| 1 |
On Sun, 27 Jan 2013 09:44:23 -0300 |
| 2 |
Alexis Ballier <aballier@g.o> wrote: |
| 3 |
|
| 4 |
> On Sun, 27 Jan 2013 12:21:01 +0100 |
| 5 |
> Michał Górny <mgorny@g.o> wrote: |
| 6 |
> |
| 7 |
> > The installed headers are supposed not to change between ABIs. If they |
| 8 |
> > do, we need to do something special about them or everything is going |
| 9 |
> > to end up real bad. |
| 10 |
> > |
| 11 |
> > Therefore, do a checksum of headers installed in /usr/include after |
| 12 |
> > each ABI's src_install() and die if they don't match. |
| 13 |
> |
| 14 |
> sounds a bit restrictive IMHO, but such a kind of check is probably |
| 15 |
> necessary and has to be added before the eclass is widespread. |
| 16 |
> you could probably add a way to disable those checks for the |
| 17 |
> hypothetical case there is a false positive (ie: yes the headers |
| 18 |
> differ but it doesn't matter). |
| 19 |
|
| 20 |
Well, since this is eclass and not the PMS, we can add it anytime when |
| 21 |
it becomes really necessary ;). |
| 22 |
|
| 23 |
-- |
| 24 |
Best regards, |
| 25 |
Michał Górny |
Archives