features/trinity-applications-stable-second-attempt #189
Merged
SlavekB
merged 2 commits from features/trinity-applications-stable-second-attempt
into master
3 years ago
Loading…
Reference in new issue
There is no content yet.
Delete Branch 'features/trinity-applications-stable-second-attempt'
Deleting a branch is permanent. It CANNOT be undone. Continue?
Version 14.0.8 ebuilds for as many packages in trinity-apps as I could make build back in June, plus trinity-base/kontact, which somehow got missed before. (Please junk the old features/trinity-applications-stable branch, which contains the same stuff, but in a crufted-up format that would have made it extremely painful to merge.)
pkgcheck doesn't seem to have much problem with these (if we ignore spew about whitespace, extra files in Manifests, and unrecognized i18n USE flags).
MR re-introduces superfluous
14.0.7
manifests. This is easily fixed by runningrepoman manifest
insidetrinity-apps
directory. You can simplygit commit --amend
afterwards. You can also use this to change the commit message of current HEAD when needed.Right now you have two commits, where the second one is just fixing stuff from the first; we don't need review noise in origin/master, so you can squash the commits easily using
git rebase -i HEAD~2
(that's interactively choosing the desired action for the rebase).Always run
git status
before and after such operations so you know where you are right now and that the previous command did what you expected.In gentoo.git we have adopted a common workflow where we always prefix commits with
category/packagename:
and a limit to ~70 characters, in this case where you add new packages in one commit to the same category it would becategory:
prefix. It is good practice to be curt and precise in the git message (likex.x.x version bump
) and have any detailed description or full sentences in the message body.All style comments I made are not necessary must-have fixes, if there is a
9999
version I will most likely have already gone over it. If there is no9999
yet, it would be nice to have your release version tidied up so I don't need to check again whenever a9999
will be introduced (and likely use the release version as template).All non-style comments would ideally be fixed in separate commits to the
9999
version, so they are not lost on subsequent version bumps.# Distributed under the terms of the GNU General Public License v2
EAPI="7"
TRINITY_MODULE_TYPE="applications"
Style: For all ebuilds: There should always be an empty line after
EAPI
.inherit trinity-base-2
DESCRIPTION="A complex calculator for TDE"
Style: For all ebuilds: During my time submitting PRs to kde overlay I was always told to get rid of filler words at the beginning of
DESCRIPTION
. That's just a nitpick, but when you think about it, really nothing is gained if every description starts withA/An
etc.inherit trinity-base-2
DESCRIPTION="A complex calculator for TDE"
KEYWORDS="~amd64 ~x86"
Style: For all ebuilds:
Of course, in ebuilds relying on the deprecated special functions
need-trinity
,set-trinityver
andneed-arts
you need to make sure those are called before those vars as well.[1] https://gitweb.gentoo.org/repo/gentoo.git/tree/skel.ebuild#n25
[2] https://wiki.gentoo.org/wiki/Basic_guide_to_write_Gentoo_Ebuilds#How_to_create_an_ebuild
SLOT="${TRINITY_VER}"
src_configure() {
mycmakeargs=(
For all ebuilds: Should be
local mycmakeargs
inherit trinity-base-2
DESCRIPTION="Advanced music player for TDE."
It's not a sentence so we generally don't use full stop in
DESCRIPTION
.dev-lang/ruby:*
media-libs/taglib
dev-db/sqlite
xine? ( <media-libs/xine-lib-1.2.10 )
Style: For all ebuilds: Dependencies should always be sorted alphanumerically, with all unconditional deps before the use-conditional ones.
SLOT="${TRINITY_VER}"
IUSE+=" css dvd dvdr vcd debug alsa ffmpeg ffmpeg_all_codecs \
No need for
\
within any quotations.IUSE+=" +ssl asus"
# SSL support might need tdelibs build with +ssl USE.
Could you test that though? In that case it should depend on tdelibs[ssl?].
Note: I just checked
kbiff-4.0
and it just depends onqtcore:4[ssl]
(by default for all ebuilds using kde4-base.eclass, so no one really ever made an effort to check that per-package), not however on kdelibs. It is possible though that, as KDE contributions moved into Qt during subsequent major porting efforts, it previously relied on functionality provided bykdelibs:3[ssl]
.inherit trinity-base-2
DESCRIPTION="Autostart module for KControl "
Cleanup trailing space
SLOT="${TRINITY_VER}"
# NVControl support needs the nvidia-settings package,
nvidia-settings
was really merged intonvidia-drivers
, and it is available viax11-drivers/nvidia-drivers[tools]
according toequery u nvidia-drivers
.DESCRIPTION="Trinity personal information manager"
KEYWORDS="~amd64 ~x86"
COMMON_DEPEND="~trinity-base/libtdepim-${PV}
Superfluous
COMMON_DEPEND
, as below we can seeDEPEND=${RDEPEND}
anyway.DEPEND+=" $COMMON_DEPEND"
RDEPEND+=" $COMMON_DEPEND"
TSM_EXTRACT_ALSO="libtdepim/ libtdenetwork/"
all eclass meta vars should always be defined before
inherit
.I have yet to be able to get repoman to do anything useful with this overlay (it throws a fit about metadata files and quits, and I don't have the energy to figure out what's going on), and in any case, as long as the metadata files contain the needed signatures, extras don't matter. Only the automated tooling ever looks inside those files.
I'll try to get to the substansive problems (that is, the things that aren't just style complaints) after Christmas.
Upon further review, I see that:
-All the issues are inherited from the live ebuilds, rather than being things that I introduced.
-All of them except the dependency issues with kima and kbiff are stylistic even when not marked that way—that is, they don't break the ebuild. Actually, even the kima dep issue doesn't break the ebuild, it just removes some functionality.
As for kbiff, I don't actually use it, so I wouldn't be able to recognize abnormal behaviour short of it outright crashing. tdelibs-14.0.8 won't build for me with -ssl (the flag seems to be broken: it tries to compile SSL-related files anyway, then errors out when it can't find a related definition), so further testing is not possible at this time.
Some poking around with ldd suggests that kbiff doesn't directly link libssl or libcrypto when built with the ssl flag active, so it's probably drawing from tdelibs. (I suppose whoever coded it might also have directly reimplemented SSL, but why would you?)
Either merge the 14.0.8 ebuilds in here as-is or discard the whole thing altogether; I lack the energy to care which right now and just want the whole thing to go away.
Efforts have been made here and there is no point in throwing it away. I suggest merging it and resolving any issues and comments afterwards. Any objections?
494d42a599
into master 3 years agoReviewers
494d42a599
.