================
@@ -3761,13 +3761,17 @@ Decl
*TemplateDeclInstantiator::VisitNonTypeTemplateParmDecl(
if (IsExpandedParameterPack)
Param = NonTypeTemplateParmDecl::Create(
SemaRef.Context, Owner, D->getInnerLocStart(), D->getLocation(),
- D->getDepth() - TemplateArgs.getNumSubstitutedLevels(),
+ D->getDepth() - (TemplateArgs.retainInnerDepths()
+ ? 0
+ : TemplateArgs.getNumSubstitutedLevels()),
D->getPosition(), D->getIdentifier(), T, TSI,
ExpandedParameterPackTypes, ExpandedParameterPackTypesAsWritten);
else
Param = NonTypeTemplateParmDecl::Create(
SemaRef.Context, Owner, D->getInnerLocStart(), D->getLocation(),
- D->getDepth() - TemplateArgs.getNumSubstitutedLevels(),
+ D->getDepth() - (TemplateArgs.retainInnerDepths()
+ ? 0
+ : TemplateArgs.getNumSubstitutedLevels()),
----------------
zyn0217 wrote:
Ugh, I don't think I have any tests? Some of our longstanding issues were
actually fixed by the 'wrong' preservation patch, which IIRC there should have
been cases where lambdas are nested within a template parameter type?
That said, I'm still unclear what we should do to move forward: do you consider
it necessary to convert everything to Rewrite, at the risk of rewriting almost
everything since #161671? Note that some of our cases are dependent on your
several NTTP patches, which I think it already makes the situation worse to
move forward to that direction.
I agree it's a bit arbitrary to introduce the third mode flag outside of that
enumeration, and we could always refactor it.
Again @cor3ntin for opinions
https://github.com/llvm/llvm-project/pull/210381
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
