C++ API: Union types are missing

C++ API: Union types are missing

[Marc Nieper-Wißkirchen]

The C++ analog for gcc_jit_context_new_union_type is missing in
libgccjit++.h.

I can send a patch adding it and anything else I may find missing as well.

Dave, as only one header file is concerned, would a single patch be okay? I
hope it could still make it into GCC 12 as no component outside of the file
would be touched.

Marc

Re: C++ API: Union types are missing

[David Malcolm]

On Sun, 2022-01-30 at 20:34 +0100, Marc Nieper-Wißkirchen wrote:
> The C++ analog for gcc_jit_context_new_union_type is missing in
> libgccjit++.h.
> 
> I can send a patch adding it and anything else I may find missing as
> well.

Yes please; ideally with test coverage.

> Dave, as only one header file is concerned, would a single patch be
> okay? I
> hope it could still make it into GCC 12 as no component outside of the
> file
> would be touched.

It's technically something I ought to ask permission for from the
release managers given that the rest of gcc 12 is in deep feature
freeze, but I think there's a good case for this - especially if the
patch has test coverage (if nothing else, I've found that it's
impossible to gauge how good an API is without writing code to consume
that API).

Please send patches for libgccjit to both the jit mailing list and the
gcc-patches mailing list (and ensure the title has "jit" in it), and
indicate what testing you've done. [1]

Thanks
Dave

[1]
https://gcc.gnu.org/onlinedocs/jit/internals/index.html#submitting-patches
has more notes on patches, fwiw (perhaps too much)

Re: C++ API: Union types are missing

[Marc Nieper-Wißkirchen]

Am Mo., 31. Jan. 2022 um 15:57 Uhr schrieb David Malcolm <
dmalcolm@redhat.com>:

> On Sun, 2022-01-30 at 20:34 +0100, Marc Nieper-Wißkirchen wrote:
> > The C++ analog for gcc_jit_context_new_union_type is missing in
> > libgccjit++.h.
> >
> > I can send a patch adding it and anything else I may find missing as
> > well.
>
> Yes please; ideally with test coverage.
>
> > Dave, as only one header file is concerned, would a single patch be
> > okay? I
> > hope it could still make it into GCC 12 as no component outside of the
> > file
> > would be touched.
>
> It's technically something I ought to ask permission for from the
> release managers given that the rest of gcc 12 is in deep feature
> freeze, but I think there's a good case for this - especially if the
> patch has test coverage (if nothing else, I've found that it's
> impossible to gauge how good an API is without writing code to consume
> that API).
>

A patch is already out with a test case. :)

Another one will follow that adds the missing support for function pointer
types in the C++ API.

Marc