[libcxx-commits] [PATCH] D91292: [libc++] [P0935] [C++20] Eradicating unnecessarily explicit default constructors from the standard library.

Tue Dec 8 10:13:51 PST 2020

tcanens added a comment.

In D91292#2440351 <https://reviews.llvm.org/D91292#2440351>, @ldionne wrote:

> In D91292#2440252 <https://reviews.llvm.org/D91292#2440252>, @tcanens wrote:
>
>> LWG doesn't do official DRs like ever,
>
> I'm not sure I follow. Aren't all LWG issues considered DRs and "back applied" to earlier standards, hence what we do in the libraries wrt LWG issues?

LWG doesn't move issue resolutions as formal DRs the way CWG does. And not all issues are DRs - see, e.g., LWG2911 <https://wg21.link/lwg2911> aka `is_aggregate`. I think the "file an issue if you want a paper back applied" process (as in LWG3494 <https://wg21.link/lwg3494>) is a fairly new invention.

>> so cppreference policy for the library is that we look at what implementations do (or make an educated guess/prediction about what they will do) to call things "DRs". We are trying to document things for programmers, not standard archaeologists or language lawyers.
>
> I think it's misleading to call it a DR if it isn't one. I believe it would be more correct (and equally useful) to say that most implementations support it back in C++11, without using the word DR. It's a bit pedantic, but I think nothing is lost by being more precise.

I think that adds unnecessary confusion and complexity, not to mention that ISO rules say that DRs only apply to the immediately previous standard so we still have similar problems with stuff that goes back farther than that. But I don't think this is the right place to discuss cppreference policy.

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D91292/new/

https://reviews.llvm.org/D91292