Bug#544981: debian-policy: Discourage native packages that are not tightly specific to Debian
On Fri, Sep 04, 2009 at 08:26:10PM +0200, Luk Claes wrote:
> Wouter Verhelst wrote:
> >>Given the recent thread in debian-devel[1], I think we should document in
> >>policy that packages that are not tightly related to Debian shouldn't be
> >>native.
> >
> >*sigh*
> >
> >So I spent a whole subthread trying to explain that I think this is
> >*not* true, and seemed to get consensus on that, and now you want to get
> >this into policy?
>
> Consensus is a big word, you managed to get people agree that if
> maintainers really considered all the downsides even our complaints
> from time to time that it would be acceptable...
Why, yes, indeed, that's what I'm saying: that there seemed to be
consensus on the fact that it is acceptable if people do indeed consider
the downsides; that the "shouldn't be native" statement is wrong.
> >Gee.
> >
> >Whether or not a native package makes sense should be the maintainer's
> >prerogative, not decided by policy. As I said in the thread on -devel,
> >there can be good reasons for making a package native. E.g., the
> >maintainer doesn't have to deal with two releases (one upstream and one
> >for debian) for every code change, but can just do one; there is
> >immediate use of a translation team; releases are at least tested on
> >Debian's architectures when they are released; etc.
>
> When using a non-native package, the maintainer does not have to do
> any separate release as the upstream tarball is in orig.tar.gz
True. However, there will be no significant difference between making the
package native or not in that case, IMHO.
> The translation team focuses on native packages (next to other
> Debian specific translation), because it does not have the resources
> to do all of it and native packages are considered Debian
> specific... so this is actually in some kind abusing the translation
> team if the package does not have to be native.
Hmm. It could be seen that way, I guess.
> >There are also obvious downsides to doing so, and it's probably a good
> >idea to document these somewhere (though I doubt policy is the place for
> >that; this is more something for the devref). However, outright claiming
> >that it should not be done, will a) make a bunch of packages
> >insta-buggy (which is bad, as far as policy is concerned), and b) is not
> >the right thing to do, IMO.
>
> They are already buggy IMHO.
Perhaps, but that does not mean that they in fact are. It has been okay
for quite a while to do this, and several packages are in fact doing so;
changing policy does make them insta-buggy.
What I'd like to see before I would support this proposal (or anything
like it), is how exactly the practice of releasing non-Debian-specific
software as native packages is causing harm to either Debian, or the
greater free software community as a whole; since I don't think it does,
and I don't think we should forbid a practice which may make a
maintainer's workflow easier if it is indeed harmless.
In other words: what kind of problems do you think this will cause that
have an effect on anyone *but* the maintainer? As said, I agree that
documenting the problems with maintaining a package natively is a good
thing to do, so that anyone thinking about going down that road can make
an informed decision; but that is a far cry from what's being proposed
here.
Sure, if something is released as a native package, that does mean that
people repackaging the software for other distributions may want to skip
a few releases now and then -- but I do not see how that is any
different from, say, the vim release model, where packagers may want to
collect a few patches before uploading a new package, rather than
uploading a new package every time Bram releases a patch (which happens
about every other day or some such, AIUI). Sure, maintaining software as
a native package does introduce the requirement that the
other-distribution-packagers know what they're doing, and that they keep
up with development with the Debian developer who maintains the package;
but then I would hope they would be doing that anyhow.
--
The biometric identification system at the gates of the CIA headquarters
works because there's a guard with a large gun making sure no one is
trying to fool the system.
http://www.schneier.com/blog/archives/2009/01/biometrics.html
Reply to: