Chris,
This is a heavier lift than most darwin related fixes as FSF gcc will
have to implement the missing availability attribute support (by, I assume,
using the clang implementation as the reference code spec).
Jack
On Wed, Jul 10, 2019 at 8:17 AM Chris Jones <[email protected]>
wrote:
>
>
> On 09/07/2019 11:48 pm, Jack Howarth wrote:
> > Just saying that since gcc will be broken on the 10.15 SDK that the path
> > of least resistance (and the best long term decision) would be to put
> > some effort behind getting flang to build (as older patches exist as a
> > starting point).
>
> Whilst I happen to suspect gcc will, come the final release of macOS10,
> get around to fixing their issues (the problem is really with gcc, not
> Apple), I also agree having a build of flang would be great. Whether
> though we wait for them to fix flang, or we take a look, is another
> question...
>
> Chris
>
> >
> > On Tue, Jul 9, 2019 at 12:35 PM Ryan Schmidt <[email protected]
> > <mailto:[email protected]>> wrote:
> >
> >
> >
> > On Jul 8, 2019, at 14:37, Jack Howarth wrote:
> >
> > > Since the legacy header support for avoiding the availability
> > attribute has been dropped from the 10.15 SDK...
> > >
> > > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90835
> > >
> > > and it is unclear when and if the missing availability attribute
> > support will be added to FSF gcc, the path of least resistance might
> > be to get a flang package into MacPorts to replace gfortran.
> > > Currently flang doesn't build on macOS...
> > >
> > > https://github.com/flang-compiler/flang/issues/678
> > > https://github.com/flang-compiler/flang/issues/212
> > >
> > > https://github.com/flang-compiler/flang/pull/592
> >
> > Certainly, a prerequisite for being able to add a port for something
> > is that it builds.
> >
> >
>
