1 |
On 9/19/15 8:56 AM, Ian Stakenvicius wrote: |
2 |
> Sent from an iPhone, sorry for the HTML... |
3 |
> |
4 |
> On Sep 19, 2015, at 8:31 AM, Vadim A. Misbakh-Soloviov <mva@×××.name> wrote: |
5 |
> |
6 |
>>> So, if an arch developer tests the package(s) on one architecture, he is |
7 |
>>> allowed to stabilize/keyword for all. |
8 |
>> And how about the |
9 |
>>> some arches rquires additional tests during stabilization, like so: mips*, |
10 |
>> arm*, and some more exotic ones |
11 |
>> definition in developer manuals? :) |
12 |
>> |
13 |
>> -- |
14 |
>> Best regards, |
15 |
>> mva |
16 |
> The ALLARCHES keyword is meant for stuff that is essentially platform-agnostic. Python code shouldn't act any different on arm than arm64 than amd64. Using that keyword on something like ffmpeg would be a bad use of the keyword. |
17 |
|
18 |
Even with python where, for example, `import ctypes' might trigger |
19 |
something on one arch and not another. Anyhow, thanks for the reminder |
20 |
ago. I see that keyword as "risky" but an unfortunate risk we have to |
21 |
take because we are short of manpower and don't have automated arch |
22 |
testing. I say use it where it appears reasonable (eg interpreted |
23 |
script and not compiled code) but be ready that every once in a while |
24 |
we'll get bitten. |
25 |
|
26 |
-- |
27 |
Anthony G. Basile, Ph.D. |
28 |
Gentoo Linux Developer [Hardened] |
29 |
E-Mail : blueness@g.o |
30 |
GnuPG FP : 1FED FAD9 D82C 52A5 3BAB DC79 9384 FA6E F52D 4BBA |
31 |
GnuPG ID : F52D4BBA |