Aran,

It's possible that the problem is related to autoconf. I have seen similar
problems that were related to determing where and how libdl is supported.
libdl is *critical* for any of the named primitive stuff - no named
primitives (e.g., all of the Balloon prims) will work without libdl. You
could try to poke around in this stuff using 'nm' but it's hard to tell for
sure.

  Andreas

> -----Original Message-----
> From: Aran Lunzer [mailto:aran at meme.hokudai.ac.jp]
> Sent: Monday, February 21, 2000 3:03 PM
> To: Squeak Mail List
> Cc: recipient list not shown
> Subject: SPARC Solaris Squeaks Successfully
> 
> 
> (Re-post of bounced message.)
> 
> -------- Original Message --------
> Subject: SPARC Solaris Squeaks Successfully
> Date: Fri, 18 Feb 2000 19:18:07 +0900
> To: Squeak Mail List <squeak at cs.uiuc.edu>
> CC: Nathanael Schaerli <n.schaerli at gmx.net>
> 
> After various attempts, I now have Squeak 2.7 running happily on our
> local SPARC Solaris server thanks to the kind efforts of Nathanael
> Schaerli.
> 
> So the question is: why did Nathanael have to re-compile to get a
> version that run successfully (at least for him and for me) - 
> i.e., that
> didn't have BalloonEngine problems?
> 
> Here are the results of "squeak -version" for Nathanael's VM:
> 
>   sparc-sun-solaris2.6 2.7 #1 XShm Thu Feb 17 09:53:57 MET 2000 gcc
> egcs-2.91.66
>   SunOS kilana 5.6 Generic_105181-15 sun4u sparc SUNW,Ultra-250
> 
> and for the public distribution:
> 
>   sparc-sun-solaris2.5.1 2.7 #1  Thu Jan 13 16:24:22 MET 2000 gcc
> egcs-2.91.66
>   SunOS fricotin.inria.fr 5.5.1 Generic_103640-28 sun4u sparc
> 
> Notes:
>   - exactly the same compiler
>   - the version that works was compiled on 2.6 instead of 2.5.1 - but
> then, my system is running 2.5.1 anyway!
>   - some difference indicated by the presence of "XShm"?
>   - some difference indicated by the presence of "SUNW,Ultra-250"?
> 
> Also something undeniably suspicious: all Nathanael's binaries are a
> little bigger than the distribution version, apart from Profiler.so
> which is a massive 50 times as big (1.6MB)!
> 
> It would be nice if we could figure out what's going on here.  Is the
> public version simply missing something?
> 
> 
> Incidentally - Nathanael wrote:
> > By the way: The Solaris version is quite slow. I compiled it 
> > using different compilers (sun-cc, egcs), but it was always 
> > as slow.
> 
> Running the simple benchmarks for 2.7 on our super-macho UltraSPARC
> gives 390,000 sends, 9,000,000 bytecodes per second - perfectly fast
> enough for use, but evidently less than half the speed of some desktop
> Macs/PCs.  But then, the poor thing's already two years old :-)
> 
> Bye for now
> 
> Aran
> -- 
> Aran Lunzer                 aran at bigfoot.com
> Meme Media Laboratory       lab: +81 11 706 7262 / fax 7808
> Hokkaido University
> Sapporo 060-8628, JAPAN     http://ca.meme.hokudai.ac.jp/people/aran/
>