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

Bug#543417: README.source patch system documentation requirements considered harmful



Raphael Hertzog <hertzog@debian.org> (25/08/2009):
> > > > The existence of a debian/patches directory proves that the
> > > > package uses some patch system and that he should investigate
> > > > more.
> > > 
> > > But this assertion is not true once new source packages are
> > > “ready”. :) Some forward-looking can't hurt when we design policy.

Oh, and I failed to catch that, which was too obvious: I said that
debian/patches is a signal of possible patch(es| system) and that one
should investigate. So your asserting my assertion is not true once
new source packages are ready is false.

> > And fortunately, Chris's concern will go away when that's the
> > case, since AFAICT the patches are going to be applied (according
> > to my reading of dpkg-source(1)), and there will be no reason to
> > be bothered with README.source at all.

Sorry, sure, I should have mentioned “for the 3.0 (quilt) source
packages”. And old source packages still will contain debian/patches,
and people might still understand from that directory that patches may
exist.

> That's my point. Without README.source (assuming the rules are
> changed to not force the creation of that file for common patch
> systems), seeing debian/patches/ is not enough to know if the
> patches are already applied or not.

Fortunately, dpkg-source is there for that:
| $ LC_ALL=C dpkg-source -x ../python-networkx_0.99-2.dsc
| dpkg-source: warning: extracting unsigned source package (../python-networkx_0.99-2.dsc)
| dpkg-source: info: extracting python-networkx in python-networkx-0.99
| dpkg-source: info: unpacking python-networkx_0.99.orig.tar.gz
| dpkg-source: info: unpacking python-networkx_0.99-2.debian.tar.gz
| dpkg-source: info: applying all patches with quilt push -a -q
| Applying patch 10_doc_relocation
| Now at patch 10_doc_relocation

> With the current rule in policy, only 3.0 (quilt) source package
> would not have a README.source and the lack of README.source fact
> can be used to decide that no further investigation is required.

No. *Suggesting* to have a README.source when a patch system is used
doesn't mean that you can infer that no patch system is used if
there's no README.source. On the contrary, having a debian/patches
directory is a fairly good indicator that there are patches. Whether
they automatically got applied when unpacking can be seen when…
unpacking, see above.

> > Thanks for playing, though.
> 
> Thanks for trying to understand other's people point of view. :)

No problem.

> That said, I don't care much about this specific argument but I like
> to highlight the fact that you should try to understand the point of
> view of the other proponents if you ever want to reach some (other)
> decision that satisfies everybody.

I might have troubles trying to follow your logical mistakes. Sorry
for that.

Mraw,
KiBi.

Attachment: signature.asc
Description: Digital signature


Reply to: