[libc-dev] Possibility of llvm-libc use in Chrome?

[Ebrahim Byagowi via libc-dev]

Three years ago when Chrome switched <https://crbug.com/801780> to llvm's
STL even on Windows, the switch resulted in many advantages such as better
crash reports as better call stack I think, being able to use new STL
features consistently, enabling more of runtime sanitizers, etc.

When llvm libc project was announced I left this comment
<https://twitter.com/ebraminio/status/1143628963106787328> in Twitter which
ofc a full self contained libc won't be a possibility without having things
as WASI but that's a different matter, yet, having access to more of libc
perhaps can be useful to embed system of different kind.

Considering those, is llvm-libc use for Chrome on all its targets something
expected to happen? I know the project is expected to be something stable
in two years just that maybe my question helps on better planning for the
project.

And as a side question, do you think a renterant qsort, qsort_r, is a
possibility on llvm's libc? At least that can be used in BoringSSL
<https://source.chromium.org/chromium/chromium/src/+/master:third_party/boringssl/src/crypto/stack/stack.c;l=368-376?q=qsort_r&ss=chromium>
and llvm's STL
<https://source.chromium.org/chromium/chromium/src/+/master:third_party/swiftshader/third_party/llvm-subzero/include/llvm/ADT/STLExtras.h;l=423-432?q=qsort_r&ss=chromium>.
Maybe if there was a __builtin_qsort and __builtin_qsort_r with the
capability those projects could benefit from it even sooner. Ofc that alone
maybe isn't a good reason to make a popular library tied to a specific
compiler but maybe as tiny motivation.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/libc-dev/attachments/20210201/53e6cff0/attachment.html>