Difference between revisions of "Da Vinci Machine"
From JVMLangSummit
Jump to navigationJump to searchLine 22: | Line 22: | ||
* fixnums and tuples handled better in the JVM | * fixnums and tuples handled better in the JVM | ||
− | = Key Issues for Discussion | + | = Key Issues for Discussion = |
''(please expand cooperatively)'' | ''(please expand cooperatively)'' |
Latest revision as of 11:50, 16 September 2009
Contents
The Da Vinci Report
John Rose - Sun
- Project
- http://openjdk.java.net/projects/mlvm
- Blog
- http://blogs.sun.com/jrose
- Slides
- File:DaVinciMachineTalk.pdf
Abstract
We will survey the present state of the Da Vinci Machine Project, with special emphasis on the implementation and use of method handles and invokedynamic.
Background
- project goal: changes to the JVM bytecode architecture
- JVM has been language-independent from the beginning, but hasn't necessarily been optimal for non-Java languages
- first code posted in 2008, along with draft specifications for reviewer (JSR 292)
- balancing specification with implementation: tail-calls, interface injection, invoke dynamic ("indy")
Current Status
- active developer community on mailing list, IRC
- programmers assigned to a number of key tasks in JSR 292
Future
- fixnums and tuples handled better in the JVM
Key Issues for Discussion
(please expand cooperatively)