Problem with the fix to PR 10234

Problem with the fix to PR 10234

[Joe Buck]

Sorry to have to bring this up, but I was updating my bug list for
3.2.3 and just noticed this.

The fix to PR 10234 was erroneous.  Sorry I missed it, but it should
have been closed as "not a bug".  The fact that the URL printed by
released compilers for bug report instructions points to

http://www.gnu.org/software/gcc/bugs.html

and not to gcc.gnu.org was something that RMS asked for and that the
SC agreed to.  This goes back to the days of the egcs/FSF re-merger,
and was just one of many petty battles from those days.  It shouldn't
really be a big deal, but I don't think that the SC can break a promise
to RMS.  So either this "fix" has to be reverted or we need to ask RMS
if it's OK.  Since the named URL is just a mirror of the gcc.gnu.org
bugs page, it's simpler to just revert the fix.

Apologies to Wolfgang, Janis, and Gaby, none of who knew about this
foolishness.

Re: Problem with the fix to PR 10234

[Janis Johnson]

On Wed, Apr 02, 2003 at 02:51:52PM -0800, Joe Buck wrote:
> 
> Sorry to have to bring this up, but I was updating my bug list for
> 3.2.3 and just noticed this.
> 
> The fix to PR 10234 was erroneous.  Sorry I missed it, but it should
> have been closed as "not a bug".  The fact that the URL printed by
> released compilers for bug report instructions points to
> 
> http://www.gnu.org/software/gcc/bugs.html
> 
> and not to gcc.gnu.org was something that RMS asked for and that the
> SC agreed to.  This goes back to the days of the egcs/FSF re-merger,
> and was just one of many petty battles from those days.  It shouldn't
> really be a big deal, but I don't think that the SC can break a promise
> to RMS.  So either this "fix" has to be reverted or we need to ask RMS
> if it's OK.  Since the named URL is just a mirror of the gcc.gnu.org
> bugs page, it's simpler to just revert the fix.
> 
> Apologies to Wolfgang, Janis, and Gaby, none of who knew about this
> foolishness.

There was a different fix to 3.3 and mainline for this, and I also
updated the web page itself to give its permanent location as the GCC
site.  At the time these changes were made, attempts to get to
http://www.gnu.org/software/gcc/bugs.html just showed a blank page.
There are often problems with the copy (not a real mirror) of GCC
pages from http://www.gnu.org.

Shall I revert these?

Janis

Re: Problem with the fix to PR 10234

[Joe Buck]

 wrote:
> > The fix to PR 10234 was erroneous.  Sorry I missed it, but it should
> > have been closed as "not a bug".  The fact that the URL printed by
> > released compilers for bug report instructions points to
> > 
> > http://www.gnu.org/software/gcc/bugs.html
> > 
> > and not to gcc.gnu.org was something that RMS asked for and that the
> > SC agreed to.
 
On Wed, Apr 02, 2003 at 03:40:50PM -0800, Janis Johnson wrote:
> There was a different fix to 3.3 and mainline for this, and I also
> updated the web page itself to give its permanent location as the GCC
> site.  At the time these changes were made, attempts to get to
> http://www.gnu.org/software/gcc/bugs.html just showed a blank page.
> There are often problems with the copy (not a real mirror) of GCC
> pages from http://www.gnu.org.
> 
> Shall I revert these?

Please hold off just a bit.  Since, as you point out, the mirroring is
frequently broken, it may be best to just ask RMS to let us point
direct to gcc.gnu.org.  After several years of the current GCC arrangement
he may feel more comfortable with it.

Gaby, this is something that we'll have to settle before any 3.2.3
release, but I'll try to get it solved quickly.
 

Re: Problem with the fix to PR 10234

[Gabriel Dos Reis]

Joe Buck <jbuck@synopsys.com> writes:

[...]

| Gaby, this is something that we'll have to settle before any 3.2.3
| release, but I'll try to get it solved quickly.

Joe,

  You certainly have more information about this issue than I do.  So,
I'm OK with anything you think is best for it.

-- Gaby

Re: Problem with the fix to PR 10234

[Joe Buck]

On Sat, Apr 12, 2003 at 03:43:48PM +0200, Gabriel Dos Reis wrote:
> Joe Buck <jbuck@synopsys.com> writes:
> 
> [...]
> 
> | Gaby, this is something that we'll have to settle before any 3.2.3
> | release, but I'll try to get it solved quickly.
> 
> Joe,
> 
>   You certainly have more information about this issue than I do.  So,
> I'm OK with anything you think is best for it.

This one is settled; RMS gave us permission to use gcc.gnu.org in links.
You don't need to do anything.

Re: Problem with the fix to PR 10234

[Gerald Pfeifer]

On Wed, 2 Apr 2003, Janis Johnson wrote:
> Shall I revert these?

Joe already contacted RMS, so I suggest to wait for his response.

(Given what I wrote in
  http://gcc.gnu.org/ml/gcc-patches/2003-03/msg02243.html
I do hope we don't have to revert your changes.)

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

Re: Problem with the fix to PR 10234

[Joe Buck]

Once again:

I wrote:
> > The fix to PR 10234 was erroneous.  Sorry I missed it, but it should
> > have been closed as "not a bug".  The fact that the URL printed by
> > released compilers for bug report instructions points to

> > http://www.gnu.org/software/gcc/bugs.html
> > 
> > and not to gcc.gnu.org was something that RMS asked for and that the
> > SC agreed to.

On Wed, Apr 02, 2003 at 03:40:50PM -0800, Janis Johnson wrote:
> Shall I revert these?

We heard back from RMS.

RMS still wants all GNUware to point to http://www.gnu.org/software/xxx
(though there's a proposal to use xxx.gnu.org for all major tools and
this may happen in the future).  So the decision is that we need to
revert, at least for now.  He wants all the tools to follow a consistent
policy, so we might be able to change it to the way many would prefer
at some future point (at which time there would be a gdb.gnu.org,
binutils.gnu.org, etc, though these might just be virtual hosts).

Please adjust all references in the documentation and GCC itself
accordingly.  This has to be done before any new release.  Sorry.

Joe


-- 
Q. What's more of a headache than a bug in a compiler.
A. Bugs in six compilers.  -- Mark Johnson

Re: Problem with the fix to PR 10234

[Janis Johnson]

On Thu, Apr 03, 2003 at 05:27:56PM -0800, Joe Buck wrote:
> Once again:
> 
> I wrote:
> > > The fix to PR 10234 was erroneous.  Sorry I missed it, but it should
> > > have been closed as "not a bug".  The fact that the URL printed by
> > > released compilers for bug report instructions points to
> 
> > > http://www.gnu.org/software/gcc/bugs.html
> > > 
> > > and not to gcc.gnu.org was something that RMS asked for and that the
> > > SC agreed to.
> 
> On Wed, Apr 02, 2003 at 03:40:50PM -0800, Janis Johnson wrote:
> > Shall I revert these?
> 
> We heard back from RMS.
> 
> RMS still wants all GNUware to point to http://www.gnu.org/software/xxx
> (though there's a proposal to use xxx.gnu.org for all major tools and
> this may happen in the future).  So the decision is that we need to
> revert, at least for now.  He wants all the tools to follow a consistent
> policy, so we might be able to change it to the way many would prefer
> at some future point (at which time there would be a gdb.gnu.org,
> binutils.gnu.org, etc, though these might just be virtual hosts).
> 
> Please adjust all references in the documentation and GCC itself
> accordingly.  This has to be done before any new release.  Sorry.

Joe,

I understand the desire for consistency among the components of the
GNU Project.  It's vitally important, though, for GCC to provide URLs
that can be accessed reliably.  I'd like to push back on this before
reverting the patches.

The changes to the ICE messages in the active branches were made after
http://www.gnu.org/software/gcc/ was inaccessible for a few days.  Some 
people who saw ICE messages and couldn't access the URL cared enough to
track down the problem and report the bad link; others probably gave up
and didn't find the page at the GCC site and we missed out on receiving
problem reports.

I don't know of other cases when the GCC pages have been inaccessible
from the GNU site, but quite frequently we get complaints about the web
pages at http://www.gnu.org/software/gcc being out of date.  Usually
they are updated every few days, but sometimes they lag much further
behind the pages at gcc.gnu.org, where updates are visible immediately.
The page referenced in ICE messages, bugs.html, was updated April 2 but
the update is not yet available at the main GNU site.

Before changing back the URL in ICE messages I'd like to know what the
problems were that caused the old URL to get only blank pages for a few
days, and to have some assurances that the URL can be accessed reliably
in the future.

Janis Johnson
GCC Web Pages Maintainer

Re: Problem with the fix to PR 10234

[Joe Buck]

On Fri, Apr 04, 2003 at 11:17:49AM -0800, Janis Johnson wrote:

> I understand the desire for consistency among the components of the
> GNU Project.  It's vitally important, though, for GCC to provide URLs
> that can be accessed reliably.  I'd like to push back on this before
> reverting the patches.

You can hold off a little more, Gerald is still trying to convince
RMS.  But we may have to do it.  The deadline for deciding is the 3.2.3
release; by that release, either we need to put the old URL back or
we need to get permission to change it.

> Before changing back the URL in ICE messages I'd like to know what the
> problems were that caused the old URL to get only blank pages for a few
> days, and to have some assurances that the URL can be accessed reliably
> in the future.

We'll try to get a commitment from RMS that he lean on the webmasters
to be more responsive about fixing any problems.  He was puzzled (as
was I) as to why a broken mirror wouldn't just have a stale copy, but
the key issue with me is that the FSF webmasters aren't diligent about
quickly responding to problems.

Re: Problem with the fix to PR 10234

[Gerald Pfeifer]

On Fri, 4 Apr 2003, Janis Johnson wrote:
> I understand the desire for consistency among the components of the
> GNU Project.  It's vitally important, though, for GCC to provide URLs
> that can be accessed reliably.  I'd like to push back on this before
> reverting the patches.

To which Joe Buck replied:
> You can hold off a little more, Gerald is still trying to convince
> RMS.  But we may have to do it.  The deadline for deciding is the 3.2.3
> release; by that release, either we need to put the old URL back or
> we need to get permission to change it.

Janis,

I just received mail from RMS where he gave us permission to "keep it
as gcc.gnu.org", which means we don't need to revert these patches.

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

Re: Problem with the fix to PR 10234

[Gabriel Dos Reis]

Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at> writes:

| I just received mail from RMS where he gave us permission to "keep it
| as gcc.gnu.org", which means we don't need to revert these patches.

This is good news.  Thanks to all of you!

-- Gaby