Gentoo Archives: gentoo-dev

From: Peter Volkov <pva@g.o>
To: gentoo-dev@l.g.o
Cc: qa@g.o
Subject: Re: [gentoo-dev] [gentoo-commits] gentoo-x86 commit in dev-util/netbeans: ChangeLog netbeans-6.9-r3.ebuild
Date: Thu, 22 Jul 2010 08:46:16
Message-Id: 1279788261.6915.45.camel@tablet
In Reply to: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-util/netbeans: ChangeLog netbeans-6.9-r3.ebuild by "Miroslav Šulc (fordfrog)"
1 В Срд, 21/07/2010 в 17:18 +0200, "Miroslav Šulc (fordfrog)" пишет:
2
3 > > epatch_user() from euitls.eclass
4
5 > so this is undocumented feature? i did not know about this cool
6 > feature till now.
7
8 It's undocumented since it's better to avoid this feature in ebuilds
9 because this hack makes impossible to know what patches user have and
10 why package fails the way it fails. So better drop this code. If user
11 wants some useful patch {s}he should open the bug or use overlay.
12
13 epatch_user() was created for rare cases where package requires random
14 third-party patches for additional functionality that is impossible to
15 maintain in the tree. Good example is iptables: it has tons of
16 third-party patches. This patches are released days/months after
17 iptables release and even skip some iptables releases, don't have
18 upstream and thus making impossible to maintain them in the tree.
19
20 In general I think this is problem with our tools - we need some way to
21 apply custom patches and make emerge --info pkg could give us all
22 information about that custom patches... But until such or similar
23 solution implemented, please, avoid epatch_user() in the tree.
24
25 CC'ing QA: guys, I think this is your domain so in case you have any
26 comments here, please, do :)
27
28 --
29 Peter.

Replies