[PATCH] D47073: Document and Enforce new Host Compiler Policy

JF Bastien via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Tue Jan 15 21:44:38 PST 2019


jfb added a comment.

> No: the RFC does not include this discussion. If you want to claim this, then please source it precisely.

This patch's description links to: http://lists.llvm.org/pipermail/llvm-dev/2018-May/123238.html

Which says:

> The 3-years/1.5 years would result in our minimum GCC/Clang becoming: GCC5.1/Clang3.6.  We would WARN on anything older than GCC7.1/Clang3.8

That 3 / 1.5 cutoff was discussed on this patch, e.g. http://lists.llvm.org/pipermail/llvm-dev/2018-May/123249.html
and came from this discussion: http://lists.llvm.org/pipermail/llvm-dev/2018-May/123182.html
It started off as "which compilers give us the features we want" and modified itself into "for now we get what we want with a 3 year lookback, let's use that as a hand wavy policy going forward".

So I'd say that it was sourced quite precisely, and re-listing the entire discussion will lose all the bikeshedding minutia which led to where we are. At this point Erich is patiently trying to add nuance on top of nuance to a warning about a policy, pertaining to a discussion that's been going on for 7 months... So yes I totally see where the frustration comes from, all of that to migrate away from C++11 which has overwhelming support. I think it's critical to hard-warn (i.e. error with opt-out option) people in the LLVM 8 timeframe. Sure we can cherry-pick a patch, but the later we do so the more potential hurt there is. I'd really rather not shoulder that hurt, because I think it's better to just tell people compiling LLVM 8 now: we're going to move away from older compilers, get ready.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D47073/new/

https://reviews.llvm.org/D47073





More information about the llvm-commits mailing list