1 |
I need people to test portage 2.0.48-r6. We have some |
2 |
fairly hefty changes to *DEPEND stuff that is going to be |
3 |
(actually, it already is) breaking quite a few packages. |
4 |
|
5 |
End-users should be ok once we get this stable and onto rsync1 |
6 |
so that metadata gets fixed, but I need some really heavy |
7 |
testing with everything you can throw at it. /msg me _anything_ |
8 |
you think is hoky and I'll look-at/fix it and/or tell you it's |
9 |
right. |
10 |
|
11 |
Ebuild developers and cvs tree users will be affect by this as |
12 |
metadata isn't particularly right with any version of portage |
13 |
prior to this one, but the changes are very noticable right now. |
14 |
|
15 |
** ** POST BUGS ** ** if I don't respond. |
16 |
|
17 |
Things that have changed: |
18 |
|
19 |
All DEPEND variables are now treated incrementally in ebuilds |
20 |
and eclasses. Neither should concern themselves with the others. |
21 |
Preferably, use DEPEND="${DEPEND} new/stuff" type syntaxes within |
22 |
eclasses if you define variables like that. |
23 |
|
24 |
newdepend is now fixed to properly set up variables for eclasses |
25 |
AND ebuilds and should be freely usable to do so. *** The function |
26 |
of newdepend has changed slightly. *** It no longer defines RDEPEND |
27 |
for anything except '/c' and '/autotools'. Use 'newrdepend' to |
28 |
define it instead. 'newcdepend' and 'newpdepend' are also available. |
29 |
|
30 |
Watch the dependencies that are being put into the cache on your |
31 |
boxes. Make sure things seem right. If you see an oddity or an |
32 |
action that you have to take to get proper results, inform me, |
33 |
post a bug, and see what else it affects if you can. |
34 |
|
35 |
--NJ |
36 |
|
37 |
|
38 |
-- |
39 |
gentoo-dev@g.o mailing list |