1 |
I'm deliberately starting a new thread here, because this is meandering |
2 |
well off the original topic of the global useflag proposals. |
3 |
|
4 |
On Thu, Oct 16, 2008 at 02:18:16AM +0200, Marius Mauch wrote: |
5 |
> There are also other issues, e.g. it breaks the generation of the |
6 |
> @installed package set as the installed slots can't be found in the |
7 |
> tree. As portage-2.2 makes increased use of slot atoms internally for |
8 |
> vdb handling we got a few bugreports simply due to the cache constraint |
9 |
> violation by USE=multislot. |
10 |
Ok, I found bug #174184 on this matter. |
11 |
|
12 |
kugelfang proposed to include this in EAPI=1, but I don't find it in |
13 |
there, what happened? |
14 |
|
15 |
The short-term fix of use.mask just stops current breakage, but we need |
16 |
a real solution so that we can keep using it where we actually need it. |
17 |
|
18 |
Ignoring Vapier's tirade against ciaranm there, we need the |
19 |
xDEPEND-syntax for SLOTS as the real solution, however that still |
20 |
wouldn't resolve the portion that has CTARGET as part of the SLOT, since |
21 |
metadata generated on the rsyncmaster with a different CTARGET wouldn't |
22 |
match on the clients. |
23 |
|
24 |
I'm also wondering about a bad interaction with slot dependencies even |
25 |
without having CTARGET: |
26 |
Say gcc has two variants, SLOT=3.4 and SLOT=3.4.5. |
27 |
The language in the approved version of PMS (section 9.2.5) states: "A |
28 |
specification with a named slot dependency matches only if the slot of |
29 |
the matched package is squal to ths lot specified." |
30 |
|
31 |
So if package foo/bar-1.0 contains DEPEND="sys-devel/gcc:3.4" it |
32 |
wouldn't match the other variant of gcc with SLOT=3.4.5 :-(. |
33 |
|
34 |
-- |
35 |
Robin Hugh Johnson |
36 |
Gentoo Linux Developer & Infra Guy |
37 |
E-Mail : robbat2@g.o |
38 |
GnuPG FP : 11AC BA4F 4778 E3F6 E4ED F38E B27B 944E 3488 4E85 |