1 |
On 02/06/15 21:38, Mike Frysinger wrote: |
2 |
> On 02 Jun 2015 20:47, Michael Palimaka wrote: |
3 |
>> On 02/06/15 17:04, Michał Górny wrote: |
4 |
>>> Dnia 2015-06-02, o godz. 03:58:35 |
5 |
>>> "Michael Sterrett (mr_bones_)" <mr_bones_@g.o> napisał(a): |
6 |
>>>> -DEPEND="readline? ( sys-libs/readline ) |
7 |
>>>> +DEPEND="readline? ( sys-libs/readline:0 ) |
8 |
>>> |
9 |
>>> This should be actually := (or :0=) for both deps since gnugo links to |
10 |
>>> them. This also applies to your remaining 'warning silencing' commits. |
11 |
>> |
12 |
>> Why? Blindly adding the subslot dep is a bad idea. |
13 |
> |
14 |
> in this particular case, the subslot usage is what we want since we're |
15 |
> compiling+linking against it. using readline:0 vs readline is still an |
16 |
> improvement though. |
17 |
> |
18 |
> we also want a subslot on ncurses since we compile+link against it. |
19 |
> |
20 |
> i think it's pretty uncommon to use readline in a package and not want a |
21 |
> subslot. your package would have to be doing something uncommon like |
22 |
> dlopening it since the only thing readline provides is a library ... |
23 |
> -mike |
24 |
> |
25 |
|
26 |
Neither readline nor ncurses define an explicit subslot, so I don't know |
27 |
what their future meaning might be. |
28 |
|
29 |
While this is not likely to ever present a problem for ncurses or |
30 |
readline, the trend of blindly adding := to all dependencies without |
31 |
knowing what it actually means is concerning. It would be nice to have |
32 |
some information first, for example: |
33 |
|
34 |
* readline subslot will be bumped when libreadline breaks, most packages |
35 |
want the operator |
36 |
* poppler has some libraries with stable interfaces, only use the |
37 |
operator if you link against unstable libpoppler (not libpoppler-qt4) |
38 |
* libfoo has an additional private, unstable api used only by specific |
39 |
packages - don't use the operator unless you know what you're doing |