Using GNATS to track GCC patches

Using GNATS to track GCC patches

[Craig Rodrigues]

Hi,

I recall from the following thread that the idea
of using GNATS to track patches in GCC came up:

http://gcc.gnu.org/ml/gcc/2000-06/msg00264.html
http://gcc.gnu.org/ml/gcc/2000-06/msg00263.html

Has anyone made any move to implement this?
I think it is a very good idea, and would
be willing to help implement it if others thought
it was a good idea too.
-- 
Craig Rodrigues        
http://www.gis.net/~craigr    
rodrigc@mediaone.net          

Re: Using GNATS to track GCC patches

[Neil Booth]

Craig Rodrigues wrote:-

> I recall from the following thread that the idea
> of using GNATS to track patches in GCC came up:
> 
> http://gcc.gnu.org/ml/gcc/2000-06/msg00264.html
> http://gcc.gnu.org/ml/gcc/2000-06/msg00263.html
> 
> Has anyone made any move to implement this?

It might be a good idea, but I don't think that enough people take
GNATS seriously.  Only 4 or 5 of us use it regularly, and I think it's
fair to say that everyone with global write privs hardly uses it at
all.  Mark and rth seem to be the only ones that do, and only then
very occasionally.

So, I don't think this is a good idea at present, since it would just
clutter up GNATS.

I also feel that GNATS is a bit, um, flimsy and breaks easily, and is
not heavy-duty enough for what we'd really like.  We realy want proper
bug cross-referencing, cross-dependency tracking, a proper idea of
duplicates, etc, etc. much like Bugzilla does and GNATS doesn't appear
to do at all.

Neil.

Re: Using GNATS to track GCC patches

[Gerald Pfeifer]

On Sat, 17 Nov 2001, Neil Booth wrote:
> It might be a good idea, but I don't think that enough people take
> GNATS seriously.  [...]
> So, I don't think this is a good idea at present, since it would just
> clutter up GNATS.

I think it shouldn't necessarily be used for all patches, but for example
for patches that -- for whatever reason -- have not been reviewed even
after a reminder, to avoid those patches getting lost.

Then, as part of the release process, we could commit to reviewing all
GNATS reports of class "patches".

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

Re: Using GNATS to track GCC patches

[Joseph S. Myers]

On Sun, 18 Nov 2001, Gerald Pfeifer wrote:

> Then, as part of the release process, we could commit to reviewing all
> GNATS reports of class "patches".

Why only those PRs?  I think we should attempt to review all non-closed
PRs - not just regressions and patches, but an actual attempt, for each
PR, for someone competent in the relevant areas of the compiler to
diagnose the cause of the problem and find a fix.  This won't get
completely done, even in the two month stabilization period, but the
principle of actually looking at each PR would seem to be a good one.

-- 
Joseph S. Myers
jsm28@cam.ac.uk

Re: Using GNATS to track GCC patches

[Jakub Jelinek]

On Sat, Nov 17, 2001 at 06:12:30PM +0000, Neil Booth wrote:
> I also feel that GNATS is a bit, um, flimsy and breaks easily, and is
> not heavy-duty enough for what we'd really like.  We realy want proper
> bug cross-referencing, cross-dependency tracking, a proper idea of
> duplicates, etc, etc. much like Bugzilla does and GNATS doesn't appear
> to do at all.

BTW: What's the reason why bugzilla isn't used instead? GNATS doesn't work
at all with some browsers and I wonder what advantages does it give over
bugzilla. Is it just political issue?

	Jakub

Re: Using GNATS to track GCC patches

[Joseph S. Myers]

On Sun, 18 Nov 2001, Jakub Jelinek wrote:

> BTW: What's the reason why bugzilla isn't used instead? GNATS doesn't work

What was the status of bugzilla on 24 Aug 1999 (the date the GCC GNATS
database started)?

> at all with some browsers and I wonder what advantages does it give over
> bugzilla. Is it just political issue?

How does bugzilla compare to the wishlist I sent in
<URL:http://gcc.gnu.org/ml/gcc-bugs/2001-10/msg00649.html>?  Advocates of
other bug tracking systems can answer for the ones they prefer.  No-one on
gnats-devel answered as to how GNATS 4 compares.

-- 
Joseph S. Myers
jsm28@cam.ac.uk

Re: Using GNATS to track GCC patches

[Bryce McKinlay]

Joseph S. Myers wrote:

>>at all with some browsers and I wonder what advantages does it give over
>>bugzilla. Is it just political issue?
>>

While the GNATS system is somewhat better than nothing, I do not think 
that it fufills the needs of a large project like GCC well at all. I 
would very much like to see a better system installed, ie Bugzilla.

>How does bugzilla compare to the wishlist I sent in
><URL:http://gcc.gnu.org/ml/gcc-bugs/2001-10/msg00649.html>?  Advocates of
>other bug tracking systems can answer for the ones they prefer.  No-one on
>gnats-devel answered as to how GNATS 4 compares.
>

Bugzilla fufills most of these wishes, although I'm actually not sure to 
the extend of which bugs can be administered through the email 
interface. Its email interface is much more powerful that GNATS though, 
eg you have lots of control over what bugs get emailed to you when the 
status changes etc. GNATS gives you no options here, which I have always 
hated. The bugzilla interface is much faster to navigate and more 
refined than the clunky GNATSweb, so I would argue there is less need 
for an email admin interface anyway.

I may be able to devote some time to investigating bugzilla, say setting 
up a sample installation and importing the existing database. I think it 
could provide a huge benefit to GCC.

Bugzilla is, I think, licensed under the MPL, which is GPL-incompatible 
for linking purposes. However, 1) the Apache license is also 
GPL-incompatible and this hasn't stopped the FSF from running Apache, 
and 2) The mozilla people are relicensing much of their code under an 
MPL/GPL/LGPL "triple license" - though afaik this hasn't been done for 
bugzilla yet at this stage.

regards

Bryce.

Re: Using GNATS to track GCC patches

[Adrian von Bidder]

Bryce McKinlay wrote:

[gcc bts publicly available]

> It is.
> 
> http://gcc.gnu.org/cgi-bin/gnatsweb.pl

Ah. I see.

I just didn't try to login - since I have no account I always stopped at
the login page; it's not exactly inviting, is it? How about a notice
about using guest/guest or whatever is proper if one hasn't got an
account? 

So sorry everybody for this.

greets from Zürich
-- vbi

Re: Using GNATS to track GCC patches

[Phil Edwards]

On Mon, Nov 19, 2001 at 10:54:47AM +0100, Adrian von Bidder wrote:
> > http://gcc.gnu.org/cgi-bin/gnatsweb.pl
> 
> Ah. I see.
> 
> I just didn't try to login - since I have no account I always stopped at
> the login page; it's not exactly inviting, is it? How about a notice
> about using guest/guest or whatever is proper if one hasn't got an
> account? 

For that matter, if the username/password cookie(s) is(are) not present, why
not just assume 'guest'/'guest' and proceed with anonymous login?  The few
people that actually have a GNATS account can always click "login again",
and even that will only need to be done once (assuming they save the cookie).

"Few" compared to the number of anonymous users, that is.


Phil

-- 
If ye love wealth greater than liberty, the tranquility of servitude greater
than the animating contest for freedom, go home and leave us in peace.  We seek
not your counsel, nor your arms.  Crouch down and lick the hand that feeds you;
and may posterity forget that ye were our countrymen.            - Samuel Adams

Re: Using GNATS to track GCC patches

[Phil Edwards]

On Mon, Nov 19, 2001 at 03:52:53PM +1300, Bryce McKinlay wrote:
> status changes etc. GNATS gives you no options here, which I have always 
> hated. The bugzilla interface is much faster to navigate and more 
> refined than the clunky GNATSweb, so I would argue there is less need 
> for an email admin interface anyway.

Structured email messages are easier to generate automatically (e.g.,
from scripts) than HTTP requests.  For those of us who grew up in the
Old School, using email commands can be far more productive then clicking
through a web interface.

Having said that, I'll add my vote for replacing GNTAS with... oh heck,
just about anything.  The few times I've used Bugzilla it was pretty nice
(although that initial bug submission form is a tad intimidating, and it
requires users to create an "account").

The system that Debian uses is very nice; for example, rather than sending
follow-on data to a general email address and using a special string in
the Subject line (a la foo/1234), in Debian you just send to the PR number
(1234@bugs.blahblahblah.blah).


Phil

-- 
If ye love wealth greater than liberty, the tranquility of servitude greater
than the animating contest for freedom, go home and leave us in peace.  We seek
not your counsel, nor your arms.  Crouch down and lick the hand that feeds you;
and may posterity forget that ye were our countrymen.            - Samuel Adams