[PATCH] D36353: Postpone instantiation of constexpr functions

[Richard Smith - zygoloid via Phabricator via cfe-commits]

rsmith added inline comments.


================
Comment at: lib/Sema/Sema.cpp:189
+
+  Context.setInstantiator(new SemaTemplateInstantiator(*this));
 }
----------------
It's not correct/reasonable to instantiate whenever an expression happens to be evaluated, in numerous ways:

1) You'll get the "point of instantiation" wrong -- the standard only permits instantiation (as-if) immediately at the point of use or at the end of the translation unit
2) You risk instantiating at some points where no instantiation is permitted at all -- for instance, if SemaChecking decides it wants to try to evaluate some expression to produce a warning, and the code it tries to evaluate happens to be in an unevaluated operand, you are not permitted to instantiate things it references.
3) It risks making the set of instantiations / effective points of instantiation dependent on the warning flags that happen to be in effect, since existing code may or may not choose to evaluate an expression depending on warning flags.

Fundamentally, either constant expression evaluation needs to be side-effect free, or we need to distinguish between places where we happen to evaluate things and places where we evaluate things because the language requires them to be constant expressions. The current direction of CWG1581 is to make the "should we instantiate?" determination based solely on the syntactic context, which means that the constant expression evaluator should never trigger any instantiations.

In order to fix the case where a function/variable template specialization that is usable in a constant expression is defined after the point of use, I think we should check for pending instantiations of that function / variable template when we complete its definition, and perform those instantiations at that time.


https://reviews.llvm.org/D36353