macOS: minor fixes for < 10.9

[barracuda156]

@gaborcsardi Could you please take a look? I hope the patch is minimal enough. With this, ps builds on my 10.6 ppc.

[barracuda156]

ObjC should simply not be used outside of MacOS, so I do not really get why this warning:

❯ checking compilation flags in Makevars ... WARNING
  Non-portable flags in variable 'PKG_OBJCFLAGS':
    -std=c11

0 errors ✔ | 1 warning ✖ | 0 notes ✔
Error: Error: R CMD check found WARNINGs
Execution halted
Error: Process completed with exit code 1.

What should I do to fix this though?

[gaborcsardi]

IDK, sorry, possibly a base R bug. Maybe there is a way to do what you want without that flag? Why do you need it?

[barracuda156]

@gaborcsardi No, it is nothing about R, the compiler explicitly requires that:

/opt/local/bin/gcc-mp-13 -I"/opt/local/Library/Frameworks/R.framework/Resources/include" -DNDEBUG   -isystem/opt/local/include/LegacySupport -I/opt/local/include    -fPIC  -pipe -Os -arch ppc -fobjc-exceptions  -c arch/macos/apps.m -o arch/macos/apps.o
arch/macos/apps.m: In function 'ps__list_apps':
arch/macos/apps.m:40:3: error: 'for' loop initial declarations are only allowed in C99 or C11 mode
   40 |   for (NSUInteger i = 0; i < count; i++) {
      |   ^~~
arch/macos/apps.m:40:3: note: use option '-std=c99', '-std=gnu99', '-std=c11' or '-std=gnu11' to compile your code
make: *** [arch/macos/apps.o] Error 1

UPD. Or if you mean the CI error is an R bug, that could be, I cannot say. Could we set the standard for ObjC without passing any flags?

[gaborcsardi]

And none of those options are portable according to R? Maybe you can specify that option in your own .R/Makevars file, and then the package does not need the change.

[barracuda156]

And none of those options are portable according to R? Maybe you can specify that option in your own .R/Makevars file, and then the package does not need the change.

But that will leave the build with gcc broken for everyone else, completely unnecessarily.

[gaborcsardi]

Well, we cannot add it to the package if it fails the CRAN checks.

[barracuda156]

Well, we cannot add it to the package if it fails the CRAN checks.

Yeah, I understand that, there should be a way to add it correctly (according to CRAN).
Let me drop this commit from here, it can be dealt with separately, since it is a separate issue.

[gaborcsardi]

If it is needed for one kind of compiler, then I think it either belongs to the user Makevars file, or the system Makevars file. E.g. I don't know how you are compiling R, and it could be set there, and then it'll be the default for that R installation.

[barracuda156]

If it is needed for one kind of compiler, then I think it either belongs to the user Makevars file, or the system Makevars file. E.g. I don't know how you are compiling R, and it could be set there, and then it'll be the default for that R installation.

I have dropped that commit now.

P. S. The issue with ObjC is that on its own it does not require a specific standard (when aqua module is built), so R config won’t have it. It is also possible, I think, to build R on macOS without aqua at all, and then ObjC compiler is not used during building R itself. But it still will be pulled in here in ps.

[gaborcsardi]

Sure, but you can still specify the default OBJC flags when building R, whether it is used or not. Just to have a good default for packages.

[barracuda156]

Sure, but you can still specify the default OBJC flags when building R, whether it is used or not. Just to have a good default for packages.

We can definitely have a local fix for ObjC issue in MacPorts, no problem.

[gaborcsardi]

Thank you!