1 |
On 03/23/2013 04:53 PM, hasufell wrote: |
2 |
> After packaging the complete humble bundle last time I noticed a bit |
3 |
> of code duplication. |
4 |
> Nothing serious, but I still think there could be a small eclass |
5 |
> making things easier. |
6 |
> |
7 |
> Especially the "remove_bundled_libs" function and the is useful IMO |
8 |
> and allows to easily remove those things. |
9 |
> The GAMES_PRESERVE_BUNDLED_LIBS array would allow to preserve libs, |
10 |
> e.g. stuff like "libbass" which is not present in the tree and even |
11 |
> GAMES_PRESERVE_BUNDLED_LIBS_amd64 (in case we have a 32bit only game |
12 |
> and there are no multilib versions for some of those libs). |
13 |
> |
14 |
> It also works well with a "bundled-libs" useflag (see |
15 |
> games-bin_src_prepare), because it can happen that binary games break |
16 |
> while the tree is moving forward with library versions, or that system |
17 |
> libraries create weird blockers for the user or even cause runtime |
18 |
> breakage (which happened with the editor of games-rpg/grimrock wrt |
19 |
> #454934). |
20 |
> |
21 |
> Most bin games use some functions from unpacker.eclass, so it is |
22 |
> inherited and used in games-bin_src_unpack by default. |
23 |
> |
24 |
> I was also thinking about some check-reqs thing, but it's probably |
25 |
> better that is handled explicitly. |
26 |
> |
27 |
> |
28 |
> If you have suggestions or if you think that approach sucks, please |
29 |
> tell me. |
30 |
> |
31 |
|
32 |
If it works, I see no reason not to commit it. It has my okay. |