I am able to build all the Freeciv branches on my Mac. If there are mac-specific things which need testing, you are welcome to assign them to me and I will eventually test it.
In addition to the implementation being empty, there's second point in jdlh's report that I almost missed. The value that the configure option is supposed to set is only used in the search for 'moc'. It's not used when looking for 'qtpaths'. It might have been an intentional design decision as 'qtpaths' is supposed to provide the *default* auto-detected paths that the configure option can override, but that design is broken in several ways (including the fact that 'qtpaths' is currently used also for things other than getting the bin directory)
CI macOS build scripts have overcome these issues by explicitly setting MOCCMD, and I guess things like CPPFLAGS and LIBS, so they don't really do any search for Qt at all. Should clean those script implementations once we get this low-level part to work correctly.
I am able to build all the Freeciv branches on my Mac. If there are mac-specific things which need testing, you are welcome to assign them to me and I will eventually test it.
I hope to get patch for this submitted later tonight, after doing some runs of CI mac build.
The interaction between --with-qt?-framework from #44412 and the --with-qt?-framework-bin might be a bit unclear.
With the implementation I currently have --with-qt-framework does not affect search of binaries at all - you need to use --with-qt-framework-bin for that.
If you only use --with-qt-framework-bin, that sets also the overall framework location with qtpaths.
If you use both options, the overall framework location is one explicitly given, and the ...-bin only affects search for the 'moc' -binary (qtpaths is not needed)
Patches attached. They are not breaking our existing CI builds, but I have had no setup to test if they improve the situation either. Can jdlh test if these fix the issues for him?
I'm not absolutely sure, but it's very likely that this depends on patch on #44412 (at least textually). Probably better to test with that one (which likely goes in first, anyway) applied first.
Reply To cazfi
Can jdlh test if these fix the issues for him?
Ping.
We may need to make the 3.0.2 release sooner than what we've planned so far.
I probably push this in soon, even if it doesn't get tested before. What didn't do anything before, cannot go more broken, at least.
As reported by jdlh in #44412, --with-qt5-framework-bin / --with-qt6-framework-bin configure options have no implementation at all - there's just the helptext for them.