Gentoo Archives: gentoo-dev

From: "Anthony G. Basile" <blueness@g.o>
To: gentoo-dev@l.g.o
Subject: Re: [gentoo-dev] Re: toolchain.eclass: need to revert fixincludes commit
Date: Wed, 04 Feb 2015 11:32:45
Message-Id: 54D203E4.6050104@gentoo.org
In Reply to: [gentoo-dev] Re: toolchain.eclass: need to revert fixincludes commit by Michael Haubenwallner
1 On 02/04/15 03:06, Michael Haubenwallner wrote:
2 > On 02/03/2015 08:55 PM, Anthony G. Basile wrote:
3 >> On 02/02/15 19:06, vivo75@×××××.com wrote:
4 >>> Il 02/02/2015 23:30, Pacho Ramos ha scritto:
5 >>>> El sáb, 31-01-2015 a las 16:48 -0500, Anthony G. Basile escribió:
6 >>>>> Hi everyone,
7 >>>>>
8 >>>>> We need to revert the following change to toolchain.eclass:
9 >>>>>
10 >>>>> http://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/eclass/toolchain.eclass?r1=1.647&r2=1.648
11 >>>>>
12 >>>> Please remember to add a comment to the eclass with the reference to not
13 >>>> forget in the future why fixincludes stuff is needed ;)
14 >>>>
15 >>> fixincludes only on prefix and bsd is doable/acceptable?
16 >> @pacho. absolutely. part of the process is me learning the layers of history there.
17 >> its not like the code is hard to read, its just "why was this done?".
18 >>
19 >> @vivo75. the fixedincludes are removed after compiling, so they don't make it to $ROOT
20 >> during qmerge either for linux or bsd/prefix.
21 >> Its just that are needed during compiling for fbsd/prefix.
22 > To complete this info: At least in prefix they have to be installed as well,
23 > as subsequent packages may still use host's (libc at least) headers, and gcc
24 > requires them to be "fixed".
25
26 Thanks for the correction.
27
28 >
29 >> So a straight revert is fine.
30 > Fine for now, it's forked in prefix-overlay still.
31
32 We could build in intelligence with the appropriate `use prefix`, `use
33 userland_BSD` and `use userland_GNU`. I'll probably do a simple revert
34 this afternoon so we can have a working toolchain.eclass for fbsd, but
35 I'll look at the prefix overlay and compare eclasses and see if we can't
36 just bring in the prefix stuff into the main tree in a safe way. If
37 fixedincludes is the only issue, then I don't see a problem.
38
39 >
40 >> We need to explain this in a comment in case some "clever" future dev doesn't comes to the
41 >> same erroneous conclusion, that its okay to just disable their build.
42 > Thanks!
43 > /haubi/
44 >
45
46
47 --
48 Anthony G. Basile, Ph.D.
49 Gentoo Linux Developer [Hardened]
50 E-Mail : blueness@g.o
51 GnuPG FP : 1FED FAD9 D82C 52A5 3BAB DC79 9384 FA6E F52D 4BBA
52 GnuPG ID : F52D4BBA