fix build errors with glibc 2.26+ due to missing xlocale.h (issue #675)

[janusw]

• check for the presence of xlocale.h via cmake
• remove related logic from Standard_CLocaleSentry.hxx

[tpaviot]

@janusw thank you for your contribution

[janusw]

Thanks for merging! :)

[tpaviot]

@janusw FYI your commit broke master because a failure in osx build. My bad, I should have carefull read the travis log. Anyway, commit 1a575f2e14889fd894d448ed65217a61bcb114df is a fix for osx. Please submit any better solution

[janusw]

Oops, sorry for the breakage. Unfortunately I have no access to an Apple machine right now, so I cannot test on OSX (ditched my MacBook for an XPS13 couple of years ago).

Also I don't quite understand the problem (and the fix). If OSX wants to #include , why is the file not found via cmake? Or is it?

Can you give more detail on how it breaks?

Cheers, Janus

[janusw]

Ok, I guess I found the Travis log (https://travis-ci.org/tpaviot/oce/jobs/318008093), where cmake says:

Looking for xlocale.h - found

and then the actual error is:

/Users/travis/build/tpaviot/oce/src/Standard/Standard_CString.cxx:95:10: error: no matching function for call to 'strtod_l'
  return strtod_l (theStr, theNextPtr, Standard_CLocaleSentry::GetCLocale());
         ^~~~~~~~
[  1%] Building CXX object adm/cmake/TKernel/CMakeFiles/TKernel.dir/__/__/__/src/Standard/Standard_MMgrOpt.cxx.o
/usr/include/xlocale/_stdlib.h:40:9: note: candidate function not viable: cannot convert argument of incomplete type 'clocale_t' (aka 'void *') to 'locale_t' (aka '_xlocale *') for 3rd argument
double   strtod_l(const char *, char **, locale_t) __DARWIN_ALIAS(strtod_l);
         ^

[janusw]

Unfortunately I'm a bit lost here. In particular I don't really understand the behavior on Mac OS, as described in the previous comment:

• It seems that cmake finds xlocale.h, thus HAVE_XLOCALE_H should be defined.
• But then the error in Standard_CString.cxx seems to indicate that HAVE_XLOCALE_H is not defined (because clocale_t is void*).

I guess I must be missing something here, because the above doesn't make much sense to me. I'd appreciate any kind of comments (or testing from someone who has access to a Mac OS machine).

[janusw]

In any case it seems a workaround has been applied in 699869bdd6a9defb431fa4b151477a8f86c4ea57.