| 1 |
On Sat, 10 Oct 2009 20:29:14 +0200 |
| 2 |
Ulrich Mueller <ulm@g.o> wrote: |
| 3 |
> > Different Portage versions do different things with mtimes, |
| 4 |
> |
| 5 |
> From bug 181021 I conclude that Portage preserves timestamps since |
| 6 |
> version 2.1.2.10, and 2.1.2.11 went stable in August 2007. That's more |
| 7 |
> than two years ago. |
| 8 |
|
| 9 |
Which is well after the EAPI process started, and is thus covered by |
| 10 |
the whole "no changes to behaviour on old EAPIs" thing. |
| 11 |
|
| 12 |
It's also beside the point. EAPIs are about introducing new features, |
| 13 |
so there's no reason we should go with whatever Portage happens to do |
| 14 |
currently just because it's what Portage happens to do. Instead, we |
| 15 |
should be going with "what's the best thing we can do when we |
| 16 |
introduce this new feature?", and I have yet to see an explanation as to |
| 17 |
why installing files with a 1 Jan 1970 timestamp is "the best thing we |
| 18 |
can do". |
| 19 |
|
| 20 |
-- |
| 21 |
Ciaran McCreesh |
Archives