forcing 10.12 or 10.13 builds onto 10.14 for 32bit fix?

[Mojca Miklavec]

On Mon, 29 Oct 2018 at 21:43, Joshua Root wrote:
> On 2018-10-30 06:13 , Mojca Miklavec wrote:
> > That said, I would like it if MacPorts had somewhat better support for
> > compiling binaries targeting older OSes on newer ones, ...
>
> Unfortunately it's mostly not MacPorts that needs the support but the
> individual build systems, most of which have no concept of macOS weak
> linking. How many ports would need to be manually patched to not use
> utimensat when targeting 10.12, as just one example?

There are two possibilities:
- weak linking
- actually linking against an older SDK

Proper support for the first option would be next to mission
impossible (for anything but [Objective C] programs which have been
written with macOS in mind). Support for the second group is totally
feasible, but MacPorts doesn't even consider this at the moment.
Portfiles use something like "if this is darwin X, then do Y", but in
many cases one would need this condition to depend on the target OS
rather than on the OS where the software is being compiled.

(Anyway, I would not consider this such a high priority to do. Simply
getting Wine and alike to work by any other means sounds more
interesting and important, but even upstream developers will need to
do something soon anyway.)

Mojca