[Debian GNUstep maintainers] Misc issues with gnustep-base

[Eric Heintzmann]

Le 22/10/2017 à 18:32, Yavor Doganov a écrit :
> On Thu, 19 Oct 2017 02:23:06 +0300,
> Eric Heintzmann wrote:
>> Le 18/10/2017 à 12:32, Yavor Doganov a écrit :
>>> I found three more issues; the first two are serious:
>> OK, let's untighten the dependencies
>> OK, let's add gnustep-base-runtime to dependencies of the the lib
>> OK.
> All fixed, plus the noopt handling issue.  Please check/test.  Here's
> the changelog:
>
> gnustep-base (1.25.0-2) unstable; urgency=medium
>
>   * debian/patches/autogsdoc-reproducibility.patch: New; fix
>     reproducibility issue with autogsdoc.
>   * debian/patches/series: Update.
>   * debian/templates/control.m4: Relax some dependencies for smoother
>     transitions.
>   * debian/control: Regenerate.
>   * debian/rules: Remove --with autoreconf; not needed for compat level
>     10.  Handle "noopt" manually to take advantage of gnustep-make's
>     debug=yes option.
>     (DEB_LDFLAGS_MAINT_APPEND): Add -Wl,--no-undefined.
>     (override_dh_shlibdeps): Add -runtime to generated shlibs.
>   * debian/templates/libgnustep-baseN.lintian-overrides.m4: Override
>     shlibs-declares-dependency-on-other-package.
>
>  -- Yavor Doganov <yavor at gnu.org>  Sun, 22 Oct 2017 18:55:09 +0300
>
> A changelog entry "Upload to unstable" or similar is needed to
> eliminate a lintian pedantic tag.  I guess you'll add it when the time
> comes.
OK
> Regarding the transition, I suggest to wait until -base is built on
> mips64el before contacting the release team.
OK
During your tests, have you also checked the non Debian GNUstep
Maintainers packages ? (biococoa, oolite , openvpn-auth-ldap, sbjson,
sope, unar, sogo ...)
(There are also Gürkan's pending uploads: chess.app and pikopixel.app)

>
> Unrelated suggestion: I propose to add this snippet to
> gnustep-make/debian/addons/config.mk:
>
> ,----
> | # Optimization is handled automatically by debhelper but
> | # gnustep-make's debug=yes adds -DDEBUG -fno-omit-frame-pointer.
> | ifndef optim
> | ifneq (,$(findstring noopt,$(DEB_BUILD_OPTIONS)))
> | optim := debug=yes
> | endif
> | endif
> `----
>
> That way, packages including config.mk can delete it from their own
> debian/rules and take advantage of a uniform noopt handling.
I think it is a good idea.
And It seems to me a good idea to upload this change now, before
uploading other GNUstep packages.
But on the other hand, I don't like the idea of multiplying uploads.
What do you think ?