On Tue, Mar 30, 2010 at 8:02 PM, John Admanski <[email protected]> wrote:
> On Tue, Mar 30, 2010 at 2:43 PM, Martin Bligh <[email protected]> wrote:
>>
>> On Tue, Mar 30, 2010 at 2:30 PM, John Admanski <[email protected]>
>> wrote:
>> > Isn't setting the variable in os.environ kind of heavyweight...couldn't
>> > you
>> > set it as part of the utils.system command invocation? That would avoid
>> > the
>> > mess of having to save off values and adding more cleanup and in general
>> > introducing unnecessary state.
>>
>> Tests are a separate subprocess anyway, so I don't think we even need to
>> clean this up. That said, part of utils.system sounds fine too.
>>
>
> True, although I still think it's better to avoid messing with the environ
> if you don't really have to. But I'm sure that setting LC_ALL instead of
> LANG is the right thing; otherwise people who have LC_ALL or LC_TIME set
> will run into the exact same problem that this patch is supposed to address.
> And I know there are systems out there with one of those set, because when I
> went to try tweaking LANG on my system to try this out it had no effect
> because it turned out LC_TIME was defined. :)

Well, as long as the environ is being restored at the end of the test
I don't think it's too much of a hassle... anyway, the suggestions are
good, I am going to respin the patch.
_______________________________________________
Autotest mailing list
[email protected]
http://test.kernel.org/cgi-bin/mailman/listinfo/autotest

Reply via email to