Bugzilla

Bugzilla

[Nick Roberts]

There seems to be disconnect between GDB bugzilla and the gdb mailing lists.
Bugs and patches submitted there seem to receive scant attention and I often
recommend posting to the mailing lists.

Would it be a good idea to configure Bugzilla to CC the gdb-patches mailing
lists so that traffic there gets more attention?

-- 
Nick                                           http://users.snap.net.nz/~nickrob

Re: Bugzilla

[Doug Evans]

On Tue, Oct 13, 2009 at 5:56 PM, Nick Roberts <nickrob@snap.net.nz> wrote:
> There seems to be disconnect between GDB bugzilla and the gdb mailing lists.
> Bugs and patches submitted there seem to receive scant attention and I often
> recommend posting to the mailing lists.
>
> Would it be a good idea to configure Bugzilla to CC the gdb-patches mailing
> lists so that traffic there gets more attention?

Hi.  The mailing list web page says gdb-prs@sourceware.org is the list
to use, though there's still a problem as the archives only have one
message from 2004.  Heh.

http://sourceware.org/gdb/mailing-lists/

I like the idea of a separate list.

Re: Bugzilla

[Tom Tromey]

>>>>> "Doug" == Doug Evans <dje@google.com> writes:

Doug> Hi.  The mailing list web page says gdb-prs@sourceware.org is the list
Doug> to use, though there's still a problem as the archives only have one
Doug> message from 2004.  Heh.
Doug> http://sourceware.org/gdb/mailing-lists/

Yeah, something is screwy with the archive.
I haven't investigated it.

Doug> I like the idea of a separate list.

The mail actually does go to gdb-prs.
Or you can read it in gmane as gmane.comp.gdb.bugs.discuss.

Tom

Re: Bugzilla

[Nick Roberts]

Tom Tromey writes:
 > >>>>> "Doug" == Doug Evans <dje@google.com> writes:
 > 
 > Doug> Hi.  The mailing list web page says gdb-prs@sourceware.org is the list
 > Doug> to use, though there's still a problem as the archives only have one
 > Doug> message from 2004.  Heh.
 > Doug> http://sourceware.org/gdb/mailing-lists/
 > 
 > Yeah, something is screwy with the archive.
 > I haven't investigated it.
 > 
 > Doug> I like the idea of a separate list.
 > 
 > The mail actually does go to gdb-prs.
 > Or you can read it in gmane as gmane.comp.gdb.bugs.discuss.

I don't think it actually goes to gdb-prs.  The Reply-To: header is
sourceware-bugzilla@sourceware.org while the To: header is
gdb-prs@sourceware.org.  I hit 'reply' in VM which goes to bugzilla.  If I hit
'followup' which tries to go to both addresses, the message gets sent back to
me.  This would explain why there is only one message in gdb-prs.

In any case, my point is that GDB Bugzilla is a bit of a black hole: more
goes in than comes out.  If reports did go to gdb-prs, they would still go
unread as presumably no-one is subscribed to it. gdb and gdb-patches are
the only mailing lists that appear to be widely read.

-- 
Nick                                           http://users.snap.net.nz/~nickrob

Re: Bugzilla

[Tom Tromey]

>>>>> "Nick" == Nick Roberts <nickrob@snap.net.nz> writes:

Nick> I don't think it actually goes to gdb-prs.  The Reply-To: header
Nick> is sourceware-bugzilla@sourceware.org while the To: header is
Nick> gdb-prs@sourceware.org.  I hit 'reply' in VM which goes to
Nick> bugzilla.  If I hit 'followup' which tries to go to both
Nick> addresses, the message gets sent back to me.

Strange, I thought it worked.  You can ask overseers to fix it.
The email does go somewhere because the gmane group works and it only
gets gdb bugs -- not bugs from other components in the sourceware
bugzilla.

Nick> In any case, my point is that GDB Bugzilla is a bit of a black
Nick> hole: more goes in than comes out.  If reports did go to gdb-prs,
Nick> they would still go unread as presumably no-one is subscribed to
Nick> it. gdb and gdb-patches are the only mailing lists that appear to
Nick> be widely read.

I read all the traffic.

I thought about this and I don't mind where the email goes.  I'm going
to read it regardless.  I'm skeptical that sending it to gdb-patches
will help the situation much, but I'll set it up if that is what the
rest of the community wants.

Tom

Re: Bugzilla

[Daniel Jacobowitz]

On Fri, Oct 16, 2009 at 07:05:36PM -0600, Tom Tromey wrote:
> I thought about this and I don't mind where the email goes.  I'm going
> to read it regardless.  I'm skeptical that sending it to gdb-patches
> will help the situation much, but I'll set it up if that is what the
> rest of the community wants.

Sending it to gdb-patches will not make more people pay attention to
it, just irritate more people who have to work harder to delete it.
Those of us with time and patience to deal with the bugzilla traffic
already get it from gdb-prs (which is a Bugzilla-only list; you can't
post to it, only Bugzilla can; reply to the Bugzilla address).

If we need more people working on the bug system, that's a people
problem, not a list configuration problem.

-- 
Daniel Jacobowitz
CodeSourcery

Re: Bugzilla

[Nick Roberts]

 > Sending it to gdb-patches will not make more people pay attention to
 > it, just irritate more people who have to work harder to delete it.

Actually, I think it would initially probably do both.  Messages on
gdb-patches certainly get more attention than bug reports.

 > Those of us with time and patience to deal with the bugzilla traffic
 > already get it from gdb-prs (which is a Bugzilla-only list; you can't
 > post to it, only Bugzilla can; reply to the Bugzilla address).

If bug reports were cc'ed to a mailing list which you already are subscribed
to then presumably you could unsubscribe from gdb-prs so that you still only
receive each bug report once.

 > If we need more people working on the bug system, that's a people
 > problem, not a list configuration problem.

I disagree.  In part it is a configuration problem because the ideal outcome
of a bug report is a patch.  Currently to get attention this requires
cross-posting to gdb-patches which I think is more irritating because
unsubscribing from gdb-prs would result in missing or incomplete bug reports.

In any case, I've unsubscribed from gdb-prs now as I've found it to be a waste
of my time and I won't be submitting any more patches for bugs reported there.

-- 
Nick                                           http://users.snap.net.nz/~nickrob

Re: Bugzilla

[Daniel Jacobowitz]

On Sun, Oct 18, 2009 at 03:19:07PM +1300, Nick Roberts wrote:
>  > If we need more people working on the bug system, that's a people
>  > problem, not a list configuration problem.
> 
> I disagree.  In part it is a configuration problem because the ideal outcome
> of a bug report is a patch.  Currently to get attention this requires
> cross-posting to gdb-patches which I think is more irritating because
> unsubscribing from gdb-prs would result in missing or incomplete bug reports.
> 
> In any case, I've unsubscribed from gdb-prs now as I've found it to be a waste
> of my time and I won't be submitting any more patches for bugs reported there.

I don't get it.  If you are choosing not to receive the bug traffic,
by doing this, why would you argue in favor of making everyone receive
that traffic?

More people getting the mail doesn't mean more people with the time to
read, respond, and patch for it.

The problem isn't that no one knows how to follow the bug system, just
the limited number of people with the patience and time to do so.

-- 
Daniel Jacobowitz
CodeSourcery

Re: Bugzilla

[Joel Brobecker]

> The problem isn't that no one knows how to follow the bug system, just
> the limited number of people with the patience and time to do so.

Fully agreed.

-- 
Joel

Re: Bugzilla

[Nick Roberts]

Joel Brobecker writes:
 > > The problem isn't that no one knows how to follow the bug system, just
 > > the limited number of people with the patience and time to do so.
 > 
 > Fully agreed.

I guess you're right.  If the responsible maintainer has neither the time nor
the patience to review the patch in Bugzilla, cc'ing it to gdb-patches
probably won't help.

-- 
Nick                                           http://users.snap.net.nz/~nickrob

Re: Bugzilla

[Joel Brobecker]

> I guess you're right.  If the responsible maintainer has neither the time nor
> the patience to review the patch in Bugzilla, cc'ing it to gdb-patches
> probably won't help.

Just to be clear: If you send a patch to gdb-patches, I will do everything
I can to make sure that it is reviewed. If you notice a problem and report
it either on GDB or bugzilla, I might not be able to give it much attention.

-- 
Joel

Re: Bugzilla

['Daniel Jacobowitz']

On Wed, Mar 21, 2007 at 12:29:46PM +0100, Sascha Radike wrote:
> It might also be helpful if all discussions about enhancements and propasals
> would be moved to bugzilla - just like it is handled by the eclipse
> community.

No thanks.  I think we have too many people who are used to working
this way.  Of course, making the bugs mailing list more helpful might
change that.

-- 
Daniel Jacobowitz
CodeSourcery

Re: Bugzilla

[Sascha Radike]

>If so I can make time to get it done, but otherwise I have plenty of other
projects ;-)

I don't know much time it takes to setup bugzilla. But I think a running
bugzilla might reduce time spent for unnecessary work like identifying bugs
already fixed ;) or searching the mailing lists.

It might also be helpful if all discussions about enhancements and propasals
would be moved to bugzilla - just like it is handled by the eclipse
community.

I'd prefer bugzilla being the central platform for bugs, enhancements and
fixes rather than searching through mailing lists, changelog, gnatsweb and
problems.

But that's just my 2 cents.

Sascha

Re: Bugzilla

[Daniel Jacobowitz]

On Wed, Mar 21, 2007 at 08:59:47AM +0100, Sascha Radike wrote:
> Hi,
> 
> is there any date planned for switching to bugzilla? I think this would be
> very helpful as bugzilla is just much more comfortable than the current bug
> tracker. 
> 
> I don't know if it is possible to migrate from gnatsweb to bugzilla, but I
> guess a simple solution would be to mark the gnatsweb as 'old' and setup a
> new, fresh bugzilla. Maybe it is possible to send out a message to all users
> who raised a bug and ask them to add their bug to bugzilla if the bug still
> exists. Bugs being addressed for the next release could also be moved to
> bugzilla.
> 
> I think it would be an improvement to the GDB project.

We meant to do it ages ago.  I was actually thinking about it again
last night...

We can automatically migrate bugs to bugzilla.  My understanding is
that the only reasonable way to do it is to add ourselves to the
sourceware.org bugzilla instance, which will renumber all the PRs.
But we should be able to keep the old gnats around for read access
somehow.

Do other people think the time has come to switch over?  If so I can
make time to get it done, but otherwise I have plenty of other
projects ;-)

-- 
Daniel Jacobowitz
CodeSourcery

Bugzilla

[Sascha Radike]

Hi,

is there any date planned for switching to bugzilla? I think this would be
very helpful as bugzilla is just much more comfortable than the current bug
tracker. 

I don't know if it is possible to migrate from gnatsweb to bugzilla, but I
guess a simple solution would be to mark the gnatsweb as 'old' and setup a
new, fresh bugzilla. Maybe it is possible to send out a message to all users
who raised a bug and ask them to add their bug to bugzilla if the bug still
exists. Bugs being addressed for the next release could also be moved to
bugzilla.

I think it would be an improvement to the GDB project.

Sascha

Re: bugzilla

[Michael Elizabeth Chastain]

> So if you guys want new fields, they need to be all decided before the 
> merge, so i can add all the necessary display stuff, etc :).

Fair enough.

Here is my list of fields.  I started with the gcc bugzilla list:

  Reporter
  Product
  Version
  Component
  Priority
  Severity
  Assigned To
  CC
  Host Triplet
  Target Triplet
  Build Triplet
  Summary
  Description

The additional fields I want are:

  Target Operating System Version
    (such as red hat linux 8, etc)
  Compiler Name
    (gcc, hp ansi c, hp acc, various sun compilers, diab, etc)
  Compiler Version
    (The compiler that you built your program with,
     not the compiler that you built gcc with.
     If gdb successfully builds, then gdb works the same no
     matter what compiler you build it with.  However, gdb
     reads debug information from your program, so we need
     the name and version of the compiler which you used to
     compile your program)
  Debug Format
    (dwarf-2, stabs+, som, mdebug, ..., other)

mec> . collapse 'priority' and 'severity' into one field,
mec>   or actually document what they mean in the online help.
db> I can't do this if you use the sources bugzilla installation, since 
db> they use priority and severity.

This is low priority to me.  I can live with whatever is easiest for
you, including no changes at all.

Michael C

Re: bugzilla

[Daniel Berlin]

On Jul 8, 2004, at 2:02 PM, Michael Elizabeth Chastain wrote:

> I like bugzilla for gcc just great!
> In particular, it's much easier to add attachments.
>
> Some requests:
>
> . add target/host/build fields similar to gcc bugzilla
>
> . add a field for the compiler used to build the test program.
>   i don't care what compiler is used to build gdb itself
>   (unless it's a build failure or a failure in selftest.exp).
>   i care a lot what compiler is used to write the program that
>   gdb is reading.  occasionally i care about the binutils version
>   as well, but perhaps not enough to have a field for it.

Adding new fields is a pain in the ass, and i try to do it when only 
*absolutely necessary*.
This is because we don't use any of the generic custom fields patches 
to bugzilla,  since none of them as they stand will ever make it into 
the bugzilla proper, and thus, using one would make merging to new 
bugzilla versions incredibly difficult (they always cause code 
conflicts).

So if you guys want new fields, they need to be all decided before the 
merge, so i can add all the necessary display stuff, etc :).
Not that i've never added a custom field later on, it's just that I try 
to avoid it whenever possible.
At least until bugzilla itself gets a custom fields mechanism, which is 
probably eons away.
I'm only one person, after all, and Bugzilla was just something i 
started doing to make life easier for gcc developers.  It's not 
anywhere near my job description, and i actually hate perl.

> . add a field for debug format.  this can be drop down:
>   dwarf-2, stabs+, som, mdebug, ..., other.

> . add instructions on running the 'script' command,
>   or running gdb inside emacs and capturing the session.
>   see a recent version of doc/gdb.info.
>
> . let us know where the online help is and how to edit it.
>
> . collapse 'priority' and 'severity' into one field,
>   or actually document what they mean in the online help.

I can't do this if you use the sources bugzilla installation, since 
they use priority and severity.
It would be possible to simply hide one of the fields when displaying 
gdb bugs, of course (IE you could just use severity, or just priority, 
for gdb bugs).

Re: bugzilla

[Michael Elizabeth Chastain]

db> Bugs id's are global to a given bugzilla, not local to a given product.
db> There exist bugs starting with the number one in there already.  Thus, 
db> gdb bugs would have to start with the highest number at the time of the 
db> import.

Erk, we use the bug number in other places.  It's a KFAIL number,
and it appears in PROBLEMS, and I link to them in my reports,
and we talk about them by number.

I guess we could have a flag day.  As far as I know, there's
no automated systems that will break, just some dangling links
that we will have.

Could we convert after the 6.2 release?

Michael C

Re: bugzilla

[Michael Elizabeth Chastain]

I like bugzilla for gcc just great!
In particular, it's much easier to add attachments.

Some requests:

. add target/host/build fields similar to gcc bugzilla

. add a field for the compiler used to build the test program.
  i don't care what compiler is used to build gdb itself
  (unless it's a build failure or a failure in selftest.exp).
  i care a lot what compiler is used to write the program that
  gdb is reading.  occasionally i care about the binutils version
  as well, but perhaps not enough to have a field for it.

. add a field for debug format.  this can be drop down:
  dwarf-2, stabs+, som, mdebug, ..., other.

. add instructions on running the 'script' command,
  or running gdb inside emacs and capturing the session.
  see a recent version of doc/gdb.info.

. let us know where the online help is and how to edit it.

. collapse 'priority' and 'severity' into one field,
  or actually document what they mean in the online help.

Michael C

Re: bugzilla

[Daniel Berlin]

>> Why is a renumbering needed?  It's our database can't we do what we
>> like? :-)
>
> GNATS numbers bugs independently per product; Bugzilla doesn't.
> There's one bugzilla installation on sources.redhat.com already, and
> it's shared with binutils and glibc.
>
> Um, I suppose we could set up a third copy of Bugzilla on that machine.
> I don't know what the issues with that would be but presumably Daniel
> Berlin does.
>

The only real problem with doing that is the pain-in-the-assness of 
updating three bugzilla installations every year or so when i merge new 
stable bugzilla code.
However, if that's what you guys want, ....

Re: bugzilla

[Daniel Berlin]

On Jul 8, 2004, at 11:06 AM, Andrew Cagney wrote:

>> On Jul 8, 2004, at 10:31 AM, Andrew Cagney wrote:
>>> Please discuss :-)
>>>
>>> -- 
>>>
>>> I've cc'd DanielB and ChrisF who would need to be involved in any 
>>> logistical matters.
>>>
>> I've converted the GDB GNATS database before with little or no 
>> problems.
>> I can put the results up on www.dberlin.org/bugzilla if you want to 
>> see what it looks like
>> The only real problem in conversion is that the bugs will not have 
>> the same PR number they used to, though they will be contiguous (IE 
>> they will be offset by some number).
>
> Why is a renumbering needed?  It's our database can't we do what we 
> like? :-)

I was planning on putting you guys in the same bugzilla as the other 
sources projects (sources.redhat.com/bugzilla).
Bugs id's are global to a given bugzilla, not local to a given product.
There exist bugs starting with the number one in there already.  Thus, 
gdb bugs would have to start with the highest number at the time of the 
import.

Re: bugzilla

[Daniel Berlin]

On Jul 8, 2004, at 10:57 AM, Dave Korn wrote:

>> -----Original Message-----
>> From: gdb-owner On Behalf Of Andrew Cagney
>> Sent: 08 July 2004 15:31
>> To: gdb
>> Cc: Christopher Faylor; Daniel Berlin
>> Subject: bugzilla
>>
>> Please discuss :-)
>>
>> --
>>
>> I've cc'd DanielB and ChrisF who would need to be involved in any
>> logistical matters.
>
>
>   I'll say one thing about the current system: PR1700 is almost 
> certainly
> bogus......

And PR 1628 is 60 meg :)
>

Re: bugzilla

[Daniel Jacobowitz]

On Thu, Jul 08, 2004 at 11:06:42AM -0400, Andrew Cagney wrote:
> >
> >On Jul 8, 2004, at 10:31 AM, Andrew Cagney wrote:
> >
> >>Please discuss :-)
> >>
> >>-- 
> >>
> >>I've cc'd DanielB and ChrisF who would need to be involved in any 
> >>logistical matters.
> >>
> >I've converted the GDB GNATS database before with little or no problems.
> >I can put the results up on www.dberlin.org/bugzilla if you want to see 
> >what it looks like
> >
> >The only real problem in conversion is that the bugs will not have the 
> >same PR number they used to, though they will be contiguous (IE they will 
> >be offset by some number).
> 
> Why is a renumbering needed?  It's our database can't we do what we 
> like? :-)

GNATS numbers bugs independently per product; Bugzilla doesn't. 
There's one bugzilla installation on sources.redhat.com already, and
it's shared with binutils and glibc.

Um, I suppose we could set up a third copy of Bugzilla on that machine. 
I don't know what the issues with that would be but presumably Daniel
Berlin does.

-- 
Daniel Jacobowitz

Re: bugzilla

[Andrew Cagney]

> 
> On Jul 8, 2004, at 10:31 AM, Andrew Cagney wrote:
> 
>> Please discuss :-)
>>
>> -- 
>>
>> I've cc'd DanielB and ChrisF who would need to be involved in any logistical matters.
>>
> I've converted the GDB GNATS database before with little or no problems.
> I can put the results up on www.dberlin.org/bugzilla if you want to see what it looks like
> 
> The only real problem in conversion is that the bugs will not have the same PR number they used to, though they will be contiguous (IE they will be offset by some number).

Why is a renumbering needed?  It's our database can't we do what we 
like? :-)

Andrew

RE: bugzilla

[Dave Korn]

> -----Original Message-----
> From: gdb-owner On Behalf Of Andrew Cagney
> Sent: 08 July 2004 15:31
> To: gdb
> Cc: Christopher Faylor; Daniel Berlin
> Subject: bugzilla
> 
> Please discuss :-)
> 
> --
> 
> I've cc'd DanielB and ChrisF who would need to be involved in any 
> logistical matters.


  I'll say one thing about the current system: PR1700 is almost certainly
bogus...... 


    cheers, 
      DaveK
-- 
Can't think of a witty .sigline today....

Re: bugzilla

[Daniel Berlin]

On Jul 8, 2004, at 10:31 AM, Andrew Cagney wrote:

> Please discuss :-)
>
> --
>
> I've cc'd DanielB and ChrisF who would need to be involved in any 
> logistical matters.
>
I've converted the GDB GNATS database before with little or no problems.
I can put the results up on www.dberlin.org/bugzilla if you want to see 
what it looks like

The only real problem in conversion is that the bugs will not have the 
same PR number they used to, though they will be contiguous (IE they 
will be offset by some number).
--Dan

bugzilla

[Andrew Cagney]

Please discuss :-)

--

I've cc'd DanielB and ChrisF who would need to be involved in any 
logistical matters.