On Tue Jul 14, 2026 at 1:52 AM WEST, Jason Merrill wrote: > On 7/13/26 7:53 PM, David Malcolm wrote: >> Jason: what do you think about the approach used by this patch? >> >> Specifically, is it OK with you if we add the new langhook to >> effectively expose can_convert_eh to the analyzer? > > The approach seems reasonable, the type matching semantics aren't > otherwise available. > > But I would avoid using "may" in the name, as it suggests permission or > possibility rather than expectation, whereas this question has a certain > answer. > > I'd at least change "may" to "will". Or the analyzer > "exception_matches_type_p" name seems better yet. >
I'll resubmit it with exception_matches_type_p. Egas
