Another jdk build error on -CURRENT

[Jens Rehsack]

Greg Lewis wrote:
> On Fri, Oct 03, 2003 at 10:43:00AM +0000, Jens Rehsack wrote:
> 
>>after I patched out all '-O3' flags as recommented, the build failed
>>again at the same place with the same core dump. By the way, the hint
>>of Thomas to use the debug versions worked fine. Everytime the build
>>stops with core dump, I link the debug executables those ones without
>>debug (ln -f $file_g $file).
>>
>>Now I'm getting 2 new errors:
>>1) Because of the -Werror in j2se/make/java/nio/Makefile (already
>>   in and added by patch for bsd platform), gcc3.3.1 fails because
>>   _LITTLE_ENDIAN is already defined. Setting both occurances to
>>   -Wno-error solved that.
> 
> I have patches for this one.  It switches the JDK to use <sys/endian.h>
> rather than do its own endian thing.

Maybe it would be a good thing (looking at all new warnings generated
by new system compiler in -CURRENT) to use -Wno-error even.

> 
>>2) In the makefiles is the threading linkage different between
>>   -STABLE and -CURRENT, j2se/ext/plugin/build/solaris/GNUmakefile
>>   seems to does not. The following error results:
> 
> I'll look into this.  All we need to do is check the version and use
> -lc_r instead (or at least thats what bsd.port.mk does).
> 

I was collecting both described messages, because I assumed there
may come up more during build. And than I forgot to send them :-)

Of course, I'm sure you work further on it, but I think when I don't
have enough time to create patches best I can do is send all
errors happens here.

Can you estimate a 'release time' for the new patch-level?
It sucks a little bit fixing the -current patchset into the
jdk-1.4_3p3 port and rename *_g -> \1 each portupgrade :-(

Jens