Gentoo Archives: gentoo-portage-dev

From: Zac Medico <zmedico@×××××.com>
To: gentoo-portage-dev@l.g.o
Subject: Re: [gentoo-portage-dev] PATCH: clean PORTAGE_TMPDIR (bugs 85803 and 105706)
Date: Sun, 30 Oct 2005 20:30:09
Message-Id: 43652D2E.1000308@gmail.com
In Reply to: Re: [gentoo-portage-dev] PATCH: clean PORTAGE_TMPDIR (bugs 85803 and 105706) by Brian Harring
1 Brian Harring wrote:
2 > On Sun, Oct 30, 2005 at 04:33:28AM -0800, Zac Medico wrote:
3 >
4 >>http://bugs.gentoo.org/show_bug.cgi?id=85803
5 >>http://bugs.gentoo.org/show_bug.cgi?id=105706
6 >>
7 >>There are a few cases where unwanted directories are left behind in
8 >>PORTAGE_TMPDIR with varying amounts of files inside. When packages are
9 >>merged this may occur with the --buildpkgonly option or with binary
10 >>packages. It may also occur when a package ia unmerged. The attached
11 >>patch should clean up all of these cases. Feedback is appreciated.
12 >
13 >
14 > Correct the BUILDDIR misses in doebuild imo; this seems a bit like a
15 > bandaid over the fact doebuild's clean only cleanses ${T} .
16 >
17 > tag the shutil.rmtree into doebuild instead of adding a secondary
18 > function that must be used, is my opinion- tracing out cleanup and
19 > mydo == "clean" demonstrates where the miss is.
20 >
21 > ~harring
22
23
24 Indeed, metaphorically speaking, my patch _is_ a bandaid. However, stable portage has many bandaids. While this certainly in not the best long term solution, it seems to work quite well in the short term. ;)
25
26 I choose the bandaid method in order to keep the patch as small as possible and still keep PORTAGE_TMPDIR clean. AFAICT it would take a *much* larger patch to clean up all the same cases without the bandaid approach.
27
28 On a side note, my patch should be changed so that it doesn't try to clean_builddir when things like "--pretend" and "--fetchonly" are in emerge's myopts.
29
30 Zac
31 --
32 gentoo-portage-dev@g.o mailing list

Replies