| 1 |
Dnia 2014-12-11, o godz. 11:36:29 |
| 2 |
Michał Górny <mgorny@g.o> napisał(a): |
| 3 |
|
| 4 |
> Hello, everyone. |
| 5 |
> |
| 6 |
> Following a similar change in distutils-r1, I would like to remove |
| 7 |
> the parallel run support in multilib eclasses, and effectively from |
| 8 |
> multibuild completely. |
| 9 |
> |
| 10 |
> The goal is to make things simpler, and follow PMS properly. Right now |
| 11 |
> the parallel runs imply running commands in a subshell. This has three |
| 12 |
> important implications: |
| 13 |
> |
| 14 |
> 1. variable exports and environment changes inside the implicit subshell |
| 15 |
> don't affect the outer scope -- developers are sometimes confused by |
| 16 |
> that, |
| 17 |
> |
| 18 |
> 2. parallel runs can collide if temporary files aren't named uniquely -- |
| 19 |
> think of ffmpeg, waf, |
| 20 |
> |
| 21 |
> 3. 'die' called in a subshell is unsupported by the PMS, and didn't make |
| 22 |
> it into EAPI 6. |
| 23 |
> |
| 24 |
> The side advantages are that we can get rid of multiprocessing inherit |
| 25 |
> and locking in multibuild_merge_root(). |
| 26 |
> |
| 27 |
> Possible issues: if people were relying on commands inside parallel not |
| 28 |
> to affect the outer environment (i.e. doing 'cd' and expecting the next |
| 29 |
> call to start in the initial location, or exporting variables and |
| 30 |
> expecting them to disappear), the ebuilds will fail randomly. However, |
| 31 |
> that's quite unlikely. |
| 32 |
> |
| 33 |
> Your thoughts? |
| 34 |
|
| 35 |
And committed. |
| 36 |
|
| 37 |
Not sure if we want to keep split build logs now. Their main purpose |
| 38 |
was making around parallel runs readable. However, they still allow you |
| 39 |
to easily trace build for one variant. |
| 40 |
|
| 41 |
-- |
| 42 |
Best regards, |
| 43 |
Michał Górny |
Archives