On 09/27/2011 03:08 PM, Eric Anholt wrote: > Bitshifts are one of the rare places that GLSL allows mixed base types > without an implicit conversion occurring. > --- > src/glsl/ir_constant_expression.cpp | 4 +++- > 1 files changed, 3 insertions(+), 1 deletions(-) > > diff --git a/src/glsl/ir_constant_expression.cpp > b/src/glsl/ir_constant_expression.cpp > index f0299a2..56a4237 100644 > --- a/src/glsl/ir_constant_expression.cpp > +++ b/src/glsl/ir_constant_expression.cpp > @@ -69,7 +69,9 @@ ir_expression::constant_expression_value() > } > > if (op[1] != NULL) > - assert(op[0]->type->base_type == op[1]->type->base_type); > + assert(op[0]->type->base_type == op[1]->type->base_type || > + this->operation == ir_binop_lshift || > + this->operation == ir_binop_rshift); > > bool op0_scalar = op[0]->type->is_scalar(); > bool op1_scalar = op[1] != NULL && op[1]->type->is_scalar();
It seems weird that we allow this in the IR for shifts, but for nothing else. I'm pretty sure that -all- other binary operations require their operand types to be equal. Might be more consistent to make AST->HIR generate i2u or u2i nodes to convert op[1] to op[0]'s type. Though, those are silly, and this ought to work fine too. Certainly the existing code is broken, your new code is not broken, and ir_validate allows mixed types in the IR. *shrug* Reviewed-by: Kenneth Graunke <kenn...@whitecape.org> _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
