| 1 |
On Thu, 2019-01-03 at 21:39 +0000, James Le Cuirot wrote: |
| 2 |
> Here's a series of eclass changes I've been working on since August to |
| 3 |
> allow Python modules to be cross-compiled. We previously believed this |
| 4 |
> to be practically impossible without significant changes upstream and |
| 5 |
> in the wider Python ecosystem so getting here feels like quite an |
| 6 |
> achievement. While the approach is a little unconventional, it doesn't |
| 7 |
> feel overly hacky and the results are quite consistently good. |
| 8 |
> |
| 9 |
> These changes should also benefit prefix and I did a full prefix |
| 10 |
> bootstrap for the first time this week just to make sure I didn't |
| 11 |
> break it. |
| 12 |
> |
| 13 |
> A handful of other Python packages that don't quite follow the usual |
| 14 |
> mantra have needed fixing up. I haven't included those changes here |
| 15 |
> but you can view them on this GitHub pull request. |
| 16 |
> |
| 17 |
> https://github.com/gentoo/gentoo/pull/9822 |
| 18 |
> |
| 19 |
> Once this is in place, I can finish my long-awaited revamp of my |
| 20 |
> cross-boss project that will allow you to cross-compile @system from |
| 21 |
> scratch with very little effort. |
| 22 |
> |
| 23 |
|
| 24 |
I've reviewed a few commits but I think I had enough. If you really |
| 25 |
want to push this forward, please start by merging fixes into your |
| 26 |
previous commits because it's really not useful to have commits like |
| 27 |
'fix bugs in commit 3 you've already reviewed'. |
| 28 |
|
| 29 |
Then, you should seriously think how to make things simple. Because |
| 30 |
the eclass is very complex right now, and doubling the complexity for |
| 31 |
the sake of corner case of cross isn't going work for me. The last |
| 32 |
thing we need is more hacks on top of hacks to work around previous |
| 33 |
hacks. |
| 34 |
|
| 35 |
-- |
| 36 |
Best regards, |
| 37 |
Michał Górny |
Archives