1 |
On 01/06/2013 06:56 AM, Diego Elio Pettenò wrote: |
2 |
> I forgot to mention that (a) is what the Ruby team has been doing up |
3 |
> to now -- it feels a bit more cumbersome in some cases, but it's |
4 |
> definitely easier to spot the problems from the start than finding |
5 |
> them months after adding the package of the tree. |
6 |
> |
7 |
> Especially if you change your mind and decide that you want to add the |
8 |
> dependency _after_ the package has been keyworded by half the arches |
9 |
> out there. |
10 |
> Diego Elio Pettenò — Flameeyes |
11 |
> flameeyes@×××××××××.eu — http://blog.flameeyes.eu/ |
12 |
> |
13 |
> |
14 |
> On Sun, Jan 6, 2013 at 1:50 PM, hasufell <hasufell@g.o> wrote: |
15 |
>> I agree with "a". |
16 |
>> A problem with "b" is: the user might install one of those "optional |
17 |
>> dependencies" later, but that will not trigger a rebuild of the other |
18 |
>> package and another run through the test phase. |
19 |
>> I would find "c" a bit confusing. |
20 |
>> |
21 |
>> The most elegant way would probably be to trigger a remerge of package |
22 |
>> a, when you want to emerge package b which is also an optional |
23 |
>> dependency of package a (in case package a has a test phase ofc). But I |
24 |
>> don't see a clean and easy way to do that. |
25 |
>> |
26 |
>> On 01/06/2013 01:28 PM, Diego Elio Pettenò wrote: |
27 |
>>> Go for a. The widest and more consistent the testing, the better. |
28 |
>>> |
29 |
>>> Otherwise the day after tomorrow you'll get a bug from me that with |
30 |
>>> $foo installed, $bar fails tests. |
31 |
>>> Diego Elio Pettenò — Flameeyes |
32 |
>>> flameeyes@×××××××××.eu — http://blog.flameeyes.eu/ |
33 |
>>> |
34 |
>>> |
35 |
>>> On Sun, Jan 6, 2013 at 11:38 AM, Michał Górny <mgorny@g.o> wrote: |
36 |
>>>> Hello, |
37 |
>>>> |
38 |
>>>> There are some Python packages which have a bunch of optional tests |
39 |
>>>> utilizing external packages. For example, the dev-python/logilab-common |
40 |
>>>> runs a few additional tests if dev-python/egenix-mx-base is installed; |
41 |
>>>> if the package is not installed, it just skips those tests. |
42 |
>>>> |
43 |
>>>> Those tests can't be really considered 'heavy' or in any way suggesting |
44 |
>>>> use of an additional USE flag. |
45 |
>>>> |
46 |
>>>> Do you believe that the ebuilds should: |
47 |
>>>> |
48 |
>>>> a) depend on all optional test dependencies conditionally to USE=test, |
49 |
>>>> therefore always requesting the widest (and consistent) testing, |
50 |
>>>> |
51 |
>>>> b) not depend on the optional test dependencies, resulting in less |
52 |
>>>> dependencies for most users but also a bit inconsistent test |
53 |
>>>> experience, |
54 |
>>>> |
55 |
>>>> c) put the optional test dependencies behind an additional USE flag? |
56 |
>>>> |
57 |
>>>> -- |
58 |
>>>> Best regards, |
59 |
>>>> Michał Górny |
60 |
>>> |
61 |
>> |
62 |
>> |
63 |
> |
64 |
This is what I have been doing with my python packages. ('A', that is.) |
65 |
|
66 |
-- |
67 |
-- Matthew Thode (prometheanfire) |