| 1 |
-----BEGIN PGP SIGNED MESSAGE----- |
| 2 |
Hash: SHA1 |
| 3 |
|
| 4 |
On Friday 06 August 2004 12:52, Paul de Vrieze wrote: |
| 5 |
> What about having different eclasses for the different versions? If those |
| 6 |
> eclasses inherit a common one, this would probably be the cleanest |
| 7 |
> solution. |
| 8 |
|
| 9 |
Then a simple variable would be better, since we cannot prefix an eclass like |
| 10 |
e.g. "inherit <=kde-3.3". Both have in common that you still have more |
| 11 |
dependencies than necessary, because the minimum is to inherit kdelibs, but |
| 12 |
an application may depend on kdegraphics, which depends on kdelibs alrady. |
| 13 |
Multiple eclasses would only be dupes, imho. Best would be, if Portage would |
| 14 |
be smart enough to detect the minimum dependency set. |
| 15 |
|
| 16 |
> To introduce a set_dependencies() function would only have the |
| 17 |
> problem appear somewhere else. It still would break the cache and/or |
| 18 |
> partial parsing |
| 19 |
|
| 20 |
We'll have similar problems, when implementing a proper i18n solution. ~30 |
| 21 |
SRC_URI's in kde-i18n are not funny, k3b is another (different) sample. |
| 22 |
Either we get a function to modify such stuff or Portage has to provide |
| 23 |
decent support, but that's post .51 stuff. |
| 24 |
|
| 25 |
|
| 26 |
Carsten |
| 27 |
-----BEGIN PGP SIGNATURE----- |
| 28 |
Version: GnuPG v1.2.4 (GNU/Linux) |
| 29 |
|
| 30 |
iD8DBQFBGVuiVwbzmvGLSW8RAl43AJ9F+jPpYZOHfl6FI3Mb6iAtTtjUvQCdGR4/ |
| 31 |
GM8+MYja2JosoA1B++LTu20= |
| 32 |
=dDJV |
| 33 |
-----END PGP SIGNATURE----- |
| 34 |
|
| 35 |
-- |
| 36 |
gentoo-dev@g.o mailing list |
Archives