Revision: 1731
http://javapathfinder.svn.sourceforge.net/javapathfinder/?rev=1731&view=rev
Author: frankrimlinger
Date: 2009-07-07 21:17:39 +0000 (Tue, 07 Jul 2009)
Log Message:
-----------
Fixed several tricky bugs involving local variable display and creation of the caller frame for virtual invocation. Now try1.main() correctly sets up the invocation for System.out.println(I)V. This is as far as we can go without developing abstractions for basic system code. Several observations:
1) MangoJPF has become stable and can stand on its own. Numerous bugs have been fixed in MangoJPF that have not been sent upstream to Mango. Therefore, the time has come to retire the name "MangoJPF" and just go back to plain old Mango. The original Mango, which was the straight port from C++, will be removed.
2) There is no need to get bogged down with clinits in the handling of System.out. The hack is to just treat java.lang.System.out as if it were final. This is serviceable for a *lot* of code. The print routines just append to a make-believe field of out, so that in concrete cases the output is readable.
3) Once basic system abstractions are available, finish Marc's exception code. This should be quick. Then the rest of the baseline loop stuff. Then the rest of the system, up to arraycopy.
4) With all this in place, Mango becomes MangoBaseline, and new copy of Mango marches into the brave new invariant-look-behind case-backtracking jpf-driven rewriter world.
Modified Paths:
--------------
branches/mango/MangoJPF/Mango/src/mango/worker/workFlow/form/model/LocalArrayModel.java
branches/mango/MangoJPF/javapathfinder-mango-bridge/mango/scanner/Invocation.java
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
