Gentoo Archives: gentoo-dev

From: Alexis Ballier <aballier@g.o>
To: gentoo-dev@l.g.o
Cc: alexxy@g.o
Subject: Re: [gentoo-dev] Getting proper USE_EXPAND variable(s) for multilib
Date: Wed, 23 Jan 2013 18:48:32
Message-Id: 20130123154812.6ae04e15@gentoo.org
In Reply to: Re: [gentoo-dev] Getting proper USE_EXPAND variable(s) for multilib by Alexey Shvetsov
1 On Wed, 23 Jan 2013 21:36:22 +0400
2 Alexey Shvetsov <alexxy@g.o> wrote:
3
4 > В письме от 23 января 2013 08:03:56 пользователь Alexis Ballier
5 > написал:
6 > > On Wed, 23 Jan 2013 09:24:26 +0100
7 > >
8 > > Michał Górny <mgorny@g.o> wrote:
9 > > > On Mon, 21 Jan 2013 10:27:30 -0300
10 > > >
11 > > > Alexis Ballier <aballier@g.o> wrote:
12 > > > > > To be honest, I don't know if there's other way to hide USE
13 > > > > > flags than using USE_EXPAND_HIDDEN. If we want to use that,
14 > > > > > we'd have
15 > > > > >
16 > > > > > to split the flags per-arch, i.e. have:
17 > > > > > MULTILIB_AMD64="abi1 abi2 abi3"
18 > > > > > MULTILIB_PPC64="abi1 abi2 abi3"
19 > > > > >
20 > > > > > with appropriate USE_EXPAND_HIDDEN set by profiles.
21 > > > >
22 > > > > I don't like that at all.
23 > > > > I'd go for ABI= the union of all the MULTILIB_ABIS variables (if
24 > > > > there is no name collision)
25 > > > > we certainly want skype to depend on libitneeds[abi_x86], not
26 > > > > 'amd64? ( libitneeds[abi_amd64_x86] ) x86? ( libitneeds )'
27 > > >
28 > > > Just a quick idea.
29 > > >
30 > > > How would you feel about abi_x86_32? (similarly _64, _x32)
31 > > >
32 > > > That would be almost natural names with the trick variable being
33 > > > ABI_X86, therefore having all the fore-mentioned advantages.
34 > > >
35 > > > The deps would look like:
36 > > > libitneeds[abi_x86_32]
37 > >
38 > > Sounds good too, I just would want it to be shared between arches
39 > > that can: amd64/x86/x32, ppc/ppc64, mips, sparc32/sparc64, etc.
40 > > You mean you will hide ABI_X86 USE_EXPAND on non-x86 arches, right ?
41 > > This would have all the benefits I think, very good idea :)
42 > >
43 > > Alexis.
44 >
45 > Shared abi names are bad idea. For example
46 > mips abis :
47 > o32
48 > n32
49 > n64
50 > eabi
51 > x86:
52 > x86_32
53 > x86_x32
54 > x86_64
55 >
56 > Actualy first three one are equivalent in their internal behavior.
57 > But i dont think that its good idea to have one name for all.
58
59 Sorry, I don't get it. What was meant here was one USE_EXPAND variable
60 per group of similar arches. Different abis are, of course,
61 distinguished within each variable.
62
63 > Think
64 > about multiarch installs where you can have binaries from different
65 > architectures in one system.
66
67 It depends what we want through multilib. I personally think multiarch
68 is out of scope: there is crossdev for that.
69
70 Alexis.