R: incorrect info re compiler in the portfile

[barracuda156]

The portfile has:

# Note: gcc cannot be used for the C compiler. It will give:
#:info:build In file included from /usr/include/dispatch/dispatch.h:51:0,
#:info:build                  from /System/Library/Frameworks/CoreFoundation.framework/Headers/CFStream.h:15,
#:info:build                  from /System/Library/Frameworks/CoreFoundation.framework/Headers/CFPropertyList.h:13,
#:info:build                  from langprefs.c:30:
#:info:build /usr/include/dispatch/object.h:143:15: error: expected identifier or '(' before '^' token
#:info:build  typedef void (^dispatch_block_t)(void);
#:info:build                ^
#:info:build /usr/include/dispatch/object.h:362:3: error: unknown type name 'dispatch_block_t'
#:info:build    dispatch_block_t notification_block);
#:info:build    ^
# However, use of compiler.blacklist *gcc* would remove the GCC Fortran compilers too.

This may be true for some versions of macOS, but this is false in general. GCC certainly can be used as the C compiler with R.

16. S. Also I wonder, why no universal for R?

[kencu (Ken)]

clang supports -fblocks by default on systems thst natively support blocks.

gcc does not

​https://gcc.gnu.org/bugzilla/show_bug.cgi?id=78352

So that would be why that is there.

[barracuda156]

Replying to kencu:

clang supports -fblocks by default on systems thst natively support blocks.

gcc does not

​https://gcc.gnu.org/bugzilla/show_bug.cgi?id=78352

So that would be why that is there.

Well, I guess the statement should have it explicitly – since it is easy to verify that GCC works perfectly fine with R at least on <10.7. (And on PPC it is the only choice out there.)

[kencu (Ken)]

true.

I’m not sure if R tests for -fblocks before using it now … previously it seemed to assume it was available on darwin.