Bug 44817 - dubious RDEPEND on sed
|
Bug#:
44817
|
Product: Gentoo Linux
|
Version: unspecified
|
Platform: All
|
|
OS/Version: All
|
Status: RESOLVED
|
Severity: normal
|
Priority: P2
|
|
Resolution: WORKSFORME
|
Assigned To: qa@gentoo.org
|
Reported By: spider@gentoo.org
|
|
Component: Applications
|
|
|
URL:
|
|
Summary: dubious RDEPEND on sed
|
|
Keywords:
|
|
Status Whiteboard:
|
|
Opened: 2004-03-16 01:07 0000
|
well, I doubt most of theese are really correct RDEPENDS, however some may
warrant. (vanilla-sources, development-sources )
However since sed is in the default profile , I'm not sure how many of them
really need it mentioned at all.
sys-kernel/vanilla-sources-2.4.25/RDEPEND: !build? ( sys-apps/sed
sys-kernel/development-sources-2.6.4_rc1/RDEPEND: !build? ( sys-apps/sed
sys-apps/util-linux-2.12-r4/RDEPEND: >=sys-apps/sed-4.0.5
sys-apps/slocate-2.7-r5/RDEPEND: >=sys-apps/sed-4
sys-apps/net-tools-1.60-r7/RDEPEND: >=sys-apps/sed-4
dev-lang/tk-8.3.4-r1/RDEPEND:>=sys-apps/sed-4.0.5
net-misc/ntp-4.1.2/RDEPEND:>=sys-apps/sed-4.0.5
net-misc/rsync-2.6.0/RDEPEND: >=sys-apps/sed-4
net-misc/rdate-990821-r1/RDEPEND:>=sys-apps/sed-4
sys-fs/xfsprogs-2.3.9/RDEPEND: >=sys-apps/sed-4
app-arch/unzip-5.50-r2/RDEPEND: >=sys-apps/sed-4
media-libs/fontconfig-2.2.1/RDEPEND:>=sys-apps/sed-4
media-libs/pdflib-5.0.2/RDEPEND:>=sys-apps/sed-4
media-libs/sdl-mixer-1.2.5-r1/RDEPEND: >=sys-apps/sed-4
media-libs/libquicktime-0.9.2_pre1/RDEPEND:>=sys-apps/sed-4.0.5
x11-base/xfree-4.3.0-r5/RDEPEND: >=sys-apps/sed-4
net-nds/openldap-2.1.26/RDEPEND: >=sys-apps/sed-4
net-mail/exim-4.21/RDEPEND:>=sys-apps/sed-4.0.5
net-mail/pine-4.58/RDEPEND: >=sys-apps/sed-4
sys-boot/lilo-22.5.8-r1/RDEPEND:>=sys-apps/sed-4
app-portage/mirrorselect-0.82-r3/RDEPEND: sys-apps/sed
app-portage/genflags-0.96/RDEPEND: sys-apps/sed
media-gfx/xloadimage-4.1-r1/RDEPEND:>=sys-apps/sed-4.0.5
media-gfx/imagemagick-5.5.7.15/RDEPEND:>=sys-apps/sed-4
net-dialup/wvdial-1.53-r1/RDEPEND:>=sys-apps/sed-4
net-dns/bind-9.2.2-r2/RDEPEND: >=sys-apps/sed-4
dev-perl/ExtUtils-MakeMaker-6.20/RDEPEND:>=dev-lang/perl-5.8.0-r12
>=sys-apps/sed-4
dev-db/mysql-4.0.18/RDEPEND: >=sys-apps/sed-4
media-sound/mpg123-0.59s-r2/RDEPEND: >=sys-apps/sed-4
app-misc/screen-4.0.1-r2/RDEPEND: >=sys-apps/sed-4
app-misc/figlet-221-r1/RDEPEND: >=sys-apps/sed-4
media-video/xanim-2.80.1-r4/RDEPEND: >=sys-apps/sed-4.0.5
media-video/avifile-0.7.38.20030710/RDEPEND: >=sys-apps/sed-4
media-video/mjpegtools-1.6.1.90-r1/RDEPEND: >=sys-apps/sed-4
net-im/gaim-0.75-r10/RDEPEND: >=sys-apps/sed-4.0.0 )
games-arcade/frozen-bubble-1.0.0-r3/RDEPEND: >=sys-apps/sed-4
There are definitely false positives in there.
games-arcade/frozen-bubble-1.0.0-r3 for example does not DEPEND on sed.
Spider - can you regenerate this list against current CVS? It wasn't clear
from our interaction on irc yesterday that it was possible with the tool you're
using, but it would make targetting existing problems easier. I fixed a
few of the ebuilds and some others had already been fixed.
I'll take a look at it, the current one just accessed a clean-slate set built
from successive packages in cvs, so it has only tracked packages that actually
update revision when changed. yep, its a fault, but its also a test against
how much disrespectancy there is between what can be installed and what is in
CVS.
We fail in reproducability. packages of the same versions frequently have
different dependencies.
It sounds like you're suggesting that changing deps requires a rev bump.
The ebuild docs say, regarding -r bumps: "If you make non-trivial improvements to an existing ebuild file..." I'm not sure changing:
DEPEND="...sed..."
to
DEPEND="...sed..."
RDEPEND=""
qualifies as a non-trivial improvement. Since the installed files didn't
change, I don't think the users should be forced to recompile the package
in this case.
No, I'm in fact not advocating necessary revision bumps for things like this.
However, The issue I'm aiming at is reproducability. That the same version/Revision should behave the same way on different systems (that you could reproduce behaviours and buglets as well) Which is what fails.
Theese scripts were as you can se not ran against /usr/portage, but against one of my clean-cut roots (aka stage3 installed in november and then updated monthly afterwards, usually from a grp set)
I'll see a bout regenerating the lists from /usr/portage
dev-embedded/sdcc-2.4.0
dev-embedded/sdcc-2.4.0_p20040304
removed sed RDEPEND (and other that were not necessary) - Thanks spider
Similar fault to bug 44816. Probably should be solved the same way.
i fixed a bunch of these myself
really the only way you'll get the rest fixed is if you file a bug for each package