GafferOSL : ShadingEngine - cast doubles to float or int if required #2547
Conversation
src/GafferOSL/ShadingEngine.cpp
Outdated
| @@ -294,6 +294,22 @@ class RenderState | |||
| t.arraylen = 2; | |||
| return ShadingSystem::convert_value( value, type, src, t ); | |||
| } | |||
| else if ( it->dataView.type.basetype == TypeDesc::DOUBLE ) | |||
There was a problem hiding this comment.
This would accept a V3d as well as just a double, right? Is that intended? I don't think we support a V3f->float conversion because it's ambiguous, and I would expect the same to apply here.
There was a problem hiding this comment.
checked we're dealing with a scalar double in ba245a1
src/GafferOSL/ShadingEngine.cpp
Outdated
| else if ( it->dataView.type.basetype == TypeDesc::DOUBLE ) | ||
| { | ||
| double doubleValue; | ||
| ShadingSystem::convert_value( &doubleValue, TypeDesc::DOUBLE, src, TypeDesc::DOUBLE ); |
There was a problem hiding this comment.
This is fairly performance-critical code, so I think it's probably worthwhile preferring static_cast<const double *>( src ) over convert_value() here.
There was a problem hiding this comment.
favoured casts over function call in ba245a1
andrewkaufman
added
the
pr-revision
donboie
force-pushed
the
ShadingEngineDouble
branch
from
3dd9262
to
ba245a1
Compare
|
Perhaps we need to rewrite the Cortex -> OSL type conversion to provide a wider range of conversions? I've largely been responsible for adding special cases and perhaps we have enough examples where we can replace all these ad-hoc conversions with something more systematic. |
andrewkaufman
removed
the
pr-revision
|
Thanks!
I think the ideal might be for |
No description provided.