Bug#96597: changing policy requirements for debian native packages to _MUST_
On Mon, Mar 04, 2002 at 03:21:39PM -0800, Chris Waters wrote:
> On Sun, Mar 03, 2002 at 11:28:55PM -0500, Andres Salomon wrote:
> > dh_installchangelogs (debhelper-3.4.11) currently bails if one attempts
> > to install a non-debian changelog to
> > /usr/share/doc/<package>/changelog.gz in a debian-native package.
> Ok. I think maybe that's a bug, but I'm willing to be proven wrong.
I'm not sure. It would be nice if it _was_ a bug; make my life easier
as an upstream debian package maintainer..
> > Joey Hess has mentioned that various tools expect the changelog.gz
> > for debian-native packages to parsable as debian-style changelogs.
> What tools? I would definitely say that any such tool has a bug.
You'll have to ask him, I don't remember him giving any examples. Joey?
> > Given this expectation, one of two things should be done; either
> > policy should be changed to explicitly allow multiple changelogs for
> > debian-native packages, or explicitly disallow more than 1.
> "Given this expectation", sure, but I *don't* give that expectation.
> If changelog.Debian.gz exists, that should be considered the Debian
> changelog. If it doesn't, then changelog.gz should be considered the
> Debian changelog. A tool which doesn't work this way should be
> considered broken, IMO.
> > The current interpretation of it (section 13.8), for debian-native
> > packages, reads as "if the package only has 1 changelog, put it in
> > changelog.gz".
> That seems reasonable, and meets my definition above (there's no
> changelog.Debian.gz, so changelog.gz is the Debian changelog). Note
> that any Debian package _must_ have a Debian changelog, so, if there's
> only one changelog, it must be the Debian changelog.
> > This should be changed to something like:
> > "If the package is a debian-native package, the changelog installed in
> > /usr/share/doc/<package>/changelog.gz MUST be the debian changelog.
> No. Current policy seems fine to me.
Should be changed if it breaks tools, I mean.
> > The exact quote from the current policy is:
> > "If the package has only one changelog which is used both as the Debian
> > changelog and the upstream one because there is no separate upstream
> > maintainer then that changelog should usually be installed as
> > /usr/share/doc/package/changelog.gz; if there is a separate upstream
> > maintainer, but no upstream changelog, then the Debian changelog should
> > still be called changelog.Debian.gz."
> Frankly, as far as I'm concerned, if something is a Debian-native
> package, it should only have one changelog. If there's any reason for
> maintaining two, then that probably counts as sufficient justification
> for NOT creating a Debian-native package! A proposal which made this
> into policy might have my support. The current proposal does not.
Policy states that history should not be rewritten; ie, past changelog
entries should remain as they were. What happens when maintainership of
a package moves upstream? The old changelog entries in ./debian/changelog
have to stick around, while new changelog entries may be written to
./ChangeLog. Or, perhaps you have a package like e2fsprogs, where there
are upstream changelogs in various directories? In that case, it would
be only natural to give the debian subdirectory its own changelog..
> In other words, I doubt if I approve of what you seem to be trying to
> do in the first place. Although, without more details, it's hard to
> be sure.
> In any case, the argument for the current proposal seems to be that
> existing tools make assumptions that do not match what policy says.
> Unless and until an effort to fix those tools fails, I think we should
> leave policy as it stands. Especially since policy has been frozen
> for months now.
> So, I guess I have two questions: 1. What are these tools that joeyh
> refers to, and why can't they be fixed? 2. Why the heck are you
> trying to make a Debian-native package with two changelogs? Without
> reasonable answers to both of these questions, I'm afraid I'll have to
> oppose the proposal.
I've CC'd joey on this, hopefully he can clarify his earlier
statement(s). As for #2, see above.
> Chris Waters | Pneumonoultra- osis is too long
> email@example.com | microscopicsilico- to fit into a single
> or firstname.lastname@example.org | volcaniconi- standalone haiku
"I think a lot of the basis of the open source movement comes from
-- Andrew Tridgell <http://www.linux-mag.com/2001-07/tridgell_04.html>