[cc'ed to Brent Dax for his specific attention]
On Fri, 27 Dec 2002 20:34:33 -0500, "Dan Sugalski" wrote:
>At 1:22 AM +0000 12/27/02, Bruce Gray (via RT) wrote:
>>The gcc compiler in MinGW (win32) emits warnings for invalid
>>pragmas in win32.h (platform.h).
>>This patch uses '#ifndef __GNUC__' to hide the pragmas from gcc.
>
>Nope--there are more compilers that aren't gcc than just MinGW.
>Is there a MinGW specific #define that can be tested instead?
Could you take a second look at the patch? Your comments
sound like you think I want to:
"hide the pragmas from gcc, but let MinGW see them"
when really I want to:
"hide the pragmas from gcc (which is the same as
MinGW in this case)"
The C compiler in MinGW *is* gcc, just like the compiler in
CygWin is gcc. MinGW and CygWin are just the environments
surrounding the build tools and the runtime libs. Any
version of gcc will complain about the "#pragma warning()"
lines. "#ifndef __GNUC__" is what I meant to say, because
gcc is the only compiler that I *know* chokes on those
pragmas. It seems likely that some other compilers
(Borland?, Mars?) will also choke on them, but I don't yet
know what they are, so I did not change the code for them yet.
Having said that, I see that nested "#ifndef __GNUC__",
#ifndef __BORLAND", "#ifndef __MARS__", etc will be
sub-optimal. The better choice is to only make the pragmas
visible to the specific compiler that understands them. Now
that I have done more checking, I think that compiler is
Microsoft's. If Brent Dax will verify that his original
intent was for those pragmas to apply only to Microsoft's C
compiler, then an improvement would be to change one line of
the patch from
+#ifndef __GNUC__
to
+#ifdef _MSC_VER
Please let me know if I have misunderstood you, or have
muddied the issue further.
--
Thank you,
Bruce Gray
