Michael Buch 1c41b9a81f [lldb][TypeSystemClang] Don't treat float vector types as floating point types (#179213) ...

This patch re-implements `TypeSystemClang::IsFloatingPointType` by
forwarding it to `clang::Type::isFloatingType`. The main difference is
that the latter returns false for float vector types. The motovation
here (apart from implementation simplicity) is that this is currently a
foot-gun because most callsites probably don't consider treating float
vector types. Callers should test for vectors explicitly using
`IsVectorType` (or use `GetTypeInfo`).

This patch makes all the callers of `IsFloatingPointType` now check
`GetTypeInfo() & eTypeIsFloat`. This is set for float vector types too,
so behaviour doesn't change.

To make sure we audit all the call-sites in `ValueObject.cpp`, I added a
helper `HasFloatRepresentation` (named after the
`clang::Type::hasFloatingRepresentation` API), which does the
`GetTypeInfo` check, and added a FIXME to it.

2026-02-03 20:27:23 +00:00

138 KiB

Raw Blame History

View Raw