organization.

Part of the problem in the past was figuring out where in the
organization this kind of thing might hook in.

Although I guess the problem now is more that loss leaders are just
really hard to get funded (or maybe not, that's what I thought and
then Jim Blandy got funded for 6 months to go hack on version control,
which just goes to show that I don't really understand - or
something).

If anyone in Red Hat is interested in working on this problem, take a
look at my internal page on sourceware strategy (email me for the URL
on the intranet if you need it), and let me know your thoughts.

Re: cygwin-xfree

[Mark Galassi]

    Stan> Not true!  Don't you remember when we got the fan mail sent
    Stan> into info@cygnus, telling us how great GNU was and how we
    Stan> Cygnites were doing God's work?  Alas, it wasn't one of the
    Stan> paying customers...

Yes, but do you also remember the time we got a letter from a guy who
was really upset with us because of Cygwin32?  He said that it was
possible to run POSIX applications on NT with that, so his employer
insisted on using NT.  If Cygwin32 had not been around he would have
been able to use UNIX at work.

Re: cygwin-xfree

[Chris Faylor]

On Wed, May 03, 2000 at 07:44:27PM -0700, Jason Molenda wrote:
>> 2) I assume that if you all think that 1) is no problem then this directory
>>    should be winsup/xfree.  
>
>It almost seems like something you'd want to put in a separate
>repository.  I mean, it's _really_ Xfree86 with a few patches on
>top (OK, a lot of patches).  How would you handle e.g. make or perl
>(I'm trying to guess at programs that might have cygwin patches to
>them) if someone wanted to do the same?

Dunno.  I've been thinking that the winsup directory should be its
own little mini-build tree.  I didn't want to pollute the rest of
sourceware with stuff that was only intended for cygwin.

>> The winsup directory is not being mirrored by any of the CVS mirrors, right?
>
>If anyone is rsync'ing the entire src repository, they'll pick it
>up, but it's not the end of the world for them to end up with
>another ~150MB.  I would think about it more like "Where do I want
>third party programs that we maintain patches to be checked in?"
>
>Maybe a new repository, "cygwin-mods" or something, would be the cleanest
>approach.  But cygwin-mods is a terrible name.

Maybe cygwin-apps?

>If he ever does get ready to start importing Xfree86, please make sure
>he reads the CVS manual on tracking third party sources,
>	http://www.loria.fr/~molli/cvs/doc/cvs_13.html#SEC98
>
>or provide exact command lines for him.  It'd probably make sense
>to import the most recent release of xfree86, then the development
>version on which he's based his changes, and then check in his
>changes on top of that.

Yup.  I supplied the exact command lines for him.  And emphasized
repeatedly that he follow them exactly.  And he screwed them up...

>That way if he dies in a tragic pingpong ball manufacturing plant accident,
>people will be able to disentangle what he was up to.  It's really
>easy to import sources incorrectly, and annoying to recover from that
>at a later date.

I've repeatedly stressed this to him.  He spent all last night checking
everything in but somehow it never showed up on sourceware, or, at least
I can't find it.  He says that he understands why this happened.

>>I'm have mixed feelings about this.  If we can't provide this, the
>>developer will go to Source Forge.
>
>I don't have a strong opinion either, I did want to interject some
>facts about the system and stuff, tho.

Thanks, Jason.

cgf

Re: cygwin-xfree

[Jason Molenda]

On Wed, May 03, 2000 at 10:18:05PM -0400, Chris Faylor wrote:

> 1) What does everyone think about adding another 150MB to the server.  There
>    seems to be more than adequate disk space available for this but 150MB
>    is still a pretty massive project.  I can just imagine what will happen
>    to sourceware the first time someone checks this out.


IMHO 150MB isn't very much - I way overspec'ed the CVS drive so
there is a lot of slack there.

(On the other hand, the web drive and the htdig drive are filling
at a rate faster than I had anticipated and won't last the ~18 month
lifetime I had intended for this system :-(


> 2) I assume that if you all think that 1) is no problem then this directory
>    should be winsup/xfree.  

It almost seems like something you'd want to put in a separate
repository.  I mean, it's _really_ Xfree86 with a few patches on
top (OK, a lot of patches).  How would you handle e.g. make or perl
(I'm trying to guess at programs that might have cygwin patches to
them) if someone wanted to do the same?


> The winsup directory is not being mirrored by any of the CVS mirrors, right?

If anyone is rsync'ing the entire src repository, they'll pick it
up, but it's not the end of the world for them to end up with
another ~150MB.  I would think about it more like "Where do I want
third party programs that we maintain patches to be checked in?"

Maybe a new repository, "cygwin-mods" or something, would be the cleanest
approach.  But cygwin-mods is a terrible name.


If he ever does get ready to start importing Xfree86, please make sure
he reads the CVS manual on tracking third party sources,
	http://www.loria.fr/~molli/cvs/doc/cvs_13.html#SEC98

or provide exact command lines for him.  It'd probably make sense
to import the most recent release of xfree86, then the development
version on which he's based his changes, and then check in his
changes on top of that.

That way if he dies in a tragic pingpong ball manufacturing plant accident,
people will be able to disentangle what he was up to.  It's really
easy to import sources incorrectly, and annoying to recover from that
at a later date.



> I'm have mixed feelings about this.  If we can't provide this, the developer
> will go to Source Forge.  

I don't have a strong opinion either, I did want to interject some
facts about the system and stuff, tho.



J

Re: cygwin-xfree

[Chris Faylor]

On Thu, May 04, 2000 at 12:29:03PM +1000, Andrew Cagney wrote:
>Where is the public xfree86 CVS repository? Hmm:
>
> http://www.xfree86.org/developer.html
>> To become an XFree86 developer you first need to join The XFree86
>> Project as a non-voting member. Access to the XFree86 development
>> code is only available to project members, and being a member provides
>> the necessary legal status to allow developers to access information
>> available to the project from other sources. 
>
>there isn't one.

AFAIK, this isn't development code.  It should be released, publicly
available code.

I'm not 100% certain of that, though.

cgf

Re: cygwin-xfree

[Andrew Cagney]

Chris Faylor wrote:
> 
> I have recently offered CVS space to the the guy who has been working on
> the port of XFree86 to Cygwin.
> 
> I thought he wanted it to upload the source code for a few drivers.
> 
> It turns out that he wants to upload pretty much all of the XFree86
> source code since there were, apparently, a number of changes required
> for Cygwin.
> 
> So, two question:
> 
> 1) What does everyone think about adding another 150MB to the server.  There
>    seems to be more than adequate disk space available for this but 150MB
>    is still a pretty massive project.  I can just imagine what will happen
>    to sourceware the first time someone checks this out.
> 
> 2) I assume that if you all think that 1) is no problem then this directory
>    should be winsup/xfree.  The winsup directory is not being mirrored by
>    any of the CVS mirrors, right?
> 
> I'm have mixed feelings about this.  If we can't provide this, the developer
> will go to Source Forge.  Maybe that's the best thing.  He's already split
> the development that had started for cygwin so that there is just a generic
> msvc/mingw version of everything as well as a cygwin-specific branch.  It
> would probably be not too much work to set up on Source Forge and maybe
> it would even give Cygwin a little publicity.

Where is the public xfree86 CVS repository? Hmm:

http://www.xfree86.org/developer.html
> To become an XFree86 developer you first need to join The XFree86
> Project as a non-voting member. Access to the XFree86 development
> code is only available to project members, and being a member provides
> the necessary legal status to allow developers to access information
> available to the project from other sources. 

there isn't one.

	Andrew

Re: cygwin-xfree

[Jim Kingdon]

> I sort of thought that, in the current Silver Age, there was a
> triumverate of terror who ruled over all things sourceware.

Yup, that's the way I see it.  C'mon, Tom, you can't inspire Terror
(that's with a capital "T") by saying "I only work here".  If you are
in doubt, just keep them guessing about what the decision was and who
made it.  That's always a good way to be an Evil Overlord.

More seriously, usually Jeff, Tom and I are able to agree on stuff
with some discussion.  For the most part things end of reducing to
"who will do the work" rather than "who decides stuff like technical
direction" since the latter is more the project leaders anyway.

As for cygwin-xfree, I'm still looking for objections to suggesting
they go to sourceforge.  One of the requirements for being on
kingdon-rules.redhat.com, er, sourceware.cygnus.com, is a sponsor
within Red Hat and in the XFree case we only seem to have one partial
reluctant one (Chris).  It's different than Cygwin itself as nearly as
I can tell.

As for angering XFree.  I could write essays on the subject (short
summary: don't let such concerns gridlock development) but the point
is, that's something that a project leader would need to be willing to
deal with.  I don't see it as a matter for the Sourceware Central.  At
least, not primarily.

> I know that Jason claimed to have left office voluntarily but I've
> always had my suspicions...

And suspicions they will remain.  We can't very well maintain our aura
of palace intrigue if we tell you what *really* happened, can we?

Re: cygwin-xfree

[Tom Tromey]

Chris> Any thoughts on this, anyone?

My thought is that the very best approach is for us to be good
citizens and get this into the official X distribution.
Based on history that is a political nightmare, but we must definitely
try before we give up.

As to whether we should host it while we (a word which I sometimes use
to mean "someone else") do this, I don't know.  Hey, I just make user
accounts and occasionally complain.

Tom

Re: cygwin-xfree

[Per Bothner]

Chris Faylor <cgf@cygnus.com> writes:

> Believe me they've tried.  Cygwin's GPL licensing restrictions are
> apparently unacceptable to the xfree project.  That's a real shame.

But that shouldn't really be an issue, should it, assuming I correctly
understand that the goal of this project is to create a *port* of X
that runs under Cygwin.  How is this different from creating a port of
X that runs under SunOS?  You just have to keep separate the
modifications to X (which remain under the X license) and the
modifications to Cygwin (which are under the GPL).  Presumably, he has
to write some new glue code - which should be considered part of (and
distributed with) either X or Cygwin, depending on what makes most
sense.

Now of course there is nothing that prevents us from releasing a
CD that contain *both* Cygwin and X - just like there is nothing
preventing Sun from releasing a CD containing both Solaris
including X.
-- 
	--Per Bothner
per@bothner.com   http://www.bothner.com/~per/

htdig, was Re: cygwin-xfree

[Frank Ch. Eigler]

crash wrote:

> [...]
> (On the other hand, the web drive and the htdig drive are filling
> at a rate faster than I had anticipated and won't last the ~18 month
> lifetime I had intended for this system :-(
> [...]

For the htdig databases, adding "compression_level: 9"
to htdig.conf should reduce the disk space usage by about 20%.
Also, the rundig scripts use htdig's "-i" option (causing the
databases to be regenerated from scratch), the "db.wordlist"
files may be deleted after each search, saving another 30%.

- FChE
-- 
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.0 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE5EO0hVZbdDOm/ZT0RAWchAJ90EhzYYEqMQzMx1dkHtIVSMKKhdwCeO3FP
MEJF2ABtyxvn7TY+1mz1tWM=
=gja3
-----END PGP SIGNATURE-----

Re: htdig, was Re: cygwin-xfree

[Hans-Peter Nilsson]

On Wed, 3 May 2000, Frank Ch. Eigler wrote:
> For the htdig databases, adding "compression_level: 9"
> to htdig.conf should reduce the disk space usage by about 20%.

Tests show that the db.docdb "before merging" went from 324231168 to
180889600 bytes, so I made that change for the gcc installation.  It
is running its normally scheduled update right now, but from scratch
today to save extra.  Doh, I just think I should have thought about that
long ago.

For the curious (but not curious enough to read up) this uses zlib to
individually compress each "document head"; the text that shows up for
each search hit.  FWIW, in 2.0 compression will be done at the DB level,
which would be the right thing.

If nothing broke (I'm cautious sometimes :-) we can move that conf item 
from gcc.conf to site.conf.

> Also, the rundig scripts use htdig's "-i" option (causing the
> databases to be regenerated from scratch),

No, since we don't use the basic rundig, as Jason told you.

> the "db.wordlist"
> files may be deleted after each search, saving another 30%.

No, it's needed when you do an "update" (not running from scratch).

Thanks!

brgds, H-P

Re: cygwin-xfree

[Jason Molenda]

On Wed, May 03, 2000 at 09:12:14PM -0700, Bob Manson wrote:

> He was no kindly king.  He was the Ruler of Evil!  Evil, I tell you!
> I quaked in fear whenver he typed!  

It's true, it's all true.  Cygnus is actually a training ground
for Evil, and after half a dozen years, I had been well trained in
the Insufficient Light Side by my master, Stan Shebs, as were the
other disciples.  As I write this, we are all moving into the
Silicon Valley power structure and we will soon issue our demands
for <pinkie>one MILLION dollars</pinkie> or we'll bring SV to its
knees.

Evil is a lost art these days, and it behooves all of us to live
it every day.  I got a copy of that "ILOVEYOU" worm.  Its evilness
is so pathetic -- the author showed no pride in workmanship, no true
commitment to evil (Oooh, it overwrites your mp3 files, so destructive.)
Pathetic!



On a slightly serious note, it was the easiest path for me to take
to decide and implement (nearly) everything on the sourceware
system.  It was helpful for really rapid deployment.  I think I've
written about this in the past, but, in retrospect, it was a mistake
to not get more buy-in from Cygnus officially at some point.  Back
in the summer of '98 it would have been a huge mistake to get
marketing et al tied in to sourceware.  But the result of doing it
on the side, with managers looking the other way when sourceware
maintainers have to spend time on it, is not very effective.  We
should have (or present tense, should) convinced management to pony
up for a full time person and hooked it into the rest of the
company's organization.


Jason

Re: cygwin-xfree

[Chris Faylor]

On Thu, May 04, 2000 at 07:52:02AM -0400, Jim Kingdon wrote:
>As for cygwin-xfree, I'm still looking for objections to suggesting
>they go to sourceforge.  One of the requirements for being on
>kingdon-rules.redhat.com, er, sourceware.cygnus.com, is a sponsor
>within Red Hat and in the XFree case we only seem to have one partial
>reluctant one (Chris).  It's different than Cygwin itself as nearly as
>I can tell.

I didn't mean to portray reluctance.  I *am* really interested in having
this on sourceware.  I just wanted to make sure that everyone realized
that I was not going to stomp my feet if the consensus was that
sourceware should not host this.  I probably went a little too much
in the other direction when I mentioned Source Forge.

cgf

libstdc++-v3 things that still are not working correctly

[Benjamin Kosnik]

Please, somebody, help.


1) CVS web seems to be pointing at the old repository

2) html web pages are still not being updated, even with all the latest 
greatest helpful hints and everything. 

3) libstdc++-cvs mapped over to egcs/libstdc++-v3 checkins.

-benjamin

cygwin-xfree

[Chris Faylor]

I have recently offered CVS space to the the guy who has been working on
the port of XFree86 to Cygwin.

I thought he wanted it to upload the source code for a few drivers.

It turns out that he wants to upload pretty much all of the XFree86
source code since there were, apparently, a number of changes required
for Cygwin.

So, two question:

1) What does everyone think about adding another 150MB to the server.  There
   seems to be more than adequate disk space available for this but 150MB
   is still a pretty massive project.  I can just imagine what will happen
   to sourceware the first time someone checks this out.

2) I assume that if you all think that 1) is no problem then this directory
   should be winsup/xfree.  The winsup directory is not being mirrored by
   any of the CVS mirrors, right?

I'm have mixed feelings about this.  If we can't provide this, the developer
will go to Source Forge.  Maybe that's the best thing.  He's already split
the development that had started for cygwin so that there is just a generic
msvc/mingw version of everything as well as a cygwin-specific branch.  It
would probably be not too much work to set up on Source Forge and maybe
it would even give Cygwin a little publicity.

Any thoughts on this, anyone?

cgf

Re: cygwin-xfree

[Stan Shebs]

Jason Molenda wrote:
> 
> On Wed, May 03, 2000 at 09:12:14PM -0700, Bob Manson wrote:
> 
> > He was no kindly king.  He was the Ruler of Evil!  Evil, I tell you!
> > I quaked in fear whenver he typed!
> 
> It's true, it's all true.  Cygnus is actually a training ground
> for Evil,

Not true!  Don't you remember when we got the fan mail sent into
info@cygnus, telling us how great GNU was and how we Cygnites were
doing God's work?  Alas, it wasn't one of the paying customers...

> We
> should have (or present tense, should) convinced management to pony
> up for a full time person and hooked it into the rest of the
> company's organization.

For those who need to work on this, allow me to recommend the power
of the nag.  These days, managers are so interrupt-driven that the
one-time mail message gets lost in the noise.  Nagging serves the
dual purposes of regular reminder and effective increase in priority;
keeping the issue "paged in", as it were.  Once a week is good, daily
is usually too much, monthly is too sparse.  Be sure the nag includes
specific activities with dollars and dates, otherwise it degenerates
into whining, which is useless.

Stan

Re: libstdc++-v3 things that still are not working correctly

[Jeffrey A Law]

  In message < Pine.SOL.3.91.1000503201215.12936B-100000@cse.cygnus.com >you writ
e:
  > 
  > Please, somebody, help.
  > 
  > 
  > 1) CVS web seems to be pointing at the old repository
That would be a problem with your download.html page.  I tried to fix it,
but the change apparently didn't show up on the web server.



  > 2) html web pages are still not being updated, even with all the latest 
  > greatest helpful hints and everything. 
Well, at least one problem was the loginfo file was wrong.  I still do
not know why it isn't working though.  A "+" in a regexp means one or more
copies of the previously matched pattern.  So "libstdc++-v3" didn't work
like people expected.  Changing it to libstdc..-v3 should help.

I ran into this when fixing #3 below.



  > 3) libstdc++-cvs mapped over to egcs/libstdc++-v3 checkins.
Fixed.
jeff

Re: cygwin-xfree

[Jim Kingdon]

> Where is the public xfree86 CVS repository? Hmm: . . .
> there isn't one.

Well, yes, and we (there is no we) would like them to open up.
Progress so far has been pretty slow although there seems to be
general claims that they are moving in the openness direction.  One
bright spot is that the Direct Rendering Infrastructure is at
http://dri.sourceforge.net/ - the Precision Insight folks are doing a
lot of X work, some of it with Red Hat funding.

As for what to tell the cygwin-xfree people, sounds good to me if they
take it to sourceforge.  Unless people see some big problem with that
(as it says on the main page, we are the Hosting Site To the World at
this time).

Re: cygwin-xfree

[Chris Faylor]

On Wed, May 03, 2000 at 08:09:38PM -0700, Tom Tromey wrote:
>>> Based on history that is a political nightmare, but we must definitely
>>> try before we give up.
>
>Chris> Believe me they've tried.
>
>That's good enough for me.  If Jason says that sourceware can handle
>the load, then I'm ok with hosting it.  Ideal would be if we could
>point to somebody with final say.  When Jason was running sourceware
>it was him.  Jim, are you the guy now?  I know it isn't me.

We used to have the kindly king Jason, who ruled over sourceware in the
Golden Age.

I sort of thought that, in the current Silver Age, there was a
triumverate of terror who ruled over all things sourceware.  These three
are trying to stamp out all memory of Jason with their sinister "red
hat" squad that is attempting to wipe out all memory of the beloved
monarch by renaming the sourceware site to something else.

I know that Jason claimed to have left office voluntarily but I've
always had my suspicions...

cgf

Re: cygwin-xfree

[Chris Faylor]

On Wed, May 03, 2000 at 07:59:22PM -0700, Tom Tromey wrote:
>Chris> Any thoughts on this, anyone?
>
>My thought is that the very best approach is for us to be good
>citizens and get this into the official X distribution.
>Based on history that is a political nightmare, but we must definitely
>try before we give up.

Believe me they've tried.  Cygwin's GPL licensing restrictions are
apparently unacceptable to the xfree project.  That's a real shame.

>As to whether we should host it while we (a word which I sometimes use
>to mean "someone else") do this, I don't know.  Hey, I just make user
>accounts and occasionally complain.

Well, I just ask for user accounts and occasionally complain but I
suspect that this would have to be hosted for a long time.  I wouldn't
want to anger the XFree86 developers, though.

I hate politics.

cgf

Re: cygwin-xfree

[Jim Kingdon]

> How is this different from creating a port of X that runs under SunOS?

At the risk of starting/continuing Yet Another Pointless Licensing
Flamewar....

The SunOS libc and other required libraries aren't under a viral
license (some proprietary compiler vendors, at least in the PC world,
tried this some years ago, but they didn't get away with it.  The most
amusing crash and burn being the TI-99/4 which wanted developers to
pay to develop products on that platform, sign NDA's, etc.  With the
predictable result - no applications and the eventual death of the
platform).

Now whether that has anything to do with the difficulty in getting the
cygwin XFree patches merged in, I don't know.  Over in Cristian's
group (don't remember who specifically builds the XFree RPMs for the
Red Hat Linux product but someone over there) they ship about 40
patches to XFree (last I looked), and licensing isn't an issue in that
context.

Re: cygwin-xfree

[Bob Manson]

In message < 20000503232221.D13804@cygnus.com >, Chris Faylor writes:
>We used to have the kindly king Jason, who ruled over sourceware in the
>Golden Age.

He was no kindly king.  He was the Ruler of Evil!  Evil, I tell you!
I quaked in fear whenver he typed!  I used to tremble in my luxurious
penthouse condominium whenver he would threaten us with bodily harm.

Oh, Jason, wherefor art thou?
						Bob