[PATCH] D44313: [WebAssembly] Implement GC for imports

[Sam Clegg via Phabricator via llvm-commits]

sbc100 added a comment.

In https://reviews.llvm.org/D44313#1037653, @ruiu wrote:

> Well, I know you guys want to add the notion of "liveness" to symbols, but I think that's sematnically not correct. Symbol is essentially a label attached to something, and they are not dead or alive themselves. What we mark and garbage collect is chunks. It even feels wrong to add markLive() to a symbol is not quite right. That shouldn't belong to symbol. Can you keep it in MarkLive.cpp?


OK,  in this case we are using this information to build the imports (i.e. the dynamic symbols table) so perhaps if we go back to the idea that sections can be "Live" and symbols can be "Used", would that be ok with you?  That would match the ELF linker better perhaps?


Repository:
  rLLD LLVM Linker

https://reviews.llvm.org/D44313