1 |
this is something we discussed about a month ago and never developed an |
2 |
official policy for then did we? george can you comment on this? |
3 |
perhaps its time we do |
4 |
|
5 |
dave |
6 |
|
7 |
On Wed, 2003-05-28 at 15:04, Matthew Kennedy wrote: |
8 |
> Jon Portnoy <avenj@g.o> writes: |
9 |
> |
10 |
> > Policy is to avoid naming conflicts. :) |
11 |
> > |
12 |
> > There are indeed ebuilds that conflict. Those are bugs that there is |
13 |
> > currently no clean fix for. Committing _more_ conflicting ebuilds is |
14 |
> > definitely not okay. |
15 |
> > |
16 |
> |
17 |
> Jon (and others), |
18 |
> |
19 |
> I'm not so sure. I have app-emacs and app-xemacs and there are |
20 |
> *many* ebuilds which "conflict" in their naming, and this is |
21 |
> intentional. |
22 |
> |
23 |
> For example, the "ilisp" package appears in both app-emacs and |
24 |
> app-xemacs. So if I'm to avoid this naming "conflict" because |
25 |
> somemone typing "emerge ilisp" might get "app-xemacs/ilisp" instead |
26 |
> of "app-emacs/ilisp", then naturally I should prefix the ebuild names |
27 |
> with "emacs" and "xemacs", right? So I'd end up with |
28 |
> |
29 |
> emerge emacs-ilisp |
30 |
> or |
31 |
> emerge xemacs-ilisp |
32 |
> |
33 |
> But wait! Thats what categories are for -- ie. defining some |
34 |
> namespace. Thus there's no need, and this more or less backs up my |
35 |
> argument that the category name + package name forms a "fully resolved |
36 |
> and unique" symbol for emerge to act upon. |
37 |
> |
38 |
> So the user should do: |
39 |
> |
40 |
> emerge app-emacs/ilisp |
41 |
> or |
42 |
> emerge app-xemacs/ilisp |
43 |
> |
44 |
> What of the case where the user is surprised by the "short form" case: |
45 |
> |
46 |
> emerge ilisp |
47 |
> |
48 |
> ...well this is where emerge should abort and report back to the user |
49 |
> something like this: "ERROR: Short form 'ilisp' matches |
50 |
> app-emacs/ilisp and app-xemacs/ilisp -- please use a fully qualified |
51 |
> name instead". (Currently emerge emerges the first one it sees.) |
52 |
> |
53 |
> I really think ambiguities should be resolved by using the category |
54 |
> name -- otherwise I can't see why we sort packages into categories |
55 |
> at all. |
56 |
> |
57 |
> Matt |
58 |
-- |
59 |
Dave Nellans |
60 |
http://lucy.wox.org/~dnellans/ |
61 |
dnellans@×××××××.edu |