Static[T] fixes #7333
Static[T] fixes #7333
Conversation
zah
force-pushed
the
nimbus
branch
2 times, most recently
from
06fd219
to
e8b273b
Compare
| @@ -1396,7 +1396,10 @@ proc semTypeNode(c: PContext, n: PNode, prev: PType): PType = | |||
| fixupTypeOf(c, prev, typExpr) | |||
| result = typExpr.typ | |||
| else: | |||
| result = semTypeExpr(c, n, prev) | |||
| if c.inGenericContext > 0 and n.kind == nkCall: | |||
There was a problem hiding this comment.
why is only nkCall checked here and not all call kinds?
There was a problem hiding this comment.
This fix turned out to be a bit tricky, due to some tricks involving the use of types and operators (as in the T -> U syntax from module for example). That's why I avoid the handling of such operators here.
All in all, I'm not particularly happy with this broad application of tyFromExpr and my first attempt was to try to introduce more logic in the compiler allowing unresolved statics to pass-through sem without executing the macros where they are used as arguments, but this turned out to be quite involved. We still need a lot of work on enabling implicit statics by default and I think these developments will allow for a nicer solution in the future.
These are some
static[T]fixes for problems discovered while wrapping and authoring fixed-size message digest cryptographic libraries at Status.