| 1 |
On 26/02/17 06:58, Felix Janda wrote: |
| 2 |
> Congrats for getting it to work! |
| 3 |
> |
| 4 |
> Out of curiosity, why isn't it enough to set the CHOST (and possibly |
| 5 |
> also the subarch) to i586 in the Spec file, and use the current i686 |
| 6 |
> stage3 to bootstrap the i586 stage1? |
| 7 |
|
| 8 |
I would have though that sufficient… but I tried that some time back |
| 9 |
with some glibc-based stages, and ran into the same problems. |
| 10 |
|
| 11 |
In essence, it's the order in which some packages are built. Packages |
| 12 |
like `libtool` will build for a particular CHOST; embedding that in the |
| 13 |
files it produces. This breaks the build for `zlib`, and can break |
| 14 |
building `gcc` with nasty build errors when doing stage 2. |
| 15 |
|
| 16 |
So better to try and get a stable environment first by hand, then let |
| 17 |
catalyst rip to clean things up. |
| 18 |
|
| 19 |
I will, of course, publish the produced stages in case anyone has some |
| 20 |
old equipment they wish to repurpose with them. |
| 21 |
-- |
| 22 |
Stuart Longland (aka Redhatter, VK4MSL) |
| 23 |
|
| 24 |
I haven't lost my mind... |
| 25 |
...it's backed up on a tape somewhere. |
Archives