[PATCH] D120471: [Bitcode] Fully support opaque pointer auto upgrade

[Nikita Popov via Phabricator via llvm-commits]

nikic added a comment.

In D120471#3357501 <https://reviews.llvm.org/D120471#3357501>, @aeubanks wrote:

> sorta makes sense, but I'm not understanding in what case we'd create a new virtual type id, can you give an example?

For example, if you have a function declaration, the bitcode might encode the function type (and have a type ID for it), but the value itself would be a pointer to a function type, for which there is either no ID, or we don't know about it at this point.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D120471/new/

https://reviews.llvm.org/D120471