| 1 |
>>Both make sense, and app-editors/yudit & app-editors/yudit would |
| 2 |
>>also make sense. Instead of categories, ebuilds should have a |
| 3 |
>>CATEGORY variable. E.g. |
| 4 |
> |
| 5 |
> Portage support would be required to support deeper trees. |
| 6 |
|
| 7 |
Note that I'm not talking about trees at all. I'm talking about |
| 8 |
a flat list of ebuilds, where every ebuild belongs to several |
| 9 |
categories of different levels of specificity. |
| 10 |
|
| 11 |
> The problem |
| 12 |
> with symlinks is that we use CVS for our portage tree, |
| 13 |
|
| 14 |
I don't propose any symlinking. |
| 15 |
|
| 16 |
> I, too, would like to see support for a more heirarchical portage tree |
| 17 |
> in a future version of portage. |
| 18 |
|
| 19 |
I think a hierarchy, where every ebuild has a specific place in |
| 20 |
the tree, is a bad fit for portage or for any packaging system. |
| 21 |
One does need a taxonomy or controlled list of categories, but |
| 22 |
an ebuild can legitimately belong to many categories. |
| 23 |
|
| 24 |
As drobbins says, though, current portage code is tied to the |
| 25 |
category structure. Which is perfectly workable, really :) I |
| 26 |
just completely ignore the category, and use `emerge search` to |
| 27 |
find packages. It only bugs developers who have to decide & |
| 28 |
revise in which category to put a package. However, I'd like |
| 29 |
categories to become more useful for me as user. |
| 30 |
|
| 31 |
-- |
| 32 |
Jean Jordaan |
| 33 |
http://www.upfrontsystems.co.za |
| 34 |
|
| 35 |
|
| 36 |
-- |
| 37 |
gentoo-dev@g.o mailing list |
Archives