https://issues.dlang.org/show_bug.cgi?id=12715

Walter Bright <bugzilla@digitalmars.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bugzilla@digitalmars.com

--- Comment #1 from Walter Bright <bugzilla@digitalmars.com> ---
This is deliberate. The reason is to not break existing code. People shouldn't have to go back and annotate code with 'nothrow' that was already working.

--

https://issues.dlang.org/show_bug.cgi?id=12715

Denis Shelomovskij <verylonglogin.reg@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|regression                  |normal

--- Comment #2 from Denis Shelomovskij <verylonglogin.reg@gmail.com> ---
(In reply to Walter Bright from comment #1)
> This is deliberate. The reason is to not break existing code. People shouldn't have to go back and annotate code with 'nothrow' that was already working.

But this code isn't "working". It's wrong and is a source of heisenbugs like every other code throwing exceptions across language boundaries. I see no excuse to have incorrect C library bindings and thus force D users to pay on-going cost of fixing heisenbugs because of it.

And yes, it isn't a regression.

--

https://issues.dlang.org/show_bug.cgi?id=12715

Martin Nowak <code@dawg.eu> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |pull
                 CC|                            |code@dawg.eu
           Severity|normal                      |regression

--- Comment #3 from Martin Nowak <code@dawg.eu> ---
(In reply to Walter Bright from comment #1)
> This is deliberate. The reason is to not break existing code. People shouldn't have to go back and annotate code with 'nothrow' that was already working.

Those callbacks were annotated nothrow 4 years ago so the code breaking
argument doesn't make sense.
https://github.com/WalterBright/druntime/commit/370ffd8d6da3978f742fd17572b418aaf85db355

Pull: https://github.com/D-Programming-Language/druntime/pull/787

--

https://issues.dlang.org/show_bug.cgi?id=12715

Martin Nowak <code@dawg.eu> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|regression                  |normal

--- Comment #4 from Martin Nowak <code@dawg.eu> ---
(In reply to Martin Nowak from comment #3)
> (In reply to Walter Bright from comment #1)
> > This is deliberate. The reason is to not break existing code. People shouldn't have to go back and annotate code with 'nothrow' that was already working.
> 
> Those callbacks were annotated nothrow 4 years ago so the code breaking argument doesn't make sense. https://github.com/WalterBright/druntime/commit/ 370ffd8d6da3978f742fd17572b418aaf85db355

Sorry for the noise, global attributes do not affect function types in function signatures, so you were right about the code breakage.

--

https://issues.dlang.org/show_bug.cgi?id=12715

Iain Buclaw <ibuclaw@gdcproject.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P1                          |P3

--

https://issues.dlang.org/show_bug.cgi?id=12715

--- Comment #5 from dlangBugzillaToGithub <robert.schadek@posteo.de> ---
THIS ISSUE HAS BEEN MOVED TO GITHUB

https://github.com/dlang/dmd/issues/17276

DO NOT COMMENT HERE ANYMORE, NOBODY WILL SEE IT, THIS ISSUE HAS BEEN MOVED TO GITHUB

--