1 |
On 06/25/11 21:42, Maciej Mrozowski wrote: |
2 |
> On Saturday 25 of June 2011 19:29:58 Ulrich Mueller wrote: |
3 |
>>>>>>> On Sat, 25 Jun 2011, Maciej Mrozowski wrote: |
4 |
>>> Assuming package names are unique identifiers, tags are not |
5 |
>>> necessary to be available for ebuild.sh so metadata.xml is the best |
6 |
>>> place. |
7 |
>> |
8 |
>> But we know that package names are _not_ unique. There are many cases |
9 |
>> in the Portage tree where two or even more packages have the same |
10 |
>> name. Categories are there to avoid such collisions. |
11 |
> |
12 |
> But we also know, that making package names unique is first step to take as I |
13 |
> already noted in my first post in this thread. It's not that current package |
14 |
> naming scheme should be an unfixable obstacle preventing us from getting rid |
15 |
> of pointless categories (yes, every pkgmove in tree renders categories concept |
16 |
> broken by design, sorry to state this fact brutally). |
17 |
|
18 |
I disagree. If I put postgresql in x11-libs that's just wrong, and then |
19 |
you fix it with a package move. Doesn't mean the category system is |
20 |
broken, just means that it was in the wrong place at the wrong time. |
21 |
|
22 |
> |
23 |
> As far as app-xemacs is concerned (and probably why you commented here), it |
24 |
> should be sufficient to prepend "xemacs-" to package names from app-xemacs |
25 |
> category in order to make them distinguished from the rest. |
26 |
> It would be elegant and correct - after all when you "emerge ocaml" you don't |
27 |
> expect to be installing objective caml mode for Emacs, but ocaml interpreter |
28 |
> itself. |
29 |
|
30 |
Please don't do that. It leads to the kind of funny package names that |
31 |
some legacy distros carry around, which make it exquisitely hard to |
32 |
guess what their intent is. |
33 |
|
34 |
Right now "emerge openoffice" does what I want. Don't change that to |
35 |
"openoffice-core-core" or other aneurisms that can be found in the wild. |
36 |
Do not try to make me do extra work, that's rude and inconsiderate :) |
37 |
|
38 |
> |
39 |
>> With multiple overlays/repositories instead of one monolithic Portage |
40 |
>> tree, the collision issue gets even worse if you have a flat |
41 |
>> namespace. |
42 |
> |
43 |
> Every not Gentoo-based distro can live with unique package names, somehow |
44 |
> Gentoo is not able to? Colour me surprised. |
45 |
|
46 |
Most other distros also live without the option of downgrades, without a |
47 |
working package search infrastructure and without working init scripts. |
48 |
Do we want to define ourselves through the features we don't have?! |
49 |
|
50 |
And actually we do have unique package names, just that we don't |
51 |
obfuscate with random distractions but with a mostly working category |
52 |
system. So think what you are trying to fix, and no, XML is not the |
53 |
answer :) |
54 |
|
55 |
> |
56 |
> Btw, in above, I specifically proposed those unique packages to be placed in |
57 |
> ${PORTDIR}/ebuilds/ because when 'ebuilds' is considered like a fake category' |
58 |
> - existing atom syntax can be used and so can be current package manager |
59 |
> implementation (even with not entirely converted package tree, except |
60 |
> uniqueness is not checked in such case). |
61 |
> |
62 |
How about we remove slots too, that's only confusing, so you get a |
63 |
distinct unique package postgres-8.3, postgres-8.4, postgres-9.0 - and a |
64 |
meta package "postgres" that depends on any of those. As an upside we |
65 |
roughly double the amount of packages we have, and our dependencies get |
66 |
so much more ... OMG ... nooo ... what a nightmare. |
67 |
|
68 |
So again, what are you trying to fix, and what makes you think it was |
69 |
broken to start with? |
70 |
|
71 |
-- |
72 |
Patrick Lauer http://service.gentooexperimental.org |
73 |
|
74 |
Gentoo Council Member and Evangelist |
75 |
Part of Gentoo Benchmarks, Forensics, PostgreSQL, KDE herds |