[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: closing upstream bugs with debian/changelog

On Sat, Apr 03, 2004 at 09:11:54PM +1000, Herbert Xu wrote:
> However, it is my firm belief that all bug submitters should respect
> your work by actually verifying for themselves that the bug is indeed
> fixed.

I agree, they *should* do that.  But I can't count on it.  And by
including the *specific reason* the bug is closed in my changelog entry,
it easier for anyone who is appropriately equipped can do this task in
lieu of the original submitter.


    - XFree86 X server now supports ATI Radeon Mobility (Closes: #196810)
    - XFree86 X server now supports Intel i845G (Closes: #184322)
    - XFree86 X server now supports Intel i865G (Closes: #221686)
    - XFree86 X server now supports SiS 650, 651, and 740 (Closes: #183619)
    - XTerm allows window title reporting, disabled by default, to be enabled
      at the user's option via the allowWindowOps X resource (Closes: #200857)

The first four could be confirmed by anyone with the appropriate
hardware, and the latter should be confirmable by just about any Debian

> After all, you might have made a simple typo/think in writing or
> verifying the fix to the bug.  It'd be a pity if it weren't picked up
> just because the submitter was complacent.

I agree, but this observation doesn't support your approach to resolving
upstream bug reports in the Debian BTS any more than mine does.
Actually, it may help mine more, because my approach communicates *why*
I think the bug is resolved.  It is easier for people to impeach my
judgement, instead of encouraging people to treat me as a guru.  "Oh,
the bug was fixed upstream?  How?  Hmm, I can't see the reason.  The
maintainer must know.  No need for me to bother verifying the fix."

As far as correcting past changelog entries goes, an examination of the
SVN changelog for the XFree86 package changelog reveals multiple
occurences of this just within the past week.


svn log -r 1209 -v svn://necrotic.deadbeast.net/xfree86/trunk/debian/changelog
svn log -r 1208 -v svn://necrotic.deadbeast.net/xfree86/trunk/debian/changelog
svn log -r 1206 -v svn://necrotic.deadbeast.net/xfree86/trunk/debian/changelog
svn log -r 1198 -v svn://necrotic.deadbeast.net/xfree86/trunk/debian/changelog

> Besides, you can also provide the explanation by posting them to the
> BTS directly.  Granted this is not as convenient as slipping them into
> the Debian changelog, but alas we don't put things in the Debian
> changelog just because it is convinient.  Otherwise we'd be closing
> bugs fixed in past Debian versions (or bugs that aren't bugs at all)
> using the Debian changelog.

The reason we don't do that is because those don't represent changes in
the package payload as of that version.

I don't have a problem with people adding "(Closes: #XXXXX)" notation to
older changelog entries when the corresponding fixes actually took
place, and I have done so myself.  Of course, I still have to close the
bug manually, but I don't find that onerous.  Having the closure
documented in the changelog at the proper time and for the proper reason
is, I find, quite useful when I get questions about when exactly some bug

> So I personally find this argument to be very weak in terms of
> motivating me in putting explanations next to bug closures caused
> by upstream changes.

The problem is that you're not providing any explanation at all for
these, anywhere, except for "new upstream release".

> > "Herbert Xu would never do such a thing" is not a valid defense.  Our
> > changelogs need to be accessible to those who are not aware of your
> > reputation within the Debian Project.
> Good to see the good old Branden Robinson again.  It's rather
> disconcerting to not see personal attacks from you in mailing lists.
> It makes want to double-check the signature to make sure that
> it really is you.

Sadly, I do not have to do a double-take to understand that you have
misinterpreted a compliment as a personal attack.

I guess people see what they want to see.

G. Branden Robinson                |
Debian GNU/Linux                   |    Yeah, that's what Jesus would do.
branden@debian.org                 |    Jesus would bomb Afghanistan. Yeah.
http://people.debian.org/~branden/ |

Attachment: signature.asc
Description: Digital signature

Reply to: