Re: gcc on OSR5 status. was: bootstraps broken during last 24 hours

[Robert Lipe <robertlipe at usa dot net>]

Jean-Pierre Radley wrote:

> I just did a configure without specifying any languages.  Went away, and
> it quit in the middle of java compiling with a complaint about some iconv
> functions; weird, considering that I've got a brand new libiconv.so
> just installed yesterday (can't compile mutt without it these days).

Since my system doesn't have a libiconv.so on it, I'll let you chase
that one.

> In file included from /s/tools/gcc/libstdc++-v3/include/bits/locale_facets.h:39,
> 
>                  from /s/tools/gcc/libstdc++-v3/include/bits/std_locale.h:38,
>                  from /s/tools/gcc/libstdc++-v3/include/bits/std_fstream.h:40,
>                  from /s/tools/gcc/libstdc++-v3/src/ios.cc:36:
> /s/tools/gcc/libstdc++-v3/include/c_std/bits/std_ctime.h:52: parse error before
> 
>    `double'

Oh, yes.  That.  Sorry, I forgot to disclaim that my tree isn't quite
pristine.

On OpenServer, 'difftime' is a macro so the 'extern "C"'prototype for it
in the above file loses.  My tree contains this fix for this problem.
There was some squabbling with the libstdc++v3 guys as to how to best
fix this, so I don't think this patch was accepted as-is.

Index: include/c_std/bits/std_ctime.h
===================================================================
RCS file: /cvs/gcc/egcs/libstdc++-v3/include/c_std/bits/std_ctime.h,v
retrieving revision 1.4
diff -c -5 -p -u -r1.4 std_ctime.h
cvs server: conflicting specifications of output style
--- std_ctime.h 2001/01/02 07:38:47     1.4
+++ std_ctime.h 2001/01/09 04:45:32
@@ -39,10 +39,11 @@
 #include <bits/c++config.h>
 #include <bits/std_cstddef.h>
 
 #pragma GCC system_header
 #include <time.h>
+#undef difftime
 
 namespace std
 {
   using ::clock_t;
   using ::time_t;