| 1 |
Hi Marius |
| 2 |
|
| 3 |
On Tue, Nov 04, 2003 at 12:20:17AM +0100, Marius Mauch wrote: |
| 4 |
[...] |
| 5 |
>So my idea (as well as others) is to add a new "server" USE flag that |
| 6 |
>controls wether the server part of a package is build. This idea is |
| 7 |
>outlined in more details on bug 12499, which also had the idea of a |
| 8 |
>"client" USE flag, so that it's possible to only build the libs or only |
| 9 |
>the server, however I think that such a flag will create more problems |
| 10 |
>than benefits (see the bug for explanations). |
| 11 |
|
| 12 |
I do not support this. Debian's SAMBA has at least subpackages for |
| 13 |
- common |
| 14 |
- doc |
| 15 |
- smbclient |
| 16 |
- winbind |
| 17 |
- swat |
| 18 |
- smbfs |
| 19 |
- python |
| 20 |
- pam |
| 21 |
|
| 22 |
which is nuts. Users dont know what to install to get what they need. |
| 23 |
|
| 24 |
I recall looking at mod_php (or maybe it was php) subpackages for another |
| 25 |
distribution, and there were _dozens_ of them. Insane in the membrane. |
| 26 |
|
| 27 |
I like the way Gentoo traditionally works; you emerge something and you |
| 28 |
get it all, with some knobs to control build-time dependencies. |
| 29 |
|
| 30 |
Further, you are overloading the intended function of USE variables. |
| 31 |
Instead of controlling optional build-time functionality, now you |
| 32 |
are abusing them to control optional install-time bits. |
| 33 |
|
| 34 |
It is not natural to stop at "client" and "server" flags either. |
| 35 |
What about "dev" for .a and .h things? This is really going down |
| 36 |
the slipperly slope in my opinion. |
| 37 |
|
| 38 |
Donny |
Archives