On 14/09/2015 18:51, Jose Fonseca wrote :
I don't know if there's a strict policy.
From my POV, if people actually bother implementing it, then yeah, it
would make a lot of sense in keeping it, since as you say it can't be
easily emulated.
But if the state is just lying there, being ignored by practically
everybody, then there's little point for its existence.
After all, it's trivial to add it back if people truly want it. And
you'd probably keep ignoring it if it weren't my proposal to remove it.
As I said in on the D3D9 state tracker comment, I don't know of any
application where ignoring the state causes any visible artifacts.
In short, from my POV, "making sense" just by itself is not sufficient
to keep stuff around. It should actually be used and implemented.
And AFAIK, the only way anybody to actually use
pipe_rasterizer_state::line_last_pixel would be:
- st/nine + ilo
- st/nine + svga on old VMware Workstation on Windows
And I strongly suspect the odds of either are quite small.
But I don't care much either way.
Jose
Given the current state of the support, and the fact it's isn't really
used by apps either, I'm ok removing it.
Patch 2 should probably remove ilo logic to handle giq_last_pixel as well,
and I would have liked patch 4 to add a comment in the code (so that
people don't try reimplementing the feature).
But there can be argument in favor keeping the logic in patch 2 (means
the feature can be reimplemented more easily), and I can add myself the
comment in nine code.
Thus the series is:
Reviewed-by: Axel Davy <axel.d...@ens.fr>
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
