1 |
On Sat, 26 Jan 2013 14:43:38 -0300 |
2 |
Alexis Ballier <aballier@g.o> wrote: |
3 |
|
4 |
> On Sat, 26 Jan 2013 18:06:32 +0100 |
5 |
> Michał Górny <mgorny@g.o> wrote: |
6 |
> |
7 |
> > On Sat, 26 Jan 2013 12:30:16 -0300 |
8 |
> > Alexis Ballier <aballier@g.o> wrote: |
9 |
> > |
10 |
> > > On Sat, 26 Jan 2013 16:08:45 +0100 |
11 |
> > > Michał Górny <mgorny@g.o> wrote: |
12 |
> > > |
13 |
> > > > On Sat, 26 Jan 2013 11:54:44 -0300 |
14 |
> > > > Alexis Ballier <aballier@g.o> wrote: |
15 |
> > > > |
16 |
> > > > > On Sat, 26 Jan 2013 13:11:41 +0100 |
17 |
> > > > > Michał Górny <mgorny@g.o> wrote: |
18 |
> > > > > |
19 |
> > > > > > > (maybe protect it with has_multilib_profile if you wish) |
20 |
> > > > > > |
21 |
> > > > > > Well, the current code assumes that no flags == non-multilib |
22 |
> > > > > > profile. |
23 |
> > > > > |
24 |
> > > > > .. and I hit send to quickly: |
25 |
> > > > > coming back to the skype example, with this assumption, on x86 |
26 |
> > > > > libitneeds will have no abi useflag enabled while on amd64 it'll |
27 |
> > > > > have abi_x86_32, so you'll end with deps like: x86? ( libitneeds |
28 |
> > > > > ) !x86? ( libitneeds[abi_x86_32] ) |
29 |
> > > > |
30 |
> > > > Hmm, maybe we should make 'plain x86' use.force the abi_x86_32 |
31 |
> > > > flag (though keeping it hidden). Then the following dep will work |
32 |
> > > > for both arches: |
33 |
> > > > |
34 |
> > > > libitneeds[abi_x86_32] |
35 |
> > > |
36 |
> > > yep, that's the best thing to do IMHO, but then if you want no |
37 |
> > > multilib -> returns empty list then you'll have to check it with |
38 |
> > > has_multilib_profile :) |
39 |
> > |
40 |
> > Is this really necessary? AFAIU from the endclass standpoint there's |
41 |
> > no real difference between non-multilib build and build for one ABI. |
42 |
> > I could even make the multilib code paths used only when at least 2 |
43 |
> > ABIs are enabled. |
44 |
> |
45 |
> I don't think it's necessary, but that's what your function had as a |
46 |
> spec :) |
47 |
> You could probably just use one code path, since, like you said, |
48 |
> non-multilib is multilib with only one abi :) |
49 |
> |
50 |
> (not sure if it's safe to do non-multilib when there's only one abi, it |
51 |
> may not be the default abi and need multilib_toolchain_setup) |
52 |
|
53 |
Correct. |
54 |
|
55 |
> by the way, you'd probably want _autotools-multilib_get_enabled_abis to |
56 |
> return the default ABI last with the same trick as get_all_abis so that |
57 |
> in make install the default ABI gets 'preference' as it overwrites the |
58 |
> others upon collision (it should, in theory, not matter but think |
59 |
> about installed headers that get processed by autotools) |
60 |
|
61 |
Good catch. I'll attach new patches soon ;). |
62 |
|
63 |
-- |
64 |
Best regards, |
65 |
Michał Górny |