Injecting aspects during compilation time (was: RE: [abc-users] Implementing a domain-specific language: LTL over pointcuts)

From: Eric Bodden <eric@bodden.de>
Date: Tue Jan 04 2005 - 10:41:50 GMT

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi, Oege.

Regarding your comment from early november:

Oege de Moor wrote:
>> 3.) Modify the AST I get from abc (polyglot) in such a way that an
>> aspect is generated that reflects the behaviour of a Büchi
>> automaton that would try to evaluate the appropriate formula
>> during runtime.
> Polyglot supports this sort of thing via visitor passes that
> rewrite the AST.
> >
> There are two options here. You could do the translation very early
> in Polyglot, before type checking etc. This is least work, but you
> may get strange error messages that come back from generated code
> (they will have the right position information).
>
> It's perhaps better in the long term to actually extend the
> semantic analysis phases to directly check your new AST, and only
> transform the new features away after all checks have been done.
>
> I'd recommend that you stick in a dummy aspect at the beginning
> (abc expects to know what aspects there are in the program from the
> very start of the compilation), and then generate new advice in
> that
> aspect to simulate the automata.

I am now at the point where I need to generate those aspects. During
the last weeks it bacame clear that best would be to have exactly one
aspect unit per formula. So how can I "inject" those dummy aspects
you are talking about?

Thanks a lot,
Eric

- --
Eric Bodden
Chair I2 for Programming Languages and Program Analysis
RWTH Aachen University

-----BEGIN PGP SIGNATURE-----
Version: PGP 8.0.3

iQA/AwUBQdpy7swiFCm7RlWCEQJnTQCg3kz1JXwvxZA3W1weiXg3vRXfwccAoI2n
6DY7B+ZfJf/i9Tq8AA5qpOZC
=jqTQ
-----END PGP SIGNATURE-----
Received on Tue Jan 4 10:44:41 2005

This archive was generated by hypermail 2.1.8 : Tue Jan 04 2005 - 15:50:03 GMT