pretest, devel and bug lists

pretest, devel and bug lists

[Glenn Morris]

If things sent to emacs-pretest-bug are being fed into a bug tracker
and sent to bug-gnu-emacs, then I would say there is no need for
copies to _also_ be sent to emacs-devel.

I also don't see the need for tracker control messages to be sent to
bug-gnu-emacs, but I'm less sure about that.

It would also be nice if there was a statement as to whether this
tracker is now offical and we should all start learning how to use it,
or if it is still being tested.

RE: pretest, devel and bug lists

[Drew Adams]

> If things sent to emacs-pretest-bug are being fed into a bug tracker
> and sent to bug-gnu-emacs, then I would say there is no need for
> copies to _also_ be sent to emacs-devel.

I agree. Bug tracking and bugs discussions should be in a separate mailing list
from emacs-devel. If a particular bug thread develops into a development
discussion, it can be moved to emacs-devel. (Yes, I know that pretest bugs can
often involve development discussions.)

> I also don't see the need for tracker control messages to be sent to
> bug-gnu-emacs, but I'm less sure about that.

I think I disagree. I like everything about Emacs bugs to be sent to the bugs
list. But I too am less sure about this.

Re: pretest, devel and bug lists

[Jason Rumney]

Drew Adams wrote:
> I think I disagree. I like everything about Emacs bugs to be sent to the bugs
> list. But I too am less sure about this.
>   

Many of the control messages are uninteresting - spam being closed, bugs 
being tagged with various tags. The control messages that are 
interesting mostly have alternatives that result in a message being 
added to the bug report, such as using the ###-done address instead of 
sending "close ###" to the control address.

Re: pretest, devel and bug lists

[Richard M Stallman]

I think that all the messages from the bug tracker
should be sent to a separate list, and NONE should be sent
to emacs-devel or bug-gnu-emacs.

Anyone interested in seeing those messages can put himself
on the new list.

Re: pretest, devel and bug lists

[Stefan Monnier]

> I think that all the messages from the bug tracker
> should be sent to a separate list, and NONE should be sent
> to emacs-devel or bug-gnu-emacs.

The way things are setup now:

- all mail to bug-gnu-emacs first goes through the bug-tracker.
  I.e. all mail received by bug-gnu-emacs recipients comes from the
  bug-tracker.
- emacs-pretest-bug and emacs-devel are competely separate from the
  bug-tracker.

So we have exactly what you describe, except that bug-gnu-emacs is now
the "separate bug-tracker list".


        Stefan

Re: pretest, devel and bug lists

[Sven Joachim]

On 2008-05-28 20:00 +0200, Stefan Monnier wrote:

> The way things are setup now:
>
> - all mail to bug-gnu-emacs first goes through the bug-tracker.
>   I.e. all mail received by bug-gnu-emacs recipients comes from the
>   bug-tracker.
> - emacs-pretest-bug and emacs-devel are competely separate from the
>   bug-tracker.

Does this mean that bug reports sent to emacs-pretest-bug will not
get assigned a bug number anymore?

Sven

Re: pretest, devel and bug lists

[Stefan Monnier]

>> The way things are setup now:
>> 
>> - all mail to bug-gnu-emacs first goes through the bug-tracker.
>> I.e. all mail received by bug-gnu-emacs recipients comes from the
>> bug-tracker.
>> - emacs-pretest-bug and emacs-devel are competely separate from the
>> bug-tracker.

> Does this mean that bug reports sent to emacs-pretest-bug will not
> get assigned a bug number anymore?

They never have, AFAIK (except for a temporarily bug in my mail filter
and except for those that someone manually forwarded to the bug-tracker).


        Stefan

Re: pretest, devel and bug lists

[David De La Harpe Golden]

Stefan Monnier wrote:
>> I think that all the messages from the bug tracker
>> should be sent to a separate list, and NONE should be sent
>> to emacs-devel or bug-gnu-emacs.
> 
> So we have exactly what you describe, except that bug-gnu-emacs is now
> the "separate bug-tracker list".
> 

The donarmstrong.com  emacsbugs mailing list also still exists.
It only seems to get bits and pieces though, I think maybe it's a loose
end from early stages of Don's implementation?

Re: pretest, devel and bug lists

[Don Armstrong]

On Wed, 28 May 2008, David De La Harpe Golden wrote:
> Stefan Monnier wrote:
> >> I think that all the messages from the bug tracker
> >> should be sent to a separate list, and NONE should be sent
> >> to emacs-devel or bug-gnu-emacs.
> > 
> > So we have exactly what you describe, except that bug-gnu-emacs is now
> > the "separate bug-tracker list".
> > 
> 
> The donarmstrong.com  emacsbugs mailing list also still exists.
> It only seems to get bits and pieces though, I think maybe it's a loose
> end from early stages of Don's implementation?

That list's usage was replaced by bug-gnu-emacs.


Don Armstrong

-- 
I'd never hurt another living thing.
But if I did...
It would be you.
 -- Chris Bishop  http://www.chrisbishop.com/her/archives/her69.html

http://www.donarmstrong.com              http://rzlab.ucr.edu

Re: pretest, devel and bug lists

[Nick Roberts]

 > So we have exactly what you describe, except that bug-gnu-emacs is now
 > the "separate bug-tracker list".

In other words, not exactly what RMS describes.  I think it's overkill for
everything that gets sent to bug-gnu-emacs to have a bug number.  Some things
are fixed immediately and others turn out not to be bugs.  I think a
separate mailing list is a good idea.

Using emacs-pretest-bug for the bug tracker seems to make sense but then I
would suggest that M-x report-emacs-bug should then send to emacs-devel.

-- 
Nick                                           http://www.inet.net.nz/~nickrob

Re: pretest, devel and bug lists

[Don Armstrong]

On Thu, 29 May 2008, Nick Roberts wrote:

>  > So we have exactly what you describe, except that bug-gnu-emacs is now
>  > the "separate bug-tracker list".
> 
> In other words, not exactly what RMS describes.  I think it's overkill for
> everything that gets sent to bug-gnu-emacs to have a bug number.  Some things
> are fixed immediately and others turn out not to be bugs. 

All of these cases are cases where you should use a bug tracker; in
the case where they're fixed immediately or aren't bugs, you close
them and move on. They're tracked for you, so you can avoid losing an
important bug, and you know that every message has actually been dealt
with.


Don Armstrong

-- 
When bad men combine, the good must associate; else they will fall one
by one, an unpitied sacrifice in a contemptible struggle.
 -- Edmund Burke "Thoughts on the Cause of Present Discoontents"

http://www.donarmstrong.com              http://rzlab.ucr.edu

Re: pretest, devel and bug lists

[Richard M Stallman]

     > So we have exactly what you describe, except that bug-gnu-emacs is now
     > the "separate bug-tracker list".

    In other words, not exactly what RMS describes.

I think that is a misunderstanding.  There are two questions here!

1. Whether messages people send to bug-gnu-emacs should generate bug
tracker entries.  (And likewise for emacs-pretest-bug.)

I said nothing about that, but I think that both lists should do so.

2. Whether the bug tracker should send mail to bug-gnu-emacs.

That is the question I was talking about before.
I think it should NOT send any mail to bug-gnu-emacs, or
to emacs-pretest-bug.

    Using emacs-pretest-bug for the bug tracker

"Using" in this context confuses the two questions,
since it fails to distinguish the two kinds of use.
I urge you to avoid that wording.

Re: pretest, devel and bug lists

[Don Armstrong]

On Thu, 29 May 2008, Richard M Stallman wrote:
> I said nothing about that, but I think that both lists should do so.
> 
> 2. Whether the bug tracker should send mail to bug-gnu-emacs.
> 
> That is the question I was talking about before. I think it should
> NOT send any mail to bug-gnu-emacs, or to emacs-pretest-bug.

The way bug-gnu-emacs works now, all of the messages which haven't
come from the bug tracker get sent to the bug tracker first, which
assigns new messages a bug number and modifies the headers
accordingly. This first part can't really be changed and still have a
functional bug tracker and a bug-gnu-emacs which works as it did
previously.

The only extraneous messages that the bug tracker sends (or at least,
should be sending) to bug-gnu-emacs which wouldn't previously have
been seen on that list are the control modification transcripts. These
can be easily filtered out by end users or shuffled off to a separate
mailing list. I personally think they're informative, since they deal
with the disposition of bugs, but that's a decision that can be made
either way.


Don Armstrong

-- 
"There's no problem so large it can't be solved by killing the user
off, deleting their files, closing their account and reporting their
REAL earnings to the IRS."
 -- The B.O.F.H..

http://www.donarmstrong.com              http://rzlab.ucr.edu

Re: pretest, devel and bug lists

[Richard M Stallman]

    > 2. Whether the bug tracker should send mail to bug-gnu-emacs.
    > 
    > That is the question I was talking about before. I think it should
    > NOT send any mail to bug-gnu-emacs, or to emacs-pretest-bug.

    The way bug-gnu-emacs works now, all of the messages which haven't
    come from the bug tracker get sent to the bug tracker first, which
    assigns new messages a bug number and modifies the headers
    accordingly. This first part can't really be changed and still have a
    functional bug tracker and a bug-gnu-emacs which works as it did
    previously.

I don't understand what that means, but I have a feeling you are not
talking about the same issue.  I am NOT talking about messages that
are actually sent to the list by users.  I am talking about the canned
messages that originate from the bug tracker.

    The only extraneous messages that the bug tracker sends (or at least,
    should be sending) to bug-gnu-emacs which wouldn't previously have
    been seen on that list are the control modification transcripts. 

Perhaps those are the messages I am talking about.

What I am saying is that it SHOULD NOT send those messages to
emacs-devel or bug-gnu-emacs.  Never.  It should send them to a
separate list, which people can subscribe to if they are interested.

I am not talking about any other question.

Re: pretest, devel and bug lists

[Don Armstrong]

On Fri, 30 May 2008, Richard M Stallman wrote:
> I am NOT talking about messages that are actually sent to the list
> by users. I am talking about the canned messages that originate from
> the bug tracker.

There aren't any canned messages that are sent out that get to that
list; every single message that goes out corresponds to a single mail
sent to debbugs which is then sent to the mailing list. [The canned
ack messages go directly to the person who sent the message; they
never show up in the list.]

> Perhaps those are the messages I am talking about.

It'd help me understand what is being discussed if you'd provide an
example of the type of messages that you don't wish to see on that
list.


Don Armstrong

-- 
Cheop's Law: Nothing ever gets built on schedule or within budget.
 -- Robert Heinlein _Time Enough For Love_ p242

http://www.donarmstrong.com              http://rzlab.ucr.edu

Re: pretest, devel and bug lists

[Nick Roberts]

 > I think that is a misunderstanding.  There are two questions here!
 > 
 > 1. Whether messages people send to bug-gnu-emacs should generate bug
 > tracker entries.  (And likewise for emacs-pretest-bug.)
 > 
 > I said nothing about that, but I think that both lists should do so.
 > 
 > 2. Whether the bug tracker should send mail to bug-gnu-emacs.
 > 
 > That is the question I was talking about before.
 > I think it should NOT send any mail to bug-gnu-emacs, or
 > to emacs-pretest-bug.

What does that mean?  The thread generated by a bug report is presumably
a mixture of technical discussion and admin related to the tracker.  How
does someone following bug-gnu-emacs know that a bug report has been closed
if he only sees part of the thread.

 >     Using emacs-pretest-bug for the bug tracker
 > 
 > "Using" in this context confuses the two questions,
 > since it fails to distinguish the two kinds of use.

That's probably because I am confused.

 > I urge you to avoid that wording.

-- 
Nick                                           http://www.inet.net.nz/~nickrob

Re: pretest, devel and bug lists

[Stephen J. Turnbull]

Nick Roberts writes:
 > Richard Stallman writes:

 >  > I think that is a misunderstanding.  There are two questions here!
 >  > 
 >  > 1. Whether messages people send to bug-gnu-emacs should generate bug
 >  > tracker entries.  (And likewise for emacs-pretest-bug.)
 >  > 
 >  > I said nothing about that, but I think that both lists should do so.
 >  > 
 >  > 2. Whether the bug tracker should send mail to bug-gnu-emacs.
 >  > 
 >  > That is the question I was talking about before.
 >  > I think it should NOT send any mail to bug-gnu-emacs, or
 >  > to emacs-pretest-bug.
 > 
 > What does that mean?

AFAICS, that means that the workflow continues to be organized around
the lists, and that gateway to the bug tracker is used to ensure that
issues get recorded for review.

 > The thread generated by a bug report is presumably a mixture of
 > technical discussion and admin related to the tracker.  How does
 > someone following bug-gnu-emacs know that a bug report has been
 > closed if he only sees part of the thread.

He looks at the bug tracker web interface, or joins the nosy list for
the bug.  (If that's possible, I know debbugs sends mail to the
originator and to the maintainer, presumably there's a way to add
yourself to the list of interested parties.)

This is basically the model used by the Python developers, with the
improvement that the tracker generates a weekly report containing a
summary of activity (total issues, total open, new this week, active
this week, closed this week, etc), a list of new issues with their
titles, and list of closed issues with their titles.  It sends this
report to the mailing list.  This minimizes the intrusion of
administrative detail on the list, while making people aware of the
tracker and its activity, and prompting developers to maintain the
issues that they are responsible for.

So your presumption is wrong, in that model: there is a thread of
substantive discussion on the mailing list, and there is a thread of
administration on the tracker.  The tracker is also responsible for
keeping important data such as test cases and proposed patches, which
are relatively rarely sent to the list.

 >  >     Using emacs-pretest-bug for the bug tracker
 >  > 
 >  > "Using" in this context confuses the two questions,
 >  > since it fails to distinguish the two kinds of use.
 > 
 > That's probably because I am confused.

Well, there certainly are tracker-centric workflows.  Emacs has never
been one, though, and given all the changes that are happening now I
find Richard's gradualist approach to introduction of the tracker to
be both natural and plausible, even though it probably does postpone
taking full advantage of important tracker features indefinitely.

Re: pretest, devel and bug lists

[Richard M Stallman]

    What does that mean?  The thread generated by a bug report is presumably
    a mixture of technical discussion and admin related to the tracker. 

That is exactly what I do not want.  I do not want to get all the
canned administrative messages, and I don't think anyone else does
either.

The person who reported a bug, and the persons who send commands to
the bug tracker about it, probably do want to get the administrative
messages about that bug.  But bug-gnu-emacs should not get them.

									 How
    does someone following bug-gnu-emacs know that a bug report has been closed
    if he only sees part of the thread.

People usually send a message saying "Yes, that change fixed it."  You
would get the information that way, and you would not need the canned
message.

If he wants to see those messages, he can subscribe to a new list
where they are sent.  That is easy.

Re: pretest, devel and bug lists

[Don Armstrong]

On Tue, 27 May 2008, Drew Adams wrote:
> I think I disagree. I like everything about Emacs bugs to be sent to
> the bugs list. But I too am less sure about this.

These can be filtered out on an individual basis by sending messages
with the 'X-Emacs-PR-Message: transcript' header to /dev/null using
procmail or similar. [In fact, every message that it sent out from the
bts uses these headers so filtering can be done.]


Don Armstrong

-- 
Debian's not really about the users or the software at all. It's a
large flame-generating engine that the cabal uses to heat their coffee
 -- Andrew Suffield (#debian-devel Fri, 14 Feb 2003 14:34 -0500)

http://www.donarmstrong.com              http://rzlab.ucr.edu

Re: pretest, devel and bug lists

[Chong Yidong]

Glenn Morris <rgm@gnu.org> writes:

> If things sent to emacs-pretest-bug are being fed into a bug tracker
> and sent to bug-gnu-emacs, then I would say there is no need for
> copies to _also_ be sent to emacs-devel.

I don't think this behavior is that bad, but it does contain one
annoying bug.  AFAICT, follow-up messages sent to the bug tracker aren't
forwarded to emacs-devel.  However, if I send a message to emacs-devel
and CC the bug tracker (so that the message is stored), duplicate
messages are posted on emacs-devel: one from me, and one from the bug
tracker.  Is there a way to change the bug tracker to fix this behavior?

> I also don't see the need for tracker control messages to be sent to
> bug-gnu-emacs, but I'm less sure about that.

I think this is harmless.

> It would also be nice if there was a statement as to whether this
> tracker is now offical and we should all start learning how to use it,
> or if it is still being tested.

I'm pretty happy with the bug tracker, apart from the funkiness
described above.  Maybe we should move it to Savannah and begin using it
officially.  Stefan, WDYT?

Re: pretest, devel and bug lists

[Don Armstrong]

On Tue, 27 May 2008, Chong Yidong wrote:
> Glenn Morris <rgm@gnu.org> writes:
> > If things sent to emacs-pretest-bug are being fed into a bug tracker
> > and sent to bug-gnu-emacs, then I would say there is no need for
> > copies to _also_ be sent to emacs-devel.

Uh... they aren't? [At least, they aren't being sent by the bug
tracker.]

> I don't think this behavior is that bad, but it does contain one
> annoying bug. AFAICT, follow-up messages sent to the bug tracker
> aren't forwarded to emacs-devel. However, if I send a message to
> emacs-devel and CC the bug tracker (so that the message is stored),
> duplicate messages are posted on emacs-devel: one from me, and one
> from the bug tracker. Is there a way to change the bug tracker to
> fix this behavior?

The bug tracker does not currently send any messages on to
emacs-devel, so this list only gets a single copy. [You *can* ask it
to send mail to -devel, but it does not do so by default.]

> > I also don't see the need for tracker control messages to be sent
> > to bug-gnu-emacs, but I'm less sure about that.
> 
> I think this is harmless.

They should go to the list so that you know what is being done with
the bugs and can contravene if necessary.

> > It would also be nice if there was a statement as to whether this
> > tracker is now offical and we should all start learning how to use
> > it, or if it is still being tested.
> 
> I'm pretty happy with the bug tracker, apart from the funkiness
> described above. Maybe we should move it to Savannah and begin using
> it officially. Stefan, WDYT?

Moving it to whatever machine is decided will basically need
coordination between me and whomever is administering that machine
along with the go-ahead to make it official. [Ideally I'd have a login
too, along with a debbugs user, but we can work that out.]


Don Armstrong

-- 
Three little words. (In decending order of importance.)
I
love
you
 -- hugh macleod http://www.gapingvoid.com/graphics/batch35.php

http://www.donarmstrong.com              http://rzlab.ucr.edu

Re: pretest, devel and bug lists

[Stefan Monnier]

> If things sent to emacs-pretest-bug are being fed into a bug tracker
> and sent to bug-gnu-emacs, then I would say there is no need for
> copies to _also_ be sent to emacs-devel.

AFAIK messages sent to bug-gnu-emacs pass through the bug-tracker, but
not messages sent to emacs-pretest-bug (which are redirected to
emacs-devel instead).

I think this should be changed so that messages sent to
emacs-pretest-bug do not go to emacs-devel any more but go to the
bug-tracker instead.

> I also don't see the need for tracker control messages to be sent to
> bug-gnu-emacs, but I'm less sure about that.

Yes, I'm also ambivalent about it.  I think I'd be happy to get rid of them.

> It would also be nice if there was a statement as to whether this
> tracker is now offical and we should all start learning how to use it,
> or if it is still being tested.

It is official in the sense that you should all learn to use it.


        Stefan

Re: pretest, devel and bug lists

[Karl Fogel]

Stefan Monnier <monnier@iro.umontreal.ca> writes:
>> If things sent to emacs-pretest-bug are being fed into a bug tracker
>> and sent to bug-gnu-emacs, then I would say there is no need for
>> copies to _also_ be sent to emacs-devel.
>
> AFAIK messages sent to bug-gnu-emacs pass through the bug-tracker, but
> not messages sent to emacs-pretest-bug (which are redirected to
> emacs-devel instead).

One technique is for automated mails (e.g. those emitted by a bug
tracker, say, or by a commit hook) to go to their own lists
(e.g. bug-gnu-emacs@, emacs-commit@), but for any *replies* to such
mails to be directed to the main development list, emacs-devel@.

That is, the automated systems set the "Reply-to:" header to
emacs-devel@, so that any followup discussion to a bug report or a
commit happens on the development list, where it belongs.  At the same
time, the development list not distracted with those reports and commits
that never spark a thread (while those who want to can subscribe to the
appropriate lists, to see and possibly react to the automated mails).

Whether this is appropriate for a given automated mail source depends on
the source.  Certainly, there may be some sources that should be sending
directly to emacs-devel@.  I just offer this technique as something to
consider when all-or-nothing answers don't seem quite right.  In my
experience, doing it at least for bug-tracker emails and for commit
mails works very well; YMMV.

(Finally, for those who believe Chip Rosenthal was right when he wrote
http://www.unicom.com/pw/reply-to-harmful.html, don't worry: this
technique doesn't contradict his recommendations.  The sender is always
free to set Reply-to however it wants, and in this case the sender is
the automated system.)

-Karl

Stefan Monnier <monnier@iro.umontreal.ca> writes:
>> If things sent to emacs-pretest-bug are being fed into a bug tracker
>> and sent to bug-gnu-emacs, then I would say there is no need for
>> copies to _also_ be sent to emacs-devel.
>
> AFAIK messages sent to bug-gnu-emacs pass through the bug-tracker, but
> not messages sent to emacs-pretest-bug (which are redirected to
> emacs-devel instead).
>
> I think this should be changed so that messages sent to
> emacs-pretest-bug do not go to emacs-devel any more but go to the
> bug-tracker instead.
>
>> I also don't see the need for tracker control messages to be sent to
>> bug-gnu-emacs, but I'm less sure about that.
>
> Yes, I'm also ambivalent about it.  I think I'd be happy to get rid of them.
>
>> It would also be nice if there was a statement as to whether this
>> tracker is now offical and we should all start learning how to use it,
>> or if it is still being tested.
>
> It is official in the sense that you should all learn to use it.
>
>
>         Stefan

Re: pretest, devel and bug lists

[Glenn Morris]

Stefan Monnier wrote:

>> If things sent to emacs-pretest-bug are being fed into a bug tracker
>> and sent to bug-gnu-emacs, then I would say there is no need for
>> copies to _also_ be sent to emacs-devel.
>
> AFAIK messages sent to bug-gnu-emacs pass through the bug-tracker, but
> not messages sent to emacs-pretest-bug (which are redirected to
> emacs-devel instead).

Take bug#327

http://lists.gnu.org/archive/html/bug-gnu-emacs/2008-05/msg00240.html
http://lists.gnu.org/archive/html/emacs-devel/2008-05/msg01552.html

AFAICS this was originally sent to emacs-pretest-bug@gnu.org, yet it
has a bug number, and exists both on emacs-devel and bug-gnu-emacs.

I'm finding it rather confusing having things on two lists.

> I think this should be changed so that messages sent to
> emacs-pretest-bug do not go to emacs-devel any more but go to the
> bug-tracker instead.

+1.

Re: pretest, devel and bug lists

[Stefan Monnier]

>>> If things sent to emacs-pretest-bug are being fed into a bug tracker
>>> and sent to bug-gnu-emacs, then I would say there is no need for
>>> copies to _also_ be sent to emacs-devel.
>> 
>> AFAIK messages sent to bug-gnu-emacs pass through the bug-tracker, but
>> not messages sent to emacs-pretest-bug (which are redirected to
>> emacs-devel instead).

> Take bug#327

That was a bug in my mail filter, sorry.  It's been fixed in the
mean time.


        Stefan

Re: pretest, devel and bug lists

[Stephen Berman]

On Tue, 27 May 2008 20:49:07 -0400 Stefan Monnier <monnier@iro.umontreal.ca> wrote:

>> If things sent to emacs-pretest-bug are being fed into a bug tracker
>> and sent to bug-gnu-emacs, then I would say there is no need for
>> copies to _also_ be sent to emacs-devel.
>
> AFAIK messages sent to bug-gnu-emacs pass through the bug-tracker, but
> not messages sent to emacs-pretest-bug (which are redirected to
> emacs-devel instead).

Since (and because) this redirection was instituted I have usually sent
bug reports directly to emacs-devel, except where I thought it was more
helpful or convenient to use report-emacs-bug.

> I think this should be changed so that messages sent to
> emacs-pretest-bug do not go to emacs-devel any more but go to the
> bug-tracker instead.

If this is done, should *no* bug reports be sent any longer to
emacs-devel (and if they were, would that lower their chances of getting
attended to)?  It might be helpful for a maintainer to issue a policy
statement (preferably clearly recognizable as such) about this.

Steve Berman

Re: pretest, devel and bug lists

[Nick Roberts]

Stephen Berman writes:
 > On Tue, 27 May 2008 20:49:07 -0400 Stefan Monnier <monnier@iro.umontreal.ca> wrote:
 > 
 > >> If things sent to emacs-pretest-bug are being fed into a bug tracker
 > >> and sent to bug-gnu-emacs, then I would say there is no need for
 > >> copies to _also_ be sent to emacs-devel.
 > >
 > > AFAIK messages sent to bug-gnu-emacs pass through the bug-tracker, but
 > > not messages sent to emacs-pretest-bug (which are redirected to
 > > emacs-devel instead).
 > 
 > Since (and because) this redirection was instituted I have usually sent
 > bug reports directly to emacs-devel, except where I thought it was more
 > helpful or convenient to use report-emacs-bug.
 > 
 > > I think this should be changed so that messages sent to
 > > emacs-pretest-bug do not go to emacs-devel any more but go to the
 > > bug-tracker instead.
 > 
 > If this is done, should *no* bug reports be sent any longer to
 > emacs-devel (and if they were, would that lower their chances of getting
 > attended to)?  It might be helpful for a maintainer to issue a policy
 > statement (preferably clearly recognizable as such) about this.

Bug reports for unreleased versions of Emacs should go to emacs-devel and those
for released versions to bug-gnu-emacs.  This happens automatically if you use
M-x report-emacs-bug (since emacs-pretest-bug is an alias for emacs-devel).

If all messages sent to bug-gnu-emacs pass through the bug-tracker I think it
will accumulate a lot of crud.  I would suggest bug reports go just to the
mailing list first then if they are good reports and not immediately fixable,
the OP is encouraged by the maintainer (or others) to post the report to the
tracker.  Of course, this process could be bypassed but it might increase the
signal to noise ratio. 

-- 
Nick                                           http://www.inet.net.nz/~nickrob

Re: pretest, devel and bug lists

[Stephen Berman]

On Wed, 28 May 2008 21:45:56 +1200 Nick Roberts <nickrob@snap.net.nz> wrote:

> Stephen Berman writes:
>  > On Tue, 27 May 2008 20:49:07 -0400 Stefan Monnier <monnier@iro.umontreal.ca> wrote:
>  > 
>  > >> If things sent to emacs-pretest-bug are being fed into a bug tracker
>  > >> and sent to bug-gnu-emacs, then I would say there is no need for
>  > >> copies to _also_ be sent to emacs-devel.
>  > >
>  > > AFAIK messages sent to bug-gnu-emacs pass through the bug-tracker, but
>  > > not messages sent to emacs-pretest-bug (which are redirected to
>  > > emacs-devel instead).
>  > 
>  > Since (and because) this redirection was instituted I have usually sent
>  > bug reports directly to emacs-devel, except where I thought it was more
>  > helpful or convenient to use report-emacs-bug.
>  > 
>  > > I think this should be changed so that messages sent to
>  > > emacs-pretest-bug do not go to emacs-devel any more but go to the
>  > > bug-tracker instead.
>  > 
>  > If this is done, should *no* bug reports be sent any longer to
>  > emacs-devel (and if they were, would that lower their chances of getting
>  > attended to)?  It might be helpful for a maintainer to issue a policy
>  > statement (preferably clearly recognizable as such) about this.
>
> Bug reports for unreleased versions of Emacs should go to emacs-devel
> and those for released versions to bug-gnu-emacs.  This happens
> automatically if you use M-x report-emacs-bug (since emacs-pretest-bug
> is an alias for emacs-devel).

This is the status quo, but I understand Stefan to be favoring changing
this, and that is why I asked for a policy statement.

Steve Berman

Re: pretest, devel and bug lists

[Stefan Monnier]

> If all messages sent to bug-gnu-emacs pass through the bug-tracker I think it

They do.


        Stefan

Re: pretest, devel and bug lists

[David De La Harpe Golden]

Nick Roberts wrote:

> If all messages sent to bug-gnu-emacs pass through the bug-tracker I think it
> will accumulate a lot of crud.  I would suggest bug reports go just to the
> mailing list first then if they are good reports and not immediately fixable,
> the OP is encouraged by the maintainer (or others) to post the report
> to the tracker

I'd say that's the wrong way round.  I'd be inclined to say all bugs,
devel or release, belong in the tracker _first_:

Bug trackers are _for_ sorting out useful nuggets from crud,
much more so than mailing lists.  Bug trackers allow easy merging of
redundant bug reports and killing of noise bug reports, unlike typical
discussion-oriented mailing lists and mailing list archives.

It also tends to be possible to do much more focussed searches, so a
"please try to check the bug tracker database before reporting" is a
more reasonable request than "please try to wade through the mailing
list archives before reporting".   [In the debian bug tracker case, I
think a bug database index can be retrieved by email, for those who
dislike (or can't easily use) the web search frontend, though it would
be nicer if it supported by email all the search functionality available
via the web interface]

Re devel and release in the tracker:

A tag/version could be presumably assigned automagically by the bug
tracker based on submission email address in the emacs case (and/or
included as a pseudoheader in report-bug, though that is perhaps prone
to users editing it away).

Once such a tag/version exists, the bug tracker could/should be able to
hit emacs-devel with only the emacs-pretest-bug reports if that's desired.

Not too familiar with debian bug tracking system's capabilities
specifically, but given the "trick" of using bug-gnu-emacs as the
"maintainer", maybe auto*-setting the bug "owner" to emacs-devel for the
pretest-bug case would be sufficient?

Though individual emacs developers might want to be assigned as
the real bug "owner" on a per-bug basis while the discussion continues
to be cc'd to the list(s), so maybe auto*-adding emacs-devel to the bug
cc list (X-Debbugs-CC) would be better.

(*manually might be preferable, e.g. not sure about hitting emacs-devel
with reports of typos just because they're typos found in the devel
tree, maybe only "interesting" bugs need escalation to mailing list)

Re: pretest, devel and bug lists

[Stefan Monnier]

> If this is done, should *no* bug reports be sent any longer to
> emacs-devel (and if they were, would that lower their chances of getting
> attended to)?  It might be helpful for a maintainer to issue a policy
> statement (preferably clearly recognizable as such) about this.

Bug reports linked to transient problems (e.g, bootstrap problems)
should not go to the bug tracker.  Any other bug report which you want
people to attend to eventually should go to the bug tracker (where it's
more difficult to let them disappear).


        Stefan

Re: pretest, devel and bug lists

[Stephen Berman]

On Wed, 28 May 2008 15:48:58 -0400 Stefan Monnier <monnier@iro.umontreal.ca> wrote:

>>> The way things are setup now:
>>> 
>>> - all mail to bug-gnu-emacs first goes through the bug-tracker.
>>> I.e. all mail received by bug-gnu-emacs recipients comes from the
>>> bug-tracker.
>>> - emacs-pretest-bug and emacs-devel are competely separate from the
>>> bug-tracker.
>
>> Does this mean that bug reports sent to emacs-pretest-bug will not
>> get assigned a bug number anymore?
>
> They never have, AFAIK (except for a temporarily bug in my mail filter
> and except for those that someone manually forwarded to the bug-tracker).

But you *are* in favor of using the bug tracker for bug reports about
development Emacs, i.e., what emacs-pretest-bug is now used for, right?
At least that is how I understand your reply (>) to my post (>>) in this
thread:

>> If this is done, should *no* bug reports be sent any longer to
>> emacs-devel (and if they were, would that lower their chances of getting
>> attended to)?  It might be helpful for a maintainer to issue a policy
>> statement (preferably clearly recognizable as such) about this.
>
> Bug reports linked to transient problems (e.g, bootstrap problems)
> should not go to the bug tracker.  Any other bug report which you want
> people to attend to eventually should go to the bug tracker (where it's
> more difficult to let them disappear).
>
>
>         Stefan

I support at least routing posts to emacs-pretest-bug to the bug
tracker, and would also like to see a full-featured Emacs interface to
it.

Steve Berman

Re: pretest, devel and bug lists

[Reiner Steib]

On Wed, May 28 2008, Stefan Monnier wrote:

>> It would also be nice if there was a statement as to whether this
>> tracker is now offical and we should all start learning how to use it,
>> or if it is still being tested.
>
> It is official in the sense that you should all learn to use it.

IMHO there must be a decent Emacs interface for the bug tracker.
Several weeks ago I started a thread about the bug tracker [1] because
I wanted (and I still want) to put bug reports about Gnus into the
tracker.  But there isn't such an interface yet, though probably
`debian-bug.el' could (should) contain most of the required
features. [2]  However, I didn't have time to learn enough about the
BTS nor `debian-bug.el' to forward bug-report mails (e.g. from
ding@gnus) to the tracker in a convenient way.

Also, the official tracker should be somewhere on gnu.org and the Web
pages have to be adopted: <http://emacsbugs.donarmstrong.com/> is the
Debian page.  An Emacs user/developer has to figure out what has to be
ignore since it is Debian-only.  (Examples: "potato, woody, ...",
Links to Debian stuff, "NMU" [3]
<http://emacsbugs.donarmstrong.com/cgi-bin/pkgreport.cgi?which=pkg&data=emacs&archive=no&version=&dist=unstable#_3_2_5>, ...).

Bye, Reiner.

[1]
,----[ http://thread.gmane.org/v9abklqfyj.fsf@marauder.physik.uni-ulm.de ]
| From: Reiner Steib
| Subject: Status of the bug tracker testbed
| To: emacs-devel@gnu.org
| Date: Wed, 26 Mar 2008 23:27:00 +0100
| Message-ID: <v9abklqfyj.fsf@marauder.physik.uni-ulm.de>
`----

,----[ http://thread.gmane.org/gmane.emacs.devel/93575/focus=94279 ]
| From: Reiner Steib
| Subject: debian-bug.el in Emacs? (was: Status of the bug tracker testbed)
| To: psg@debian.org (Peter S. Galbraith)
| Cc: emacs-devel@gnu.org
| Date: Thu, 03 Apr 2008 21:40:07 +0200
| Message-ID: <v97ifeafs8.fsf_-_@marauder.physik.uni-ulm.de>
`----

[3] Accidentally, I (not using Debian) read about "NMU" a few days
    ago: It is "Non Maintainer Upload" in Debian, but I don't know
    what this is supposed to mean in the context of Emacs development.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/