[Intel-gfx] [PATCH i-g-t] tests/gem_exec_nop: Improved test run time

[Tvrtko Ursulin]

On 04/11/15 14:46, Morton, Derek J wrote:
>>
>>
>> -----Original Message-----
>> From: Tvrtko Ursulin [mailto:tvrtko.ursulin at linux.intel.com]
>> Sent: Wednesday, November 4, 2015 2:34 PM
>> To: Chris Wilson; Morton, Derek J; intel-gfx at lists.freedesktop.org; Wood, Thomas; Gore, Tim
>> Subject: Re: [Intel-gfx] [PATCH i-g-t] tests/gem_exec_nop: Improved test run time
>>
>>
>> On 04/11/15 14:24, Chris Wilson wrote:
>>> On Wed, Nov 04, 2015 at 02:10:36PM +0000, Morton, Derek J wrote:
>>>>>
>>>>>
>>>>> -----Original Message-----
>>>>> From: Chris Wilson [mailto:chris at chris-wilson.co.uk] No, just
>>>>> android appears to be broken. I was actually complaining about the use of 'S'.
>>>>
>>>> What would you prefer to be printed?
>>>
>>> Prefer? µs. us if we must.
>>
>> Is this not just about the encoding supported by the terminal of the user running the test? In case of Android it is the adb shell them which does not support UTF-8?
>
> Actually the µs works fine on android if setlocale() is not called. The problem is that when called on android it sets a very basic locale with ascii character set.
>
> I am assuming it was added for a good reason and can't just be removed so the second best option is going to be to use 'us'.

It allows programs to respect some of the locale settings, but should 
not cause raw C strings to be modified I would have thought. So to me it 
also looks like a bug somewhere in the stack. A bit curious.

Regards,

Tvrtko