Bug#998041: RFS: makedeb/11.0.1-1 [ITP] -- The modern packaging tool for Debian archives
Package: sponsorship-requests
Followup-For: Bug #998041
Control: tags -1 moreinfo
Lintian's still unhappy:
(Please integrate lintian in your workflow, as those lintian remarks had been
found in previous reviews already.)
I've added some remarks.
E: makedeb: extended-description-is-empty
W: makedeb: description-synopsis-starts-with-article
W: makedeb source: no-debian-changes
W: makedeb: no-manual-page usr/bin/makedeb-makepkg
W: makedeb: no-manual-page usr/bin/makepkg-template
W: makedeb: script-not-executable [etc/makepkg.conf]
-- Remark: possibly this file should not be a script!
X: makedeb source: debian-watch-does-not-check-gpg-signature [debian/watch]
P: makedeb source: package-uses-old-debhelper-compat-version 12
P: makedeb source: silent-on-rules-requiring-root [debian/control]
P: makedeb source: trailing-whitespace debian/rules (line 4)
P: makedeb source: trailing-whitespace debian/rules (line 5)
P: makedeb source: trailing-whitespace debian/rules (line 6)
P: makedeb source: update-debian-copyright 2021 vs 2022 [debian/copyright:11]
X: makedeb source: upstream-metadata-file-is-missing
Beside:
- /etc/makepkg.conf:
-you are packaging arch:all but have hard-coded arch-dependent settings for
amd64. Is this intentional?
-I don't think that should be a script needing a shebang, should it?
- d/makedeb-docs.docs d/README d/README.source should not be needed (and they
do not contain useful information)
- Somewhere in the documentation it still says "the modern packaging tool for
Debian archives", which needs still to be changed, accordingly to your
message in #998039#32.
Disclaimer: I'm not going to sponsor this package.
I believe that makedeb creates packages in a way that might cause problems for
our users. (e.g as laid out in 998039#22; it makes dependency handling a
user-problem, which is a core task of a packaging management system. So I
thinkg it needs still some development before it should be uploaded.
I was playing with makedeb, but unfortunatly with very mixed result:
For example, I test-built "moonlight-qt" (selected because it was on the
frontpage of the homepage [¹] and not arch-all generates this Depends: line
for the generated binary package:
> Depends: libegl1-mesa-dev, libgl1-mesa-dev, libopus-dev, libqt5svg5-dev, libsdl2-dev, libsdl2-ttf-dev, libssl-dev, libavcodec-dev, libva-dev, libvdpau-dev, libxkbcommon-dev, qt5-qmake, qtbase5-dev, qtdeclarative5-dev, qtquickcontrols2-5-dev, wayland-protocols, qml-module-qtquick-controls2, qml-module-qtquick-layouts, qml-module-qtquick-window2, qml-module-qtquick2, ffmpeg
(note that it would also depend on qt5-default, but I had to edit it because this package is gone in sid, and I used a sid container for my tests.)
Another package, zotero, claims "zotero is not available for the 'x86_64' architecture.**", and I
found many others with the exact same problem... As those working using "x86_64" in the PKGBUILD
and those which don't "amd64", makes me wonder if the PKGBUILD used or how it
is parsed by makedeb is stable. Also, my i386 (on amd64 hardware -- Multiarch)
chroot claims to be 64 bit... I guess there are bugs here...
Also (IIRC gh (some github tool)) created an arch:all package which having arch-dependent binaries.
(Frankly, I ran accross so many broken PKGBUILD recipes, all sorts of errors... Is there
(automated) QA?)
--
tobi
[¹] https://mpr.makedeb.org/
Reply to: