Tue Aug 23 14:28:08 CEST 2005
Joe Armstrong (AL/EAB) a écrit :
> Hello world,
> Open questions
> 4) Which style of component specification is best
> CSP, or RCP?
To get intesting guaranties from the contract checker, I would like to
keep some of the state information of CSP.
What I like in UBF is this state information and I think that state view
is consistent with a Finite State Machine approach to concurrent
> 2) Dynamic type checking of these types is trivial - but can a type
> checker *prove* that your program follows a given component spec.
Previous works have demonstrated (I think), that if your erlang program
is not too complex (that is it does not use some dynamic code: eval,
building of atoms, ...). We can prove some correctness.
The remaining problem is that of type equivalence in the presence of
type aliasing. After "Type fileName = str()." do you consider that str
== filename or not?
- If you consider them equivalent then the type checker could <<*prove*
that your program follows a given component spec.>>, but then the only
interest of defining fileName type is documentation of the protocol.
- If you do not consider them as equivalent, the type checker must be
able to relate Erlang value with those types (perhaps the programmer
have to add type annotation to his program).
Fabien Dagnat -- Maître de Conférences
Web : perso-info.enst-bretagne.fr/~fdagnat
Tel : (0|33) 2 29 00 14 09 Fax : (0|33) 2 29 00 12 82
Adr : Ecole Nationale Superieure des Télécommunication de Bretagne
Technopôle Brest-Iroise - CS 83818 - 29238 Brest Cedex 3
More information about the erlang-questions