https://issues.dlang.org/show_bug.cgi?id=24769

Paul Backus <snarwin+bugzilla@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |snarwin+bugzilla@gmail.com

--- Comment #1 from Paul Backus <snarwin+bugzilla@gmail.com> ---
Where is the proposal for "the proposed phobos.sys.* package"?

--

https://issues.dlang.org/show_bug.cgi?id=24769

--- Comment #2 from Richard (Rikki) Andrew Cattermole <alphaglosined@gmail.com> ---
It is part of PhobosV3, see Adam Wilson's DConf 2024 talk.

A more concrete example of a module that would benefit from it is core.int128, but there are others such as the WinAPI bindings that have symbols declared that are not a binding.

Oh and I've implemented it https://github.com/dlang/dmd/pull/16848

--

https://issues.dlang.org/show_bug.cgi?id=24769

Adam D. Ruppe <destructionator@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |destructionator@gmail.com

--- Comment #3 from Adam D. Ruppe <destructionator@gmail.com> ---
When is this actually useful? The case of the windows bindings' helper functions is already solved by a linker.

If you try to separate compile anything with this pragma two things that depend on it in a diamond shape, you're liable to get duplicate symbols. Easy to solve with -i, but to clear how it would be solved with this pragma.

--

https://issues.dlang.org/show_bug.cgi?id=24769

--- Comment #4 from Richard (Rikki) Andrew Cattermole <alphaglosined@gmail.com> ---
With dmd everything is weak, and will handle duplicates by-default.

It is ldc and gdc which I have concerns with for duplicates.

--

https://issues.dlang.org/show_bug.cgi?id=24769

--- Comment #5 from Richard (Rikki) Andrew Cattermole <alphaglosined@gmail.com> ---
The -i switch suffers the same duplicate issues as the pragma, you should not have to be altering your compile command to remove this, nor tell the linker how to handle duplicates.

Solving it for one, solves it for both.

--