Re: [boost] [review] outcome broken on clang/libstdc++ (Linux) and Apple clang/libc++ and gcc 7.2

Subject: Re: [boost] [review] outcome broken on clang/libstdc++ (Linux) and Apple clang/libc++ and gcc 7.2
From: Glen Fernandes (glen.fernandes_at_[hidden])
Date: 2018-01-28 00:20:51

On Sat, Jan 27, 2018 at 7:15 PM, degski wrote:
> Sure you can mention it, but stating the obvious, the compiler shouldn't
> crash, but should instead tell you what's wrong with the code (if anything
> *is* actually wrong with the code, as it ICE-es we don't know, the only
> thing we do know is that gcc-7.2 and lower won't handle it at all)...

Sure. It would be nice if compilers didn't have defects at all. Be
they defects which result in an ICE, or defects that result in
incorrect or sub-optimal code being generated, failure to conform to
the specification, defective implementation of certain language or
library features.

One of the things I appreciated about Boost when I came to know Boost
libraries more intimately when we started using them at Microsoft, was
that they compensated for those defects (to the point where we would
use Boost equivalents even instead of our C++ implementation's
standard library facilities).

Glen