Re: OCL Parser API Project



Date view Thread view Subject view Author view Attachment view

From: Stefan Haustein (haustein@kimo.cs.uni-dortmund.de)
Date: Mon 03 Nov 2003 - 16:20:56 GMT


Jrn Guy S wrote:
> Dear Colleagues,
> At UML 2003 in San Francisco, at least 10 different implementations of OCL each sporting 
(...)
> Before we start this, I would like to see some questions discussed, (...)

Hi,

I suppose our parser is one of those 10+. In our case, OCL parsing is currently 
strongly tied with model access, and I am not sure if the benefits would justify 
the extra abstraction layer.

Well, if I could provide the parser with an XxxExp-Factory, allowing me to 
immediately create the objects that are used later in the evaluator without an 
extra transformation step, it may actually be useful for our purposes...

> C) What software platform do you believe the API should be defined for? 

The DOM API is one of the worst I have ever seen (at least from a Java 
perspective). Wouldn't it be sufficient to develop Java and C++ or C# versions 
in parallel?

> D) What process should we choose to conduct this effort? 

Why not simply create a project on sourceforge and use a mailing list for 
discussion? I think creating "official" standards from scratch is not a very 
good idea. The API can be rubberstamped later by an official body when the worst 
issues have been sorted out?

Best regards,
Stefan

-- 
Dipl.-Inform. Stefan Haustein
Univ. Dortmund, FB 4, LS 8   tel: +49 231 755 2499
Baroper Str. 301             fax: +49 231 755 5105
D-44221 Dortmund (Germany)   www-ai.cs.uni-dortmund.de

Date view Thread view Subject view Author view Attachment view