1 |
Marco Matthies posted <442A6C24.2090801@×××.net>, excerpted below, on |
2 |
Wed, 29 Mar 2006 13:14:44 +0200: |
3 |
|
4 |
> Bertrand Jacquin wrote: |
5 |
>> The first exemple I wrote was in reality : |
6 |
>> CFLAGS="-m32" emerge -avt mozilla-firefox |
7 |
>> |
8 |
>> To be clear : I don't want a chroot, because it make be 2 gentoo to |
9 |
>> maintain and a lot of things unecessary ATM. |
10 |
>> I would like portage build for me a software in 32 bit mode. |
11 |
>> If it's a lib, I would like portage to install it in /emul/linux/x86 |
12 |
>> if I tell him to do that |
13 |
> |
14 |
> Sorry, i missed the part about you not wanting a chroot. As far as I |
15 |
> know, portage currently does not support installing 32-bit and 64-bit |
16 |
> versions of the same package, i.e. it does not know anything about the |
17 |
> bitness a package was compiled for and therefore doesn't use this |
18 |
> information for dependency resolution. In other words, it will not know |
19 |
> that the X11 libs it installed are 64-bit and that the firefox you want |
20 |
> to compile for 32-bits needs the 32-bit X11 libs. |
21 |
|
22 |
That's correct. You /will/ eventually trash your system trying to do |
23 |
this, as portage will get hopelessly confused trying to merge stuff with |
24 |
dependencies it /thinks/ you have already merged, only you don't, because |
25 |
you merged them in the other bitness. |
26 |
|
27 |
The easiest way to keep portage from getting confused is to run two |
28 |
separate instances of it that don't know about each other. The way you do |
29 |
that is to run a 32-bit chroot. Yes, it /does/ require a certain amount |
30 |
of duplication, but if you use a stage-3 x86 install and the binary |
31 |
packages CD, it's not /too/ bad. You lose a bit of customization going |
32 |
pre-built binaries, but it's a trade-off between that and the time to |
33 |
compile all that stuff twice. You can always remerge the specific |
34 |
packages you want to. |
35 |
|
36 |
At some point in the future, likely with the ongoing full rewrite |
37 |
project rewrite that's very possibly a year or two away from release, |
38 |
altho the feature may well be backported some time before then, it's |
39 |
planned that portage will have a multi-bitness deps tracker and resolver. |
40 |
However, that's a /long/ way off. Meanwhile, you can either go 32-bit |
41 |
only, 64-bit only, or run a multilib system. If you run a multilib |
42 |
system, you can choose limited 32-bit support based on the 32-bit binary |
43 |
compatibility libs, or full 32-bit support based on a chroot, with all the |
44 |
time necessary to maintain it balanced against the better 32-bit support. |
45 |
|
46 |
Personally, I don't like closed source slaveryware in any case, and |
47 |
basically won't have it on my machine. If I was content being someone's |
48 |
slave because they don't respect me as a user enough to give me my rights, |
49 |
I could have stayed on MSWormOS. After all, I left a decade of knowledge |
50 |
behind when I switched, and I /could/ have just stayed where I was. I |
51 |
found it worth my time to switch, and there's no way I'm going to consent |
52 |
or allow myself to be tied down to unfreedomware now, or why did I switch |
53 |
in the first place? That makes the 32-bit/64-bit software issue much |
54 |
easier, since the non-marginal freedomware was ported a long time ago and |
55 |
has been available in 64-bit native for soome time. It's only the |
56 |
marginal stuff, and closed source slaveryware that I couldn't/wouldn't run |
57 |
anyway, that's 32-bit only, now. |
58 |
|
59 |
Of course, I'm mature enough to realize not everyone holds my values, |
60 |
neither do I expect them to, so I recognize that those that choose to run |
61 |
what I consider slaveryware can make that choice, just as I made mine. |
62 |
Just don't ask me to take part in it. |
63 |
|
64 |
-- |
65 |
Duncan - List replies preferred. No HTML msgs. |
66 |
"Every nonfree program has a lord, a master -- |
67 |
and if you use the program, he is your master." Richard Stallman in |
68 |
http://www.linuxdevcenter.com/pub/a/linux/2004/12/22/rms_interview.html |
69 |
|
70 |
|
71 |
-- |
72 |
gentoo-amd64@g.o mailing list |