1 |
Am Dienstag 01 Mai 2012, 00:44:20 schrieb Ciaran McCreesh: |
2 |
> "Andreas K. Huettel" <dilfridge@g.o> wrote: |
3 |
> > > I'm against this one in a "quick" EAPI, unless you can get a |
4 |
> > > reference implementation and extensive testing on possible use |
5 |
> > > scenarios done in time. I strongly suspect this will end up having |
6 |
> > > the problems that REQUIRED_USE had when it was shoved in at the |
7 |
> > > last minute without anyone having properly tried it out... |
8 |
> > |
9 |
> > I cannot say much myself about the complexity of the reference |
10 |
> > implementation, however the concept itself is imho pretty |
11 |
> > straightforward and (in particular) not intrusive. |
12 |
> |
13 |
> Can you enumerate every possible way the files will be used, both in |
14 |
> terms of syntax and intended effect? |
15 |
|
16 |
In the same way as package.use.mask and package.use.force. |
17 |
|
18 |
> Can you provide assurances that it |
19 |
> can't also be (ab)used to do other things not on your list? |
20 |
|
21 |
Which list? |
22 |
|
23 |
Of course someone will come up with other creative ideas how to (ab)use it, |
24 |
that's the nature of things. (I mean, people even write other package manglers |
25 |
replacing portage... :) |
26 |
|
27 |
> Can you demonstrate that introducing this in an EAPI won't require |
28 |
> upping profile EAPIs, |
29 |
|
30 |
No. Teach me, please. |
31 |
|
32 |
An indication might however be that it acts on a package level. |
33 |
|
34 |
> and that users whose package mangler doesn't do |
35 |
> EAPI 5 won't run into problems with it? |
36 |
|
37 |
Well. PMS describes the files in a profile directory. If |
38 |
* we introduce a new file via PMS that was not in there before, |
39 |
* and another package manager accesses that file but expects different |
40 |
information there not corresponding to our new definition, |
41 |
that package manager should be considered broken because it is not adhering to |
42 |
previous PMS revisions. So? |
43 |
|
44 |
> |
45 |
> The interaction of the various use related profile things is already |
46 |
> very complicated and messy. We still haven't decided what happens when |
47 |
> use dependencies become allowed in profiles, and we're keeping profile |
48 |
> EAPIs locked below 2 so we don't have to figure it out. |
49 |
|
50 |
|
51 |
-- |
52 |
|
53 |
Andreas K. Huettel |
54 |
Gentoo Linux developer |
55 |
dilfridge@g.o |
56 |
http://www.akhuettel.de/ |