[Piglit] [PATCH 4/8] shader_runner: Add a non-relative, rgba variant of relative probe rect rgb.

Eric Anholt eric at anholt.net
Fri Apr 10 13:32:34 PDT 2015


Kenneth Graunke <kenneth at whitecape.org> writes:

> On Friday, August 29, 2014 02:41:15 PM Eric Anholt wrote:
>> ---
>>  tests/shaders/shader_runner.c | 13 +++++++++++++
>>  1 file changed, 13 insertions(+)
>> 
>> diff --git a/tests/shaders/shader_runner.c b/tests/shaders/shader_runner.c
>> index 717be42..2c350cf 100644
>> --- a/tests/shaders/shader_runner.c
>> +++ b/tests/shaders/shader_runner.c
>> @@ -2221,6 +2221,19 @@ piglit_display(void)
>>  			if (!piglit_probe_pixel_rgb(x, y, &c[2])) {
>>  				pass = false;
>>  			}
>> +		} else if (sscanf(line, "probe rect rgba "
>> +				  "( %f , %f , %f , %f ) "
>> +				  "( %f , %f , %f , %f )",
>> +				  c + 0, c + 1, c + 2, c + 3,
>> +				  c + 4, c + 5, c + 6, c + 7) == 8) {
>> +			x = c[0];
>> +			y = c[1];
>> +			w = c[2];
>> +			h = c[3];
>> +
>> +			if (!piglit_probe_rect_rgba(x, y, w, h, &c[4])) {
>> +				pass = false;
>> +			}
>>  		} else if (sscanf(line, "relative probe rect rgb "
>>  				  "( %f , %f , %f , %f ) "
>>  				  "( %f , %f , %f )",
>> 
>
> I don't think %f makes much sense here - you're reading floats,
> converting them to ints, then passing them to a function expecting
> ints.  For relative probes, it makes sense.
>
> I think you should do:
>
> 		} else if (sscanf(line, "probe rect rgba "
> 				  "( %d , %d , %d , %d ) "
> 				  "( %f , %f , %f , %f )",
> 				  &x, &y, &w, &h,
> 				  c, c + 1, c + 2, c + 3) == 8) {
>
> 			if (!piglit_probe_rect_rgba(x, y, w, h, c)) {
> 				pass = false;
> 			}
> 		}
>
> With that changed,
> Reviewed-by: Kenneth Graunke <kenneth at whitecape.org>

Nope, that seems obviously good.  And it might have caught an earlier
bug in my code generation :)  Thanks!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 818 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/piglit/attachments/20150410/1a12c205/attachment.sig>


More information about the Piglit mailing list