[PATCH] D152891: [ADT] Add deduction guide for iterator_range

[Balázs Benics via Phabricator via llvm-commits]

steakhal updated this revision to Diff 532384.
steakhal added a comment.

- Factor out ADl utilities
- Add SFINAE check for iterator_range constructor

I measured the clean build time before and after the patch on my Ubuntu 22.04.1 system.
I used `gcc (Ubuntu 11.3.0-1ubuntu1~22.04.1) 11.3.0` with the  `mold` linker.

Baseline:

  time cmake -S llvm -B build/rel-old -DLLVM_ENABLE_PROJECTS="clang;clang-tools-extra" -DCMAKE_BUILD_TYPE=Release
  real    0m8.446s
  user    0m6.587s
  sys     0m1.691s
  
  time ninja -C build/rel-old/
  [5445/5445]
  real    19m23.414s   1163.414s
  user    431m9.467s
  sys     18m20.037s

With this patch:

  time cmake -S llvm -B build/rel-new -DLLVM_ENABLE_PROJECTS="clang;clang-tools-extra" -DCMAKE_BUILD_TYPE=Release
  real    0m8.943s
  user    0m6.956s
  sys     0m1.818s
  
  time ninja -C build/rel-new/
  [5445/5445]
  real    19m32.817s   1172.817s
  user    433m30.939s
  sys     18m20.295s

This means that we would have `0.808224759` % compile time regression with this patch. Is it acceptable?
(I only run the measurement once)


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152891/new/

https://reviews.llvm.org/D152891

Files:
  llvm/include/llvm/ADT/ADL.h
  llvm/include/llvm/ADT/STLExtras.h
  llvm/include/llvm/ADT/iterator_range.h

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D152891.532384.patch
Type: text/x-patch
Size: 8839 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20230617/fb904237/attachment.bin>