Re: [Hecl-devel] Issue with synchronized eval
Brought to you by:
davidw
From: Wolfgang S. K. <wol...@da...> - 2008-01-21 08:09:14
|
David Welton wrote: >> I spend some more time and Thread.yield(0 is the bad guy. I will make >> modifications and run tests to identify the performance enhancements we >> can get. > > Cool... so eliminating that fixes the problem? > I managed to avoid yield() what causes some other modifications. This reduces http get/post times dramatically (is costed some 3 seconds per call at least on mobile devices before). Various postings in several places indicate that other programmers observe identical behavior. I am currently investigating in which methods of the interpreter should be synchronized. Definitly, ist stack manipulation and variable access. I think eval need not be synchronized, but it requires some more time to make it robust. Since the changes I made are very subtle, I need to download the latest souorce from svn and compare every changed bit with my changes to create a combined robust result. Hopefully I will be able to finalize the work during this week. Unfortunatly, I detected a new problem causes by the change of the preprocessor directives; the new antenna directives are not understand by j2meposlish, so it is currently not possible to create a j2me-polished version of hecl (i.e. integrate Hecl into a j2mepolished application). I fear I need to address this as well, since we use Hecl together with this development tool as well. Regards -- Wolfgang Kechel mailto:wol...@da... data2c GmbH http://www.data2c.com Bierstadter Straße 7 Fax: +49-(0)611-1731-31 D-65189 Wiesbaden Phone: +49-(0)611-1731-611/+49-(0)174-3454260 |