<div dir="ltr"><div>It looks like get_sample_position should be in pipe_screen too, but you don't have to worry about that.<br><br></div>Marek<br></div><div class="gmail_extra"><br><div class="gmail_quote">On Wed, May 9, 2018 at 9:26 AM, Rhys Perry <span dir="ltr"><<a href="mailto:pendingchaos02@gmail.com" target="_blank">pendingchaos02@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">It was put in pipe_context because get_sample_position was there. It's<br>
not really a strong reason to have it there though, so I'll put it in<br>
pipe_screen.<br>
<div class="HOEnZb"><div class="h5"><br>
On Wed, May 9, 2018 at 2:37 AM, Marek Olšák <<a href="mailto:maraeo@gmail.com">maraeo@gmail.com</a>> wrote:<br>
> BTW, is there any reason for get_sample_pixel_grid to be in pipe_context? It<br>
> seems that pipe_screen is a better place, because the returned values are<br>
> immutable. pipe_context also doesn't like functions that return something,<br>
> because it limits multithreading.<br>
><br>
> Marek<br>
><br>
> On Tue, May 8, 2018 at 9:32 PM, Marek Olšák <<a href="mailto:maraeo@gmail.com">maraeo@gmail.com</a>> wrote:<br>
>><br>
>> The cso_context changes are unnecessary. It looks like that meta ops don't<br>
>> enable MSAA, so they are unaffected by sample locations, and thus no need to<br>
>> have any code for them in cso_context.<br>
>><br>
>> The names don't have to use the _state suffix, i.e. set_sample_locations.<br>
>><br>
>> I wouldn't add struct pipe_sample_locations. The whole array can be passed<br>
>> to set_sample_locations via a pointer, and NULL can mean disabled.<br>
>><br>
>> Marek<br>
><br>
><br>
</div></div></blockquote></div><br></div>