December 3, 2008

Attendees

Intel: Shengnan
Adobe: Edwin Smith, Steven Johnson, Tom Harwood, Leslie, Lars, Jeff Dyer, Rob Winchell, Rob Borcic, Trevor Baker, Dan Schaffer, Brent Baker, Dan Smith, Rick Reitmaier, Rishit Shah

All acceptance tests passing - revision 703 of TC is stable
Trevor to send changelist to list and post on other locations.
Code size - 79 K bigger than VM in Flash 10
- Prior to merge from Redux, we had reduced by 85K
Performance - faster across the board. Most dramatic improvement linux using interpreter flash player 10 up to 182% faster
Memory usage
- mac using 8% less memory
- every other platform using more (worst is using forceMIR on Windows using about 22% more memory)
ACTION for QE to be added to backlog: Need a measure relative to the flash 10 VM
Release notes (detailing changelist) will be provided.

PPC Nanojit port, now at Beta quality, code here - [1]
Tracking bug here: [2]
Performance is decent, ranging from slightly slower to 25% faster on sunspider/* and misc/*.

Now it can optimize and run most programs
working on a deployment guide
Performance - code size reduced by 15-20% (with standard deviation about the same)
Still improving the code generated on the backend
Will look at LVM integration and getting better error behavior
GO will be posted on opensource.adobe.com (when it is in its final state)

Will post notes to tamarin devel list and check into tamarin repository
Need a doc directory in Tamarin (off of the root)
Feel free to review doc when posted
Will create a separate bug to track the verifier documentation
Next step is to do some refactoring of the verifier to separate code generation and verification logic.

Brent will ping Leon (bug for Sun Studio compiler) to see if there is an update as to when the compiler will be fixed.
current issue is to modify code to get around issues for sun studio compiler for nanojit to work. Leon seemed to think it will be fixed in the sun compiler.