| 1 |
Simon Stelling wrote: |
| 2 |
> Stephen P. Becker wrote: |
| 3 |
> |
| 4 |
>>> Using a single keyword would make us unable to mark for example |
| 5 |
>>> helixplayer (source) x86 and -amd64 at the same time (as it's now). |
| 6 |
>> |
| 7 |
>> |
| 8 |
>> |
| 9 |
>> So package.mask it in the (now hypothetical) amd64 sub-profile, and it |
| 10 |
>> is fixed. |
| 11 |
> |
| 12 |
> |
| 13 |
> That's exactly why i don't like the idea of merging keywords: You loose |
| 14 |
> the ~arch state. |
| 15 |
|
| 16 |
We weren't talking about ~arch, we were talking about -arch. |
| 17 |
|
| 18 |
> |
| 19 |
> Also, you can't compare sparc32/sparc64 to x86/amd64: sparc64 is just a |
| 20 |
> 64bit kernel with a 32bit userland. For users who want that, there is |
| 21 |
> already a keyword: x86. |
| 22 |
|
| 23 |
Wrong again. On mips, we have 64-bit kernels with *three* different |
| 24 |
possible userlands, n64, n32, and o32, and we do just fine (although as |
| 25 |
of right now, we haven't bothered to make any n64 stages since they |
| 26 |
would run slower than n32 and o32 on all of our supported hardware). |
| 27 |
|
| 28 |
-Steve |
| 29 |
-- |
| 30 |
gentoo-dev@g.o mailing list |
Archives