[dsdp-tcf-dev] [Bug 311952] New: Build fix for MinGW build of TCF agent

[bugzilla-daemon@xxxxxxxxxxx]

https://bugs.eclipse.org/bugs/show_bug.cgi?id=311952 
Product/Component: Target Management / TCF

           Summary: Build fix for MinGW build of TCF agent
    Classification: DSDP
           Product: Target Management
           Version: 3.2
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Severity: normal
          Priority: P3
         Component: TCF
        AssignedTo: dsdp.tm.tcf-inbox@xxxxxxxxxxx
        ReportedBy: ed.swartz@xxxxxxxxx
         QAContact: martin.oberhuber@xxxxxxxxxxxxx


We build the Windows TCF agent using Mingw 3.4.5.  We include some C++ sources
in our side, and when using these, the compiler issues errors due to this line
in mdep.h:

(line 131)
#define snprintf _snprintf

with such reports as, e.g.:

C:/MinGW/bin/../lib/gcc/mingw32/3.4.5/../../../../include/c++/3.4.5/cstdio:167:
error: `::snprintf' has not been declared
C:/MinGW/bin/../lib/gcc/mingw32/3.4.5/../../../../include/c++/3.4.5/cstdio:177:
error: `__gnu_cxx::snprintf' has not been declared
C:/MinGW/bin/../lib/gcc/mingw32/3.4.5/../../../../include/c++/3.4.5/mingw32/bits/c++locale.h:72:
error: `snprintf' is not a member of `std'

This kind of #define pattern seems like an MSVC-ism (to access extension
functions which aren't in standard C).  This pattern seems to be unnecessary in
GCC-derived builds.

Is it safe to just guard these as:

(mdep.h line 128)
#ifdef _MSC_VER
#define utimbuf _utimbuf
#define utime   _utime
#define futime  _futime
#define snprintf _snprintf
#endif

I did this and the build succeeded under MinGW, so it seems safe...

-- 
Configure bugmail: https://bugs.eclipse.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching the assignee of the bug.