On Thu, 26 Nov 1998, Todd Vierling wrote:

> On Thu, 26 Nov 1998, Thorsten Frueauf wrote:
> 
> : But (I somehow expected that, as java is really a monster) it seems as
> : java uses unfortunalty some unimplemented syscalls :-(
> 
> :    367 java     CALL  #209 (unimplemented resolvepath)
> 
> Hm.  I suppose we probably should look into implementing some more (newer)
> Solaris syscalls.  resolvepath() looks almost trivial from what I see in the
> Solaris manpage.
> 
> : So is it hopeless to get the jdk 1.2 from Solaris to work on NetBSD/i386?
> 
> Not if you use "green_threads", but that probably means implementing a few
> more syscalls than just this.  To explain what "green_threads" is:
> 
> On Thu, 26 Nov 1998, Scott Bartram wrote:
> 
> : It's unlikely that the Solaris JDK will work for quite some time if ever
> : since they use native threads and we don't support that (yet). 
> 
> I believe the JDK still ships "green_threads" as one of the thread methods,
> which is not Solaris native threads; it's setjmp/longjmp/signal based (MIT
> pthreads-like) threading.  Otherwise that ktrace probably would have an
> `unimplemented _lwp_create' as the bad syscall.

The "JavaSoft reference" versions are green_thread based and that will
probably always be true for the 1.1 branch. The question is if HotSpot
(the new JVM due out soon after 1.2) will continue to support both. 

The Sun Production versions are all native thread based. I don't know if
they will work with green_threads. 

scott
---
Scott Bartram
scottb@orionsoft.com
http://www.orionsoft.com/