[PATCH] D34667: [Demangler] [DO NOT SUBMIT] Initial patch for Microsoft demangler.

Reid Kleckner via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Wed Jun 28 14:02:23 PDT 2017


rnk added a comment.

Broadly speaking, having a general `Type` or `Node` class that forms a tree seems like the right approach. I'm sure it'll grow in complexity over time. Pre-allocating enough of them to demangle most function prototypes without memory allocation is also great. I think this is the right direction.

It might be worth completely eliminating the STL dependency so that this can be run reliably without memory allocation in a crashing process, but it's probably not worth it for now.


https://reviews.llvm.org/D34667





More information about the llvm-commits mailing list