Re: dropping bug-buddy reports for old version of gnome automatically

2007-08-11 Thread Olav Vitters
On Sat, Apr 07, 2007 at 05:13:04PM +0200, Paolo Borelli wrote:
 As far as I am concerned I'd like to drop any reports from 2.16 and
 below, but I see how that may be a little to aggressive. What about
 dropping reports from 2.14 and below?

Bug-Buddy reports from 2.14 and older are now rejected at the mail
server (with a pointer to either upgrade or use the webinterface).

-- 
Regards,
Olav
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list


Re: dropping bug-buddy reports for old version of gnome automatically

2007-04-11 Thread Federico Mena Quintero
El mar, 10-04-2007 a las 19:02 -0600, Elijah Newren escribió:
 On 4/9/07, Federico Mena Quintero [EMAIL PROTECTED] wrote:
  Still, this can be an indication that your fixes are not reaching the
  affected people.  Maybe you need to explicitly poke the distro(s) in
  question and tell them, this fixes a bug that 5000 of your users have
  reported; can you release a new package through your update channels?.
 
 Which won't necessarily reach them either, as that assumes the users
 will all update their packages and there won't be new installations
 without automatic updates before usage.  (Bug 94625 received duplicate
 filings for _years_ after RedHat pushed out updates to fix the
 problem.  In fact, it received another one just last month.)

Could this happen:

1. bug-buddy tells bugzilla which distro/version you are using, and the
package version/revision.

2. bugzilla tells bug-buddy I know this has been fixed in
version'/revision' for that distro

3. bug-buddy tells you, This bug is already fixed in your distro.
Please update to package-version'-revision'.

This assumes that someone feeds bugzilla with the info it needs to do
(2), but that sounds doable with a little coordination in
distributor-list :)

Or maybe this can't happen during a run of bug-buddy, since the
duplicate bug needs to be triaged.  I don't know if we can steal some
ideas from here to make triaging automatic for frequent crashers:
http://www.fogcreek.com/FogBugz/docs/40/Articles/GetCrashReportsFromUsersA.html

[Are we ensuring that bugs reported through bug-buddy have a valid email
address?  Those bugs would be rather useless for the reporter without
one, because we can't notify the reporter later when the bug is fixed.]

  Federico

___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list

Re: dropping bug-buddy reports for old version of gnome automatically

2007-04-11 Thread Federico Mena Quintero
El mié, 11-04-2007 a las 10:31 -0600, Elijah Newren escribió:

 In other words, yes, what you ask can be done and _is_ being done.

This is totally awesome :)  I didn't know our Bugzilla infrastructure
was so nice.

 But to avoid false-positives, we require human intervention (and only
 a few people have the appropriate privileges) involving manual entry
 of data that should uniquely match the given bug.

OK.  What sort of data would you need?  Or can it be inferred from (say)
a bunch of duplicate bugs?

   If you have such a
 bug you'd like put on the auto-rejection list, contact the bugsquad
 with the information about the bug and a specifically tailored message
 you want bug-buddy to show the user.

This is great to know.  Is there a way we can notify all maintainers
about this?

[I don't seem to be getting bugmail for sabayon... I'll go poke the
bugmasters] :)

  Federico

___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list

Re: dropping bug-buddy reports for old version of gnome automatically

2007-04-11 Thread Elijah Newren
On 4/11/07, Federico Mena Quintero [EMAIL PROTECTED] wrote:
 This is totally awesome :)  I didn't know our Bugzilla infrastructure
 was so nice.

Olav rocks.  :-)

  But to avoid false-positives, we require human intervention (and only
  a few people have the appropriate privileges) involving manual entry
  of data that should uniquely match the given bug.

 OK.  What sort of data would you need?  Or can it be inferred from (say)
 a bunch of duplicate bugs?

edittraces.cgi currently asks for
  1) Product
  2) Product version   (can be blank, makes comparison test ignore this field)
  3) GNOME version  (can be blank; bug-buddy can't always detect GNOME version)
  4) Complete stack trace
  5) Reason (explanatory text sent to user when report is rejected)
I didn't develop it and haven't yet used it myself; further questions
are probably best directed to Olav or perhaps Andre or Karsten.

If you have such a
  bug you'd like put on the auto-rejection list, contact the bugsquad
  with the information about the bug and a specifically tailored message
  you want bug-buddy to show the user.

 This is great to know.  Is there a way we can notify all maintainers
 about this?

Good question.  I have no idea.

Cheers,
Elijah
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list


Re: dropping bug-buddy reports for old version of gnome automatically

2007-04-10 Thread Elijah Newren
On 4/9/07, Federico Mena Quintero [EMAIL PROTECTED] wrote:
 El sáb, 07-04-2007 a las 16:39 +0100, Richard Hughes escribió:

  I'm sure other maintainers must be getting as demoralised as myself when
  dealing with so many duplicates of a fixed bug.

 Still, this can be an indication that your fixes are not reaching the
 affected people.  Maybe you need to explicitly poke the distro(s) in
 question and tell them, this fixes a bug that 5000 of your users have
 reported; can you release a new package through your update channels?.

Which won't necessarily reach them either, as that assumes the users
will all update their packages and there won't be new installations
without automatic updates before usage.  (Bug 94625 received duplicate
filings for _years_ after RedHat pushed out updates to fix the
problem.  In fact, it received another one just last month.)
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list


Re: dropping bug-buddy reports for old version of gnome automatically

2007-04-10 Thread Elijah Newren
On 4/8/07, Olav Vitters [EMAIL PROTECTED] wrote:
 On Sun, Apr 08, 2007 at 04:01:14PM +0200, Josselin Mouette wrote:
  Given the length of Redhat, SuSE and Debian release cycles, this will
  just make bug-buddy useless for many stable users. Ideally you could

 Honestly, I think that =2.14 bugreports are useless. Dropping =2.16
 feels way too soon, however, I also want to keep Bugzilla usable (the
 incoming bug rate is way too high). Further, as soon as a GNOME is
 released as stable, the old one is forgotten about.
 Note: we don't receive many =2.14 bug-buddy bugreports (mostly due to
 using sendmail to send it). And it was going to be ceased with the
 release of 2.20 anyway.

[Responding a bit slowly...]

I'm all in favor of dropping = 2.14 bugreports, mainly because of the
sendmail and yet-another-bugzilla-parsing-script issue.  So +1 from me
on that.

I wouldn't want to see 2.16 bugreports dropped, especially not this
early.  It'd be really cool if we could make this kind of thing
product-specific and have it be a maintainer choice on which
bugreports are considered old enough to be automatically dropped in
the future.

Just my $0.02 (even though you probably already knew my answers) since
I didn't see too many other people responding to this specific
issue...

Elijah
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list


Re: dropping bug-buddy reports for old version of gnome automatically

2007-04-09 Thread Olav Vitters
On Sun, Apr 08, 2007 at 04:19:12PM +0200, Olav Vitters wrote:
 On Sun, Apr 08, 2007 at 04:01:14PM +0200, Josselin Mouette wrote:
  Le dimanche 08 avril 2007 à 14:20 +0200, Olav Vitters a écrit :
   On Sat, Apr 07, 2007 at 05:13:04PM +0200, Paolo Borelli wrote:
As far as I am concerned I'd like to drop any reports from 2.16 and
below, but I see how that may be a little to aggressive. What about
dropping reports from 2.14 and below?
   
   If there are no objections, I'll start with dropping bug-buddy reports
   created with bug-buddy =2.14 (meaning: GNOME versions before =2.14)
   starting Mon 9 April. I'll probably setup an [EMAIL PROTECTED] auto
   responder which tells them to upgrade their GNOME to the latest version
   available.
  
  Given the length of Redhat, SuSE and Debian release cycles, this will
  just make bug-buddy useless for many stable users. Ideally you could
 
 Honestly, I think that =2.14 bugreports are useless. Dropping =2.16
 feels way too soon, however, I also want to keep Bugzilla usable (the
 incoming bug rate is way too high). Further, as soon as a GNOME is
 released as stable, the old one is forgotten about.
 Note: we don't receive many =2.14 bug-buddy bugreports (mostly due to
 using sendmail to send it). And it was going to be ceased with the
 release of 2.20 anyway.

Adding some numbers (I mentioned this in
http://blogs.gnome.org/view/ovitters/2007/04/08/0):

Based on server stats over the last week, we:

 * Created 23 GNOME =2.14 bug-buddy reports. Update: To clarify, the
   =2.14 bug-buddy programs updated themselves 9603 times (it does that
   when you start it and it wasn't updated for at least 1 day, plus the
   server config changed). Obviously that method of reporting bugs is so
   broken that killing it wouldn't have a big impact.
 * Received 5513 XML-RPC GNOME 2.15 + 2.16 bug-buddy reports. I
   assume 60% are auto-rejected (received!=created).
 * Received 475 XML-RPC bug-buddy report for other GNOME versions
   (sometimes the bug-buddy version couldn't be determined).
 * Created 2405 bugreports in total (over the last 7 days instead of
   last week).

I hope above clarifies why I will kill =2.14, but want further input on
the other versions.


Note: The blog also has a comment (finally found why my comments weren't
working!:) about very hard to understand XML error messages. Someone
seen that as well?


-- 
Regards,
Olav
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list


Re: dropping bug-buddy reports for old version of gnome automatically

2007-04-09 Thread Federico Mena Quintero
El sáb, 07-04-2007 a las 16:39 +0100, Richard Hughes escribió:

 I'm sure other maintainers must be getting as demoralised as myself when
 dealing with so many duplicates of a fixed bug.

Still, this can be an indication that your fixes are not reaching the
affected people.  Maybe you need to explicitly poke the distro(s) in
question and tell them, this fixes a bug that 5000 of your users have
reported; can you release a new package through your update channels?.

  Federico

___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list

Re: dropping bug-buddy reports for old version of gnome automatically

2007-04-08 Thread Olav Vitters
On Sat, Apr 07, 2007 at 05:13:04PM +0200, Paolo Borelli wrote:
 As far as I am concerned I'd like to drop any reports from 2.16 and
 below, but I see how that may be a little to aggressive. What about
 dropping reports from 2.14 and below?

If there are no objections, I'll start with dropping bug-buddy reports
created with bug-buddy =2.14 (meaning: GNOME versions before =2.14)
starting Mon 9 April. I'll probably setup an [EMAIL PROTECTED] auto
responder which tells them to upgrade their GNOME to the latest version
available.

For dropping =2.16 I want to wait a bit longer for objections.


-- 
Regards,
Olav
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list


Re: dropping bug-buddy reports for old version of gnome automatically

2007-04-08 Thread Josselin Mouette
Le dimanche 08 avril 2007 à 14:20 +0200, Olav Vitters a écrit :
 On Sat, Apr 07, 2007 at 05:13:04PM +0200, Paolo Borelli wrote:
  As far as I am concerned I'd like to drop any reports from 2.16 and
  below, but I see how that may be a little to aggressive. What about
  dropping reports from 2.14 and below?
 
 If there are no objections, I'll start with dropping bug-buddy reports
 created with bug-buddy =2.14 (meaning: GNOME versions before =2.14)
 starting Mon 9 April. I'll probably setup an [EMAIL PROTECTED] auto
 responder which tells them to upgrade their GNOME to the latest version
 available.

Given the length of Redhat, SuSE and Debian release cycles, this will
just make bug-buddy useless for many stable users. Ideally you could
detect their distribution and tell them to report the issue directly in
the distribution's tracking system.

BTW, I'm working on a non-GNOME-dependent replacement to bug-buddy,
which will report crashes in the Debian BTS instead. In the future, it
should save you the Debian-related bugs (if we can manage the number of
bugs...)

-- 
 .''`.
: :' :  We are debian.org. Lower your prices, surrender your code.
`. `'   We will add your hardware and software distinctiveness to
  `-our own. Resistance is futile.


signature.asc
Description: Ceci est une partie de message	numériquement signée
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list

Re: dropping bug-buddy reports for old version of gnome automatically

2007-04-08 Thread Olav Vitters
On Sun, Apr 08, 2007 at 04:01:14PM +0200, Josselin Mouette wrote:
 Le dimanche 08 avril 2007 à 14:20 +0200, Olav Vitters a écrit :
  On Sat, Apr 07, 2007 at 05:13:04PM +0200, Paolo Borelli wrote:
   As far as I am concerned I'd like to drop any reports from 2.16 and
   below, but I see how that may be a little to aggressive. What about
   dropping reports from 2.14 and below?
  
  If there are no objections, I'll start with dropping bug-buddy reports
  created with bug-buddy =2.14 (meaning: GNOME versions before =2.14)
  starting Mon 9 April. I'll probably setup an [EMAIL PROTECTED] auto
  responder which tells them to upgrade their GNOME to the latest version
  available.
 
 Given the length of Redhat, SuSE and Debian release cycles, this will
 just make bug-buddy useless for many stable users. Ideally you could

Honestly, I think that =2.14 bugreports are useless. Dropping =2.16
feels way too soon, however, I also want to keep Bugzilla usable (the
incoming bug rate is way too high). Further, as soon as a GNOME is
released as stable, the old one is forgotten about.
Note: we don't receive many =2.14 bug-buddy bugreports (mostly due to
using sendmail to send it). And it was going to be ceased with the
release of 2.20 anyway.

 detect their distribution and tell them to report the issue directly in
 the distribution's tracking system.

You mean in the auto responder right? I can add a comment that their
should report it to their distribution. I assume people will know what
distribution they use.

 BTW, I'm working on a non-GNOME-dependent replacement to bug-buddy,
 which will report crashes in the Debian BTS instead. In the future, it
 should save you the Debian-related bugs (if we can manage the number of
 bugs...)

Suggest to work together on this. Ubuntu has something which already
works (apport). Fedora wants something similar, etc for other
distributions. Oh, and a warning: without having some system in place to
detect duplicates automatically, you will be overwhelmed. 
I do wonder about the 'non-GNOME-dependent' comment though? Does it mean
'GTK+' / 'Qt' or just like apport that it works for every app (kernel
integrated)?

Ideally the major distributions would have something like apport, where
at least most of the code is shared between the distributions. This
should work for every application in the distribution.

It would still need to handle the jhbuild / GARNOME case though.


-- 
Regards,
Olav
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list


Re: dropping bug-buddy reports for old version of gnome automatically

2007-04-08 Thread Alex Jones
On Sun, 2007-04-08 at 16:01 +0200, Josselin Mouette wrote:
 BTW, I'm working on a non-GNOME-dependent replacement to bug-buddy,
 which will report crashes in the Debian BTS instead. In the future, it
 should save you the Debian-related bugs (if we can manage the number of
 bugs...)

What's wrong with Apport?

___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list


Re: dropping bug-buddy reports for old version of gnome automatically

2007-04-08 Thread Josselin Mouette
Le dimanche 08 avril 2007 à 16:19 +0200, Olav Vitters a écrit :
 Oh, and a warning: without having some system in place to
 detect duplicates automatically, you will be overwhelmed. 

Definitely. This, and the fact such reports are mostly useless without
debugging symbols available. I have some ideas on how to handle the
first issue, but it's quite entangled with the BTS and I want to see the
system in action before. For the debugging symbols issue, we have some
ideas about their automated installation, but it remains to be done.

Reports on stable releases are another issue, and I'm reluctant to ship
such automated tools (including bug-buddy) in stable releases unless we
find ways to identify bugs that we'll really want to fix (like security
issues).

 I do wonder about the 'non-GNOME-dependent' comment though? Does it mean
 'GTK+' / 'Qt' or just like apport that it works for every app (kernel
 integrated)?

I chose the LD_PRELOAD way, so that it remains entirely optional and can
run with any kernel.

 Ideally the major distributions would have something like apport, where
 at least most of the code is shared between the distributions. This
 should work for every application in the distribution.

It turns out that the shared code would be very small. The LD_PRELOAD
library (that's mostly stolen from libgnomeui) is a tiny piece of code,
and I suspect the kernel patch is likewise. The more important part of
the code is the integration with the distribution's own packaging system
and bug tracker.

 It would still need to handle the jhbuild / GARNOME case though.

Currently Debreaper doesn't override applications setting their own
signal handlers, but for GNOME it overrides gnome_segv2. As GARNOME
installations have their own gnome_segv2, this should bring bug-buddy as
expected.

-- 
 .''`.
: :' :  We are debian.org. Lower your prices, surrender your code.
`. `'   We will add your hardware and software distinctiveness to
  `-our own. Resistance is futile.


signature.asc
Description: Ceci est une partie de message	numériquement signée
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list

Re: dropping bug-buddy reports for old version of gnome automatically

2007-04-07 Thread Thomas Wood
On 07/04/07 16:13, Paolo Borelli wrote:
 Hi,
 
[...]
 
 Personally I am not able anymore to handle my bugmail anymore and even
 useful bugreports get lost in the noise.

I usually mark any Crash: ... bugmail as read without even looking at 
it nowadays.

 As a first step, what do you think of just dropping silently at least
 crash reports from older versions of gnome? All bugs coming from there
 are either already fixed or will be reported anyway if still present in
 newer versions.

Is there any way to do it non-silently? There needs to be something to 
at least inform the user they have an ancient version of GNOME and 
should consider upgrading.

Regards,

Thomas
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list


Re: dropping bug-buddy reports for old version of gnome automatically

2007-04-07 Thread Richard Hughes
On Sat, 2007-04-07 at 17:13 +0200, Paolo Borelli wrote:
 Personally I am not able anymore to handle my bugmail anymore and even
 useful bugreports get lost in the noise. 

Tell me about it. gnome-power-manager 2.18.0 had a bug where it would
segfault when you locked the screen if HAL was not running. 2.18.1 fixed
the bug, but that still means I get about 5-10 duplicate bugzillas a
day. The guys triaging this stuff have been great, but it's still more
and more emails for me.

I'm really thinking the auto-submit thing to bugzilla should have more
text like:

1. You do not have debuginfo packages installed, the trace may not be
useful to the developer.

2. [click here to check to see if the crash was already reported]

3. You are running an old version. This crash may be fixed in newer
versions of the product.

Point 1 assumes we can tell what useful debuginfo packages should be
installed for each package.
Point 2 assumes we can automatically check for similar stacktraces.
Point 3 assumes the bugbuddy can query the servers for the latest
version.

I'm sure other maintainers must be getting as demoralised as myself when
dealing with so many duplicates of a fixed bug.

Richard.


___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list


Re: dropping bug-buddy reports for old version of gnome automatically

2007-04-07 Thread Andrew Sobala
Richard Hughes wrote:

On Sat, 2007-04-07 at 17:13 +0200, Paolo Borelli wrote:
  

Personally I am not able anymore to handle my bugmail anymore and even
useful bugreports get lost in the noise. 



Tell me about it. gnome-power-manager 2.18.0 had a bug where it would
segfault when you locked the screen if HAL was not running. 2.18.1 fixed
the bug, but that still means I get about 5-10 duplicate bugzillas a
day. The guys triaging this stuff have been great, but it's still more
and more emails for me.
  


We can auto reject specific stack traces [1] for this very reason. Get 
in touch with your local bugsquadder/bugmaster for more information.

I'm really thinking the auto-submit thing to bugzilla should have more
text like:

1. You do not have debuginfo packages installed, the trace may not be
useful to the developer.
  

This happens [2] if we manage to detect it..

Thanks,

Andrew

[1] http://blogs.gnome.org/view/ovitters/2006/11/12/0
[2] http://blogs.gnome.org/view/ovitters/2007/01/06/0
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list


Re: dropping bug-buddy reports for old version of gnome automatically

2007-04-07 Thread Elijah Newren
[Just noticed that Andrew made most of my email obsolete before I
finished.  Maybe you'll see something useful, though, so I'll still
send it.]

On 4/7/07, Richard Hughes [EMAIL PROTECTED] wrote:
 On Sat, 2007-04-07 at 17:13 +0200, Paolo Borelli wrote:
  Personally I am not able anymore to handle my bugmail anymore and even
  useful bugreports get lost in the noise.

 Tell me about it. gnome-power-manager 2.18.0 had a bug where it would
 segfault when you locked the screen if HAL was not running. 2.18.1 fixed
 the bug, but that still means I get about 5-10 duplicate bugzillas a
 day. The guys triaging this stuff have been great, but it's still more
 and more emails for me.

You can have such bugs automatically rejected with a specialized
notice (Thanks for the bug report, but this issue has been fixed in
gnome-power-manager 2.18.1; please update your software.)  One
example is bug 352592 which has had over 3000 duplicates automatically
rejected[1].  See http://live.gnome.org/Bugsquad/AutoReject for more
details.  I believe Olav set the system up so that only a few users
(maybe 3 or 4?) had permissions to add bugs to the auto-reject list,
so you'll probably need to email the bugsquad to ask someone to do it
for you.

 I'm really thinking the auto-submit thing to bugzilla should have more
 text like:

 1. You do not have debuginfo packages installed, the trace may not be
 useful to the developer.

http://live.gnome.org/ImprovingDebugging

 2. [click here to check to see if the crash was already reported]

We have no fail proof method for determining if two bugs are indeed a
duplicate.  Having that would be great, but we simply don't have it.
Help welcome (Olav may already be working on it, so ask him first
instead of starting from scratch.)

 3. You are running an old version. This crash may be fixed in newer
 versions of the product.

Rejecting on age would need to be product-specific, as which versions
to reject probably depends on the volume of email the product
receives.  We've asked maintainers if they wanted to have bugsquadders
automatically close bugs if against versions that are too old and what
too old meant for their product, so moving that logic into bugzilla
itself seems like a reasonable enhancement.  Default would be to not
reject anything based on version, of course.

 Point 1 assumes we can tell what useful debuginfo packages should be
 installed for each package.
 Point 2 assumes we can automatically check for similar stacktraces.
 Point 3 assumes the bugbuddy can query the servers for the latest
 version.

 I'm sure other maintainers must be getting as demoralised as myself when
 dealing with so many duplicates of a fixed bug.

[1] I suggest only doing this for fixed bugs, where further reports
won't be helpful; otherwise, you miss out on potential extra details
that future bug submitters might provide.

Hope that helps,
Elijah
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
http://mail.gnome.org/mailman/listinfo/desktop-devel-list