Re: [SCM] Debian package checker branch, master, updated. 2.5.0-rc1-9-gb88de18
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256
On 2011-03-14 18:21, Russ Allbery wrote:
> Niels Thykier <niels@thykier.net> writes:
>
>> Are you referring to #539760 ? That has just been tagged wontfix without
>> any reason given from any Lintian maintainer as far as I can tell. The
>> only thing I see is a wontfix from you with no explanation and a retitle
>> from Adam to (presumably) improve the readability/sorting of the BTS.
>
> Bleh. Apparently all the discussion actually happened outside of the
> bug. That's not very useful; sorry about that.
>
:)
>> Yet, we have removed versioned nagging for quilt for versions in Squeeze
>> in 2.5.0~rc1
>
> Right, but quilt doesn't explicitly say to use a versioned dependency in
> its documentation. The rationale here was that the debhelper
> documentation explicitly recommends that you set a minimum version bound
> that matches the compat file, and I'm fairly sure that in one of the
> previous discussions of this I confirmed with Joey that he did mean to
> recommend that.
>
True, but if you consider the current Debian Policy (4.2):
"""
[...] In particular, this means that version clauses should be used
rigorously in build-time relationships so that one cannot produce bad or
inconsistently configured packages when the relationships are properly
satisfied.
"""
Then that ought to hold for quilt as well. That being said...
> For versions that are older than what's in stable, it's really a pedantic
> tag, though. I was hesitant to remove the check until I did the work to
> split it so that the pedantic check could be retained, but maybe it's
> better to just make it go away for right now and add the pedantic check
> later.
>
I am okay with using a pedantic tag for this kind of thing. Any
suggestions for the name? Current tag is:
package-lacks-versioned-build-depends-on-debhelper
So "package-lack-pedantic-versioned-build-depends-on-debhelper"?
> This used to be much more of an issue since there was a much larger
> version spread between unstable, stable, and oldstable in terms of compat
> levels than we have now. I agree that in the current world where the only
> debhelpers anyone could conceivably use are 7 and 8, it looks kind of
> silly. The main advantage is to make sure people stay in the habit of
> bumping compat and debhelper in sync (before Lintian was widespread, I
> used to see packages all the time in Debian that had a higher compat level
> than their debhelper dependency), but that's really a pedantic-level tag.
>
Of course the compat <=> B-D bumping is a good argument. On a related
note, I noticed that debhelper in git is (most likely) going to open
compat 9 with the next upload (which was expected to be 8.1.3 last I
checked).
~Niels
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/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=ID/X
-----END PGP SIGNATURE-----
Reply to: