[PATCH] D27960: Use exact vector capacities to store DWARF line tables

[Adrian Prantl via Phabricator via llvm-commits]

aprantl added a comment.

For a tool like llvm-dsymutil, that reads hundreds of object files on after another and immediately discards the line table after it's finished processing one, this seems like it would always be slower, right?

What's a use-case were we would benefit from this optimization? Did you benchmark it?


https://reviews.llvm.org/D27960