Please revert commit ceaca7bec6adf5735d51c1ca7efea5a2e05f470e

Please revert commit ceaca7bec6adf5735d51c1ca7efea5a2e05f470e

[Philipp Stephani]

[-- Attachment #1: Type: text/plain, Size: 149 bytes --]

This doesn't work because noexcept is not part of a function's type. See
the discussion in https://github.com/aaptel/emacs-dynamic-module/issues/27.

[-- Attachment #2: Type: text/html, Size: 251 bytes --]

Re: Please revert commit ceaca7bec6adf5735d51c1ca7efea5a2e05f470e

[Paul Eggert]

Philipp Stephani wrote:
> This doesn't work because noexcept is not part of a function's type.

That was fixed long ago in emacs-25, in commit 
7210f5130abb909aebcbb09a8a71558100fb9f96. Why are you still running into the 
problem, exactly?

Re: Please revert commit ceaca7bec6adf5735d51c1ca7efea5a2e05f470e

[Philipp Stephani]

[-- Attachment #1: Type: text/plain, Size: 695 bytes --]

Paul Eggert <eggert@cs.ucla.edu> schrieb am So., 10. Jan. 2016 um 20:51 Uhr:

> Philipp Stephani wrote:
> > This doesn't work because noexcept is not part of a function's type.
>
> That was fixed long ago in emacs-25, in commit
> 7210f5130abb909aebcbb09a8a71558100fb9f96. Why are you still running into
> the
> problem, exactly?
>

I've looked at that commit, but that also doesn't really work, because
noexcept is not part of the function type. This currently happens to work
in both Clang and GCC due to bugs (
https://llvm.org/bugs/show_bug.cgi?id=16469,
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=49332), but I guess we
shouldn't rely on compiler bugs. I'd suggest to revert both commits.

[-- Attachment #2: Type: text/html, Size: 1129 bytes --]

Re: Please revert commit ceaca7bec6adf5735d51c1ca7efea5a2e05f470e

[Eli Zaretskii]

> From: Philipp Stephani <p.stephani2@gmail.com>
> Date: Sun, 10 Jan 2016 20:37:27 +0000
> 
>     That was fixed long ago in emacs-25, in commit
>     7210f5130abb909aebcbb09a8a71558100fb9f96. Why are you still running into
>     the
>     problem, exactly?
> 
> I've looked at that commit, but that also doesn't really work, because noexcept
> is not part of the function type. This currently happens to work in both Clang
> and GCC due to bugs (https://llvm.org/bugs/show_bug.cgi?id=16469,
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=49332), but I guess we shouldn't
> rely on compiler bugs. I'd suggest to revert both commits.

Can you show the sign(s) of the problem, like compiler error messages,
programs that crash, etc.?  It's hard to reason about a problem
without a good understanding of what it is.

Thanks.

Re: Please revert commit ceaca7bec6adf5735d51c1ca7efea5a2e05f470e

[Paul Eggert]

Eli Zaretskii wrote:
> Can you show the sign(s) of the problem, like compiler error messages,
> programs that crash, etc.?

That would require some actual modules written in C++ that illustrate the 
disasters of allowing exceptions to propagate into C code, which I presume would 
be some work.

I briefly looked into this and though I am by no means a C++ expert it appears 
that we went over the edge of the standard on that one, so in commit 
552694a2653b4f9cde515139d01793a5a0e964b7 I reverted both changes and moved the 
problematic typedef out of the public .h file (where it would just be a bug 
magnet anyway, apparently) and into our private .c file.