[libcxx-commits] [PATCH] D156931: [libc++] Granularize most cmath includes

[Louis Dionne via Phabricator via libcxx-commits]

ldionne added a comment.

As discussed just now, I would like to understand the concrete benefit for doing this granularization. Usually, granularizing headers that are higher in the stack helps with disentangling include cycles. This is not the case here since `<cmath>` is basically its own little "submodule". Do we get any compile-time benefit when we disable transitive includes?

If we could e.g. run the test suite with transitive includes disabled and show that this change makes a difference, that would be some motivation to make this change. Otherwise, I must say I am skeptical of the benefit of splitting up a header that was as "simple" as this (I mean with few other dependencies).


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D156931/new/

https://reviews.llvm.org/D156931