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

Bug#851606: RFS: rmlint/2.4.6-1 [ITP]




On 14/02/17 18:56, Axel Beckert wrote:
> Hi,
>
> Carlos Maddela wrote:
>> Thanks for your suggestions. I have made the fixes and re-uploaded the
>> source files as the same version (since it still has yet to be
>> officially released) to mentors. I have applied for access to
>> collab-maint (just a few minutes ago), so I probably won't be able to
>> push my changes for another 24 hours. For the meantime, my changes are
>> available in the debian branch of https://github.com/e7appew/rmlint.git.
> Carlos: Thanks for the updated package.
>
> Roger: Thanks for the review and the reminder about this RFS.
>
> I've did a short first review. Will have a closer look at it probably
> this evening.
>
> One thing I noticed is that the upstream tar ball available at
> https://github.com/sahib/rmlint/releases and in the source package are
> not identical due to different compression.
>
> Is there a reason not to use the official source tarballs from
> https://github.com/sahib/rmlint/releases? (AFAIK and according to a
> short check, GitHub no more produces different tarballs upon
> subsequent downloads as it did in the past.)
>
> (Contentwise the contained files seem to be identical, so it's ok-ish,
> but such differences always cause extra work on any review of a source
> package, so it's preferred to use official tar balls if possible --
> which seems to be possible here.)
There wasn't any specific reason, except for trying to keep file sizes
as small as possible. I have reverted to using the original tarballs now.
>
> In debian/rules I found this:
>
> # Automatically disable tests in sbuild and pbuilder.
> ifneq (,$(filter /%nonexistent,$(HOME)))
> export DEB_BUILD_MAINT_OPTIONS += nocheck
> endif
>
> If this is to avoid issues with the test suite, it's a very unlucky
> solution as all of Debian's build daemon run sbuild and some sponsors
> upload packages using pbuilder. So when the binary packages are built
> which are to become the official binary packages, the test suite is
> not run. This especially applies to all other architectures than the
> maintainer's one and hence contradicts the idea that the test suite
> should be run at build time on all architectures to find
> architecture-specific issues.
>
> If a non-existent home is really such an issue, I'd rather set it to
> some other value and create that home directory with some sane
> contents, e.g. debian/tmp-home or so.
I only disabled the tests to speed up builds. There was initially some
problems with the tests inside of sbuild, which prevented successful
builds, but they have been resolved. The test suite doesn't make use of
any $HOME directory. It was just my way of detecting if the build was
being done from a chroot.

Anyway, I have removed those lines completely, and the tests are
performed now.

Cheers,

Carlos


Attachment: signature.asc
Description: OpenPGP digital signature


Reply to: