This looks better. I was reluctant to use build numbers as a substitute
for micro as build numbers are AFAIK sequential throughout the whole
windows series.
I got the point that undef'ing is a no-no in the comp tests.
When it comes to OS version, I believe the build number is the one that
comes closest to be the equivalent of UNIX "uname -v". It would be very
helpful to hear more opinions on this one though.
Liviu
Martin Sebor wrote:
> Liviu Nicoara wrote:
>
> Thanks! (In the future, please make sure to explain what the patch
> does in case it's not obvious to everyone).
>
> Here's the output of the test I get on Windows 2000 Professional:
>
> #define _RWSTD_OS_WINDOWS
> #define _RWSTD_OS_SYSNAME "WINDOWS"
> #define _RWSTD_OS_RELEASE "5.0"
> #define _RWSTD_OS_MAJOR 5
> #define _RWSTD_OS_MINOR 0
> #undef _RWSTD_OS_VERSION
> #undef _RWSTD_OS_MICRO
>
>
>>Index: etc/config/src/UNAME.cpp
>>===================================================================
>>--- etc/config/src/UNAME.cpp (revision 219737)
>>+++ etc/config/src/UNAME.cpp (working copy)
>>@@ -103,8 +103,27 @@
>>
>> #else
>>
>>- return 1;
>>+ OSVERSIONINFO osinfo;
>>+ osinfo.dwOSVersionInfoSize = sizeof osinfo;
>>
>>+ if (!::GetVersionEx (&osinfo))
>
>
> (FYI: there's no reason to qualify the call above.)
>
>
>>+ return 1;
>>+
>>+ printf ("#define _RWSTD_OS_WINDOWS\n");
>>+ printf ("#define _RWSTD_OS_SYSNAME \"WINDOWS\"\n");
>
>
> I would suggest to #define these two unconditionally (i.e., regardless
> of whether GetVersion() succeeds or not), and probably also to avoid
> returning a non-0 status on GetVersion() failure.
>
>
>>+
>>+ // no micro on Windows
>>+ printf ("#define _RWSTD_OS_RELEASE \"%d.%d\"\n",
>>+ osinfo.dwMajorVersion,
>>+ osinfo.dwMinorVersion);
>
>
> I think this macro would be more useful if it identified the flavor
> of Windows, i.e., one of "2000", "NT 4.0", "95", "XP", "Server 2003",
> "98", or "Me". It should be possible to extract it (or most of it)
> from dwMajorVersion, dwMinorVersion, and dwPlatformId:
> http://msdn.microsoft.com/library/default.asp?url=/library/en-us/sysinfo/base/osversioninfo_str.asp
>
>
>>+
>>+ printf ("#define _RWSTD_OS_MAJOR %lu\n", osinfo.dwMajorVersion);
>>+ printf ("#define _RWSTD_OS_MINOR %lu\n", osinfo.dwMinorVersion);
>
>
> Hmmm. These are supposed to let us easily identify the version in
> preprocessor conditionals. Let's see, what would the values for
> the different flavors of Windows look like?
>
> Windows _RWSTD_OS_MAJOR _RWSTD_OS_MINOR
> +-----------+---------------+---------------+
> | 95 | 4 | 0 (*) |
> | 98 | 4 | 10 |
> | Me | 4 | 90 |
> | NT 4.0 | 4 | 0 (*) |
> | 2000 | 5 | 0 |
> |Server 2003| 5 | 2 |
> | XP | 5 | 1 |
> +-----------+---------------+---------------+
>
> Okay, that's not too bad, except that the special case (*) would
> have to be handled. It might be useful to also #define a helper
> macro corresponding to each flavor of Windows:
>
> _RWSTD_OS_WINDOWS_95
> _RWSTD_OS_WINDOWS_98
> _RWSTD_OS_WINDOWS_ME
> _RWSTD_OS_WINDOWS_NT
> _RWSTD_OS_WINDOWS_2000
> _RWSTD_OS_WINDOWS_2003
> _RWSTD_OS_WINDOWS_XP
>
>
>>+
>>+ // not applicable
>>+ printf ("#undef _RWSTD_OS_VERSION\n");
>>+ printf ("#undef _RWSTD_OS_MICRO\n");
>
>
> These tests shouldn't #undefine any config macros. If it's not
> applicable, the _RWSTD_OS_MICRO should be #defined to 0, although
> it might be meaningful to define to the Windows build number
> (dwBuildNumber, or its low word on 95/98/Me). _RWSTD_OS_VERSION
> could be #defined to szCSDVersion.
>
> Attached is a modified version of the patch with the changes
> mentioned above. It produces the following output on the same
> Windows 2000 box:
>
> #define _RWSTD_OS_WINDOWS
> #define _RWSTD_OS_SYSNAME "WINDOWS"
> #define _RWSTD_OS_MAJOR 5
> #define _RWSTD_OS_MINOR 0
> #define _RWSTD_OS_MICRO 2195 /* build number */
> #define _RWSTD_OS_RELEASE "Windows 2000"
> #define _RWSTD_OS_WINDOWS_2000
> #define _RWSTD_OS_VERSION "Service Pack 4"
>
> Martin
>
