[Mingw-users] Re: Clock/Time problems with GNAT

[Stefan B. <sb...@sb...>]

Danny Smith wrote:

[snip]

> Try changing that #ifdef _WIN32 to  #if TARGET_IS_PE_COFF.  If that
> works, you may like to submit a bug report and/or patch to GCC.

Sounds good. With my test example everything is fine now. I will now
try our main project with 760.000 lines of Ada 95 code.

Speaking of patches, I needed another one to make linking work:

--- gcc-3.4.2-20040916-1/gcc/ada/Makefile.in    2005-03-10 14:08:55.710970566 +0100
+++ gcc-3.4.2-20040916-1/gcc/ada/Makefile.in.patched    2005-03-10 14:14:47.813280451 +0100
@@ -106,9 +106,9 @@
 AR_FLAGS = rc
 LS = ls
 # How to invoke ranlib.
-RANLIB = ranlib
+RANLIB = i586-mingw32msvc-ranlib
 # Test to use to see whether ranlib exists on the system.
-RANLIB_TEST = [ -f /usr/bin/ranlib -o -f /bin/ranlib ]
+RANLIB_TEST = [ ! -z "$(RANLIB)" ]
 
 SHELL = @SHELL@
 PWD_COMMAND = $${PWDCMD-pwd}

When not using this patch, the built libraries cannot be linked against
because the target ranlib hasn't processed them. I think it would be
nicer if "i586-mingw32msvc-ranlib" could be replaced with something like
"@RANLIB_FOR_TARGET@" but this didn't seem to work.

Where should I submit those patches? To mainstream GCC or to you? You
suggest submitting the stdcall bug to GCC, but is it fixed there
quickly?

-- 
Stefan Bellon