[libcxx-commits] [PATCH] D99789: [libc++] Add fallback standard flags and normalize corresponding feature.

[Louis Dionne via Phabricator via libcxx-commits]

ldionne added a comment.

In D99789#2679958 <https://reviews.llvm.org/D99789#2679958>, @Quuxplusone wrote:

> @ldionne wrote:
>
>> IMO we don't want to allow people to pass --param std=c++2a to Lit. We want to use --param std=c++20
>
> Unfortunately that doesn't scale. Today we //force// people to use `--param std=c++2a` (because this patch hasn't been committed); and we also //force// people to use `--param std=c++2b` (because C++2b hasn't got an official number yet).  We don't want to force everyone to have a "flag day" where they all simultaneously get forced over from `std=c++2a` to `std=c++20`; we want them to be able to adjust their scripts and stuff on their own time. So, since we need to support `std=c++2a` as a synonym for `std=c++20` here anyway at the compiler level, I think it's quite correct to just reuse that list of synonyms for argument parsing as well. Which is what this patch does.
>
> TLDR: This morning I ran `./bin/llvm-lit --param std=c++2a ../libcxx/test/`. I want to run that same command tomorrow morning.

I disagree - this is just the interface for running Lit from the command line, I think it's entirely fine if there's a "flag" day, and even if it's not official. There are some things that we need to create engineering solutions for in order not to break people, but I don't think that's one. I think pretty much the only places where there's `--param std=c++2a` is going to be in the CI scripts (which can be updated in lockstep with this patch), and in the libc++ developers' shell command history or local helper functions. I think it's fine if they just need to change it to `--param std=c++20` at some point.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D99789/new/

https://reviews.llvm.org/D99789