| 1 |
> This is the general trend that bug #12499 is supposed to address. We need |
| 2 |
> a global server/client flag to allow individual ebuilds to disable parts of |
| 3 |
> packages (i.e. when USE=-server, I only want the mysql client installed, |
| 4 |
> not full server. There are many ebuilds that would benefit from this, as |
| 5 |
> is discussed in the bug.) |
| 6 |
> |
| 7 |
> Personally, I think it'd be best to get #12499 in place than to add local |
| 8 |
> use flags on individual packages. |
| 9 |
> |
| 10 |
> Thoughts? |
| 11 |
|
| 12 |
This may indeed be a better solution but my plan was not to remove |
| 13 |
server support from the binary but just all the stuff specifically added |
| 14 |
by the ebuild, so someone could still run a server with his custom |
| 15 |
settings (and people already running a server won't need that stuff), |
| 16 |
but for all people except those who run a cvs pserver _and_ use the |
| 17 |
config supplied by the ebuild (including a new uid/gid etc.) it would |
| 18 |
just be a burden. |
| 19 |
|
| 20 |
The benefit of disabling server things in the binary is virtually zero |
| 21 |
IMHO, my point is about keeping the system more leightweight, so we |
| 22 |
don't force the user to have all that stuff. |
| 23 |
|
| 24 |
That bug you mentioned perhaps has a far bigger impact and may still |
| 25 |
need to have some things worked out, that's why I'd still like to have a |
| 26 |
cvspserver local use flag, at least temporary (if another [better] |
| 27 |
solution pops in later). |
| 28 |
|
| 29 |
It's not that I want to add the local flag by all means, personally I'd |
| 30 |
just say "it would be better". |
| 31 |
|
| 32 |
-- |
| 33 |
Rainer |
| 34 |
|
| 35 |
|
| 36 |
-- |
| 37 |
gentoo-dev@g.o mailing list |
Archives