Re: SableVM 1.1.7 *prerelease* available - please test!

On Tue, 2004-10-12 at 06:00, Laurent Martelli wrote:
> I tried unpacking the jars and removing them, but it's still slow as a
> lame snail:
> 
> 	2004-10-12 11:48:22,182 INFO perf:  application started in 34812ms
> 
> Whereas with 1.1.6:
> 
> 	2004-10-12 09:48:48,250 INFO perf:  application started in 6279ms
> 
> I even tried to unjar almost all the jars used by my app with no
> success. I'm still in the 35 seconds range  :-(

Rene Wagner was so kind to prepare tarballs with SableVM and SableVM
Classpath just before we merged in the latest GNU CP CVS changes (rev.
2967).

http://gadek.homelinux.org/1.1.7-pre/before-merge

Could you please try them on and report whether they also give you
this performance hit or not?  In case the performance problem was
still there could you please do (you need to have subversion client):

svn co -r XXX svn://svn.sablevm.org/sablevm/branches/staging \
 sablevm-staging-revXXX
svn co -r XXX svn://svn.sablevm.org/sablevm-classpath/branches/staging \
 sablevm-classpath-staging-revXXX

where XXX is revision number: 2933, 2824, 2802.  These are revisions
just before we merged in some changes from GNU CP CVS.  You only need to
run ./autogen.sh first and then install them normally to try on each of
them.

I believe the performance regression you observe is due to some changes
in Classpath, as I can't think of any way any of the changes we've done
to SableVM codebase since 1.1.6 release could give such an effect.  We
just need to track down when it happened.  (This is btw. yet another
reason why we desperately need automated nightly tests.  We've just got
somebody who's starting working on it.)

Let us know what you've found out,

				Grzegorz B. Prokopski
-- 
Grzegorz B. Prokopski           <[email protected]>
SableVM - Free, LGPL'ed Java VM  http://sablevm.org
Why SableVM ?!?                  http://sablevm.org/wiki/Features
Debian GNU/Linux - the Free OS   http://www.debian.org