|Date||14 August 2011, 1030 CET|
|Participants||Peter Gorm Larsen, Nick Battle, Shin Sahara, John Fitzgerald, Marcel Verhoef.|
See detailed report from Shin Sahara (sent to the overture-core mailing list) for the latest on VDMTools from Japan. A large number of smaller-scale but very positive engagements. Shin reported the set up of a VDM Working Group as part of the Dependable Software Forum (DSF)
See detailed report from Kenneth Lausdahl and August Ribeiro (sent to the overture-core mailing list) for the latest news on the Overture IDE. Since the IDE release only a single change has been made optimizing the update of the navigator.
RMs are up in for community discussion. People should actively discuss the outstanding issues.
Action 54/1: LB to place minutes on the Wiki in future.
Release of version 1.1
Agreed to announce this minor release in a “small splash” way to VDM Forum + twitter + all overture users.
See detailed report from Kenneth Lausdahl and August Ribeiro (sent to the overture-core mailing list) for the latest news on the AST Restructuring. This has turned out to be a larger task than originally envisaged.
“As you all might know we started the work on creating a new generated AST with visitor support for Overture. So far we have completed the AST definition and implemented changed VDMJ parser to create the new tree. Implemented the type checker so that it can check all but a few of the examples SL models in the same way as VDMJ does it. It can also check the CSK Test suite where only a few models give errors and about 50 models fail, however almost all of the 50 failures are just an offset in the reported location of warnings and errors which is not a big deal to fix.
The derived AST stuff we have been discussing turned out to be a bad idea (We implemented it for the interpreter). Here is what we discussed doing: We have a tree AST1 which has some structure, then we would create a new AST2 which has the same structure but new Java classes with no relation to AST1. This turned out to be a bad idea because e.g. the interpreter would need some functionality already implemented in the type checker e.g. the type comparator. The result of this was re-implementing the same functionality used to do some work based on the AST (In our case this might be quite a lot of code which then would be duplicate classes just taking a different Java class name with the same structure).
We changed most of the interpreter to the new AST (Which is an extended copy of the AST from the Type Checker). Missing are a few places where parsing and TC is done plus the two visitors evaluating the expressions and statements, however this work is relatively easy. The main part missing is that we need to change the AST so that it both extends the type checker AST and allows new stuff to be added. We have a solution for this but is a bit more complicated than the original idea since Java doesn’t have multiple inheritance.”
No other business.
Next meeting is due on September 18th 2011, 1300 CET