[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
Re: 2008Q3 Solaris 10 bulk build issues ...
On Tue, Jan 6, 2009 at 9:19 PM, Malcolm Herbert <mjch%mjch.net@localhost> wrote:
> On Tue, 23 Dec 2008 15:18:20 +1100, "Malcolm Herbert"
>> I managed to find and fix a few issues with lang/tcl and a few others
>> when compiling from the 2008Q3 pkgsrc tarball under Solaris 10 with the
>> Sun Studio 12 compiler, however these remaining packages are proving to
>> be more difficult ...
> Done some more work on these over the last few days, using the input
> from a number of you (thanks, btw) ... I've put the bulk build results
> at http://www.mjch.net/pub/pkgsrc/pkgstat/20090106.1648/report.html for
> I removed the gmp-fat option as some people suggested and this allows
> the compile to get a little further along - it now fails when compiling
> in the cxx subdirectory: CC appears to have major issues compiling
> ../gmp-impl.h which is strange as it has been included fine by cc when
> compiling almost all the other modules ... it's not a C++ header (.hh)
> file though ... is the format markedly different? This can be seen in
> the output in the pkgstat report above.
> I did manage to prevent the cxx directory being built and the package
> was completed, however p5-Math-GMP relies on the C++ portion of
> devel/gmp, so that failed in short order ... :)
> I'd independently found the first issue Jörn Clausen pointed out was
> able to use the patch described in PR pkg/39612 to prevent the examples
> from being compiled which allowed this package to build.
> The Makefile settings for IConv and some assumptions regarding OS names
> were incorrect, but with some alterations I was able to work around
> this, although I still have some outstanding PLIST issues to resolve
> before I submit a PR with patches for it.
> Once I managed to get textproc/xerces-c to compile I found that
> textproc/p5-XML-Xerces had big issues compiling the SWING glue to
> interact with the C library, as can be seen in the pkgstat output (1500+
> warnings per file? seems dubious)
> Still no idea on these ones at all ...
> I'd like to stay with the Sun Studio compiler if possible, but if not,
> what's the best way to tell pkgsrc to use a natively-installed gcc? I
> had a shot at compiling gcc34 in my existing environment which failed
> miserably ...
I think this effort is great and using Sun Studio is even cooler.
Thanks for doing it!
Main Index |
Thread Index |