Re: Changelogs (Re: Bug#193497: marked as done (svtools: svsetup uses bashism echo -e))
On Mon, 16 Jun 2003 08:36:20 +1000, Herbert Xu [EMAIL PROTECTED] said: Manoj Srivastava [EMAIL PROTECTED] wrote: Even if the bug is for upstream fixing a typo in a comment? :) If it caused a Debian bug to be closed, that is a significant change in status for the Debian package (it may not be for the upstream software being packaged, but it is for my package). What if the bug was reported after the new Debian package was uploaded? Why does it suddenly stop being a significant change? Good point. Shall we mandate that all bug closures be adequately documented in the ChangeLog? I would be quite happy with that. Indeed, the rationale would be that closing bugs is an important event for the Package, and needs be thus documented. I thank you for pointing out this defect in our ChangeLog standards. manoj -- If all the world is a stage, where is the audience sitting? George Carlin Manoj Srivastava [EMAIL PROTECTED] http://www.debian.org/%7Esrivasta/ 1024R/C7261095 print CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E 1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C
Re: Changelogs (Re: Bug#193497: marked as done (svtools: svsetup uses bashism echo -e))
On Sun, Jun 15, 2003 at 11:42:44PM -0500, Manoj Srivastava wrote: On Mon, 16 Jun 2003 08:36:20 +1000, Herbert Xu [EMAIL PROTECTED] said: Manoj Srivastava [EMAIL PROTECTED] wrote: If it caused a Debian bug to be closed, that is a significant change in status for the Debian package (it may not be for the upstream software being packaged, but it is for my package). What if the bug was reported after the new Debian package was uploaded? Why does it suddenly stop being a significant change? Good point. Shall we mandate that all bug closures be adequately documented in the ChangeLog? I would be quite happy with that. Ai. Er, I hope you're not planning to encourage people to upload new versions of packages just to add bug numbers to the changelog? Because that would be most inefficient and wrong. My answer to Herbert would be it would be nice, but unfortunately time travel is not yet among our capabilities. -- Colin Watson [EMAIL PROTECTED]
Re: Changelogs (Re: Bug#193497: marked as done (svtools: svsetup uses bashism echo -e))
Colin Watson [EMAIL PROTECTED] wrote: On Sun, Jun 15, 2003 at 11:42:44PM -0500, Manoj Srivastava wrote: What if the bug was reported after the new Debian package was uploaded? Why does it suddenly stop being a significant change? This is meant to be a rhetorical question. My answer to Herbert would be it would be nice, but unfortunately time travel is not yet among our capabilities. Which just proves that listing things in the changelog on the basis of bug reports is meaningless. Listing things that actually have changed in Debian is what it was meant for in the first place. -- Debian GNU/Linux 3.0 is out! ( http://www.debian.org/ ) Email: Herbert Xu ~{PmVHI~} [EMAIL PROTECTED] Home Page: http://gondor.apana.org.au/~herbert/ PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
Re: Changelogs (Re: Bug#193497: marked as done (svtools: svsetup uses bashism echo -e))
On Mon, Jun 16, 2003 at 07:51:07PM +1000, Herbert Xu wrote: Which just proves that listing things in the changelog on the basis of bug reports is meaningless. No, it only shows that it is not possible under some circumstances: specifically, when there was no bug report at the time. To say that we should not document any bugs because of this is analogous to saying that since we cannot fix all of the bugs, we should not fix any of them. -- - mdz
Re: Changelogs (Re: Bug#193497: marked as done (svtools: svsetup uses bashism echo -e))
On Mon, 16 Jun 2003 19:51:07 +1000, Herbert Xu [EMAIL PROTECTED] said: Colin Watson [EMAIL PROTECTED] wrote: On Sun, Jun 15, 2003 at 11:42:44PM -0500, Manoj Srivastava wrote: What if the bug was reported after the new Debian package was uploaded? Why does it suddenly stop being a significant change? This is meant to be a rhetorical question. It turned out to be closer to the mark than you suspected ;-) My answer to Herbert would be it would be nice, but unfortunately time travel is not yet among our capabilities. Which just proves that listing things in the changelog on the basis of bug reports is meaningless. Listing things that actually have changed in Debian is what it was meant for in the first place. I disagree. Changelogs for debian packages should document all changes that happened to the package -- and this include bugs that were closed, and the reaons thereof. No inconsistency. Thank you for helping me make my argument. manoj -- In any problem, if you find yourself doing an infinite amount of work, the answer may be obtained by inspection. Manoj Srivastava [EMAIL PROTECTED] http://www.debian.org/%7Esrivasta/ 1024R/C7261095 print CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E 1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C
Re: Changelogs (Re: Bug#193497: marked as done (svtools: svsetup uses bashism echo -e))
On Mon, 16 Jun 2003 09:36:24 +0100, Colin Watson [EMAIL PROTECTED] said: On Sun, Jun 15, 2003 at 11:42:44PM -0500, Manoj Srivastava wrote: On Mon, 16 Jun 2003 08:36:20 +1000, Herbert Xu [EMAIL PROTECTED] said: Manoj Srivastava [EMAIL PROTECTED] wrote: If it caused a Debian bug to be closed, that is a significant change in status for the Debian package (it may not be for the upstream software being packaged, but it is for my package). What if the bug was reported after the new Debian package was uploaded? Why does it suddenly stop being a significant change? Good point. Shall we mandate that all bug closures be adequately documented in the ChangeLog? I would be quite happy with that. Ai. Er, I hope you're not planning to encourage people to upload new versions of packages just to add bug numbers to the changelog? Because that would be most inefficient and wrong. *Sigh*. Has common sense totally escaped the world? I never indicated that one upload every other minute or whenever something is added to the changelog. I do think bug closures be documented in the ChangeLog (I shall attempt to do so from now on for every real bug that is closed for my packages). I shall not upload for every item in my changelog. And note this does not involve time travel; my changelog would document the bug was closed, and explain why: the fact that the change was made in the past is OK. manoj -- I owe the public nothing. J.P. Morgan Manoj Srivastava [EMAIL PROTECTED] http://www.debian.org/%7Esrivasta/ 1024R/C7261095 print CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E 1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C
Re: Changelogs (Re: Bug#193497: marked as done (svtools: svsetup uses bashism echo -e))
On Mon, Jun 16, 2003 at 04:04:55PM -0500, Manoj Srivastava wrote: On Mon, 16 Jun 2003 09:36:24 +0100, Colin Watson [EMAIL PROTECTED] said: On Sun, Jun 15, 2003 at 11:42:44PM -0500, Manoj Srivastava wrote: Good point. Shall we mandate that all bug closures be adequately documented in the ChangeLog? I would be quite happy with that. Ai. Er, I hope you're not planning to encourage people to upload new versions of packages just to add bug numbers to the changelog? Because that would be most inefficient and wrong. *Sigh*. Has common sense totally escaped the world? I never indicated that one upload every other minute or whenever something is added to the changelog. I do think bug closures be documented in the ChangeLog (I shall attempt to do so from now on for every real bug that is closed for my packages). I shall not upload for every item in my changelog. And note this does not involve time travel; my changelog would document the bug was closed, and explain why: the fact that the change was made in the past is OK. At least don't use the Closes: syntax to do so, please. That will be a very good way to confuse attempts to track version information ... (By now I can't tell whether this is regarded as common sense or not, sorry.) Cheers, -- Colin Watson [EMAIL PROTECTED]
Re: Changelogs (Re: Bug#193497: marked as done (svtools: svsetup uses bashism echo -e))
Manoj Srivastava wrote: Good point. Shall we mandate that all bug closures be adequately documented in the ChangeLog? I would be quite happy with that. I do think bug closures be documented in the ChangeLog (I shall attempt to do so from now on for every real bug that is closed for my packages). I shall not upload for every item in my changelog. Like so? debhelper (4.1.48) unstable; urgency=low * Not adding ld.so.conf parsing code for libraries, if your library is not in a usual path, use the standard -X option to dh_makeshlibs to skip it. Closes: #122174 * No, I don't think that adding rm -rf /tmp to postinsts is a good idea. Closed: #201012 * Reasigned bugs #133949 and #123043 to dh-make, since dh_make is not in debhelper. When will people learn? * debhelper won't include a dh_installrootkit until I see more demand for such a program. Closes: #37337 * I fixed the debhelper overwrites libc bug back in 1999, why am I still getting reports about it? Sheesh. Closed: #393933, #209202, #384821 * Yes, debhelper and debconf do indeed have recursive build-depends. Deal. Closed: #197602 * In the last release I made a typo, and accidentially closed bug #196343, not #196344. I've reopened the former bug. Closes: #196344 * Trimmed the last 500k of the changelog; it was 90% of the entire package size since it recorded every dismissed feature request and dh-make bug for the past 6 years. * Fixed a typo in dh_python. Closes: #197679 -- Joey Hess [EMAIL PROTECTED] Mon, 16 Jun 2003 16:58:27 -0400 -- see shy jo pgpw7OVNsQEkU.pgp Description: PGP signature
Changelogs (Re: Bug#193497: marked as done (svtools: svsetup uses bashism echo -e))
On Tue, Jun 03, 2003 at 07:42:05AM +1000, Herbert Xu wrote: Matt Zimmerman [EMAIL PROTECTED] wrote: Yes. There is no record in the package that a bug report in the Debian BTS was closed in this version. What about other users who experienced the same Why should that be in the package? What if you didn't know at the time that the bug was fixed in this version? You obviously can't document what you don't know. This discussion was about closing bugs in the changelog (which obviously implies that the maintainer knows that they are closed) without explaining why. bug? Is it so hard to explain what bug was fixed, and if possible, how it was fixed? Remember that I'm talking about closing a bug in the BTS by hand. I'm talking about changelogs. Would you upload a new kernel-source saying Closes: #242141 rather than apply patch from 2.6.48 to fix root security hole in nanosleep()? If this is fixed by upstream, yes. I would simply say * New upstream release (closes: #xxx, #yyy, #zzz). When I open up a Debian changelog, I expect to see what the Debian developer has done to it. I don't mind seeing changelog entries explaining upstream changes as long as they're clearly marked. And I would be most concerned if people start putting them in without marking them as upstream. When I open a Debian changelog, I expect to see changes which are pertinent to Debian development. This obviously includes changes which affect the status of Debian bug reports. As for attribution, I do not think this is a problem in practice, but this is the format I use: foo (2.1.1-1) unstable; urgency=high * New upstream release - use snprintf rather than sprintf to prevent a buffer overrun (Closes: #528432) - do some other things which relate to how this package is used in Debian * Build-Depends on more recent libbar * Don't forget to install the documentation (Closes: #879242) It is perfectly clear what changes are made by whom, and how these changes affect Debian bug reports. -- - mdz
Re: Changelogs (Re: Bug#193497: marked as done (svtools: svsetup uses bashism echo -e))
mdz writes: When I open a Debian changelog, I expect to see changes which are pertinent to Debian development. This obviously includes changes which affect the status of Debian bug reports. From Debian-Policy 3.5.10.0: Debian Policy Manual - Source packages (from old Packaging Manual) (7/11) C.2.3 debian/changelog This file records the changes to the Debian-specific parts of the package [72]. This would seem to preclude upstream bug-fixes. -- John Hasler [EMAIL PROTECTED] Dancing Horse Hill Elmwood, Wisconsin
Re: Changelogs (Re: Bug#193497: marked as done (svtools: svsetup uses bashism echo -e))
On Mon, Jun 02, 2003 at 07:06:12PM -0500, John Hasler wrote: mdz writes: When I open a Debian changelog, I expect to see changes which are pertinent to Debian development. This obviously includes changes which affect the status of Debian bug reports. From Debian-Policy 3.5.10.0: Debian Policy Manual - Source packages (from old Packaging Manual) (7/11) C.2.3 debian/changelog This file records the changes to the Debian-specific parts of the package [72]. I disagree with that sentence (note that it is not really part of the policy document itself, but a pseudo-merge of the old packaging manual). -- - mdz