ASIS : questions about the ASIS status against the Ada reference

ASIS : questions about the ASIS status against the Ada reference

[Hibou57 (Yannick Duchêne)]

Hello ASIS fans,

I'm interested in ASIS as it is an abstraction for representing Ada
sources. One may set up its own abstract representation, but as such a
thing already exist, it is better to use the existing one. Better,
beceause such an interface is a rather big stuff, and it is better to
reuse to avoid errors and get the benefits of long time worked issues.

But I have some doubt about ASIS. Well, at least, one.

Looking at the ASIS package specifications and even more at the ASIS
issues at ada-auth, I began to feel that a lot of efforts are still
running at making ASIS able to fully represent all possible Ada 2005
constructions.

From here, comes these fundamental questions :

- Is the ability of ASIS to represent all Ada 2005 constructions
formally provable ?
- Can ASIS be viewed as another expression of a part of the Ada 2005
reference in a special formalism ?

While I know ASIS is mainly useful for code analysies, audits,
statistics, etc, I though about another question which may give an
answser to these two latter ones : formally speaking (although perhaps
not practicable), is it theorically possible to imagine a compiler
builded around ASIS ? If the answer is Yes, then the answer to the two
previous questions would be Yes as well.

With thanks

Re: ASIS : questions about the ASIS status against the Ada reference

[Pascal Obry]

Hibou57 (Yannick Duch�ne) a �crit :
> While I know ASIS is mainly useful for code analysies, audits,
> statistics, etc, 

Don't forget code generation. For example AWS generates WSDL documents
out of Ada specs.

Pascal.

-- 

--|------------------------------------------------------
--| Pascal Obry                           Team-Ada Member
--| 45, rue Gabriel Peri - 78114 Magny Les Hameaux FRANCE
--|------------------------------------------------------
--|    http://www.obry.net  -  http://v2p.fr.eu.org
--| "The best way to travel is by means of imagination"
--|
--| gpg --keyserver keys.gnupg.net --recv-key F949BD3B

Re: ASIS : questions about the ASIS status against the Ada reference

[Stephen Leake]

"Hibou57 (Yannick Duch�ne)" <yannick_duchene@yahoo.fr> writes:

> Looking at the ASIS package specifications and even more at the ASIS
> issues at ada-auth, I began to feel that a lot of efforts are still
> running at making ASIS able to fully represent all possible Ada 2005
> constructions.

Right. 

> From here, comes these fundamental questions :
>
> - Is the ability of ASIS to represent all Ada 2005 constructions
> formally provable ?

I doubt it. But why do you want to know?

> - Can ASIS be viewed as another expression of a part of the Ada 2005
> reference in a special formalism ?

Probably.

> While I know ASIS is mainly useful for code analysies, audits,
> statistics, etc, I though about another question which may give an
> answser to these two latter ones : formally speaking (although perhaps
> not practicable), is it theorically possible to imagine a compiler
> builded around ASIS ? 

I don't see how; ASIS presumes a compiler that generates the
information ASIS uses. Why build another compiler on top of that?

-- 
-- Stephe

Re: ASIS : questions about the ASIS status against the Ada reference

[Pascal Obry]

Hibou57 (Yannick Duch�ne) a �crit :
> While I know ASIS is mainly useful for code analysies, audits,
> statistics, etc, I though about another question which may give an
> answser to these two latter ones : formally speaking (although perhaps
> not practicable), is it theorically possible to imagine a compiler
> builded around ASIS ? 

I would say yes. The internal representation of the Ada code could be
based on ASIS. Yet I don't think there is a single Ada compiler that has
followed this path.

-- 

--|------------------------------------------------------
--| Pascal Obry                           Team-Ada Member
--| 45, rue Gabriel Peri - 78114 Magny Les Hameaux FRANCE
--|------------------------------------------------------
--|    http://www.obry.net  -  http://v2p.fr.eu.org
--| "The best way to travel is by means of imagination"
--|
--| gpg --keyserver keys.gnupg.net --recv-key F949BD3B

Re: ASIS : questions about the ASIS status against the Ada reference

[Tero Koskinen]

On Sat, 06 Jun 2009 02:25:45 -0400 Stephen Leake wrote:
> "Hibou57 (Yannick Duchêne)" <yannick_duchene@yahoo.fr> writes:
...
> > not practicable), is it theorically possible to imagine a compiler
> > builded around ASIS ? 
> 
> I don't see how; ASIS presumes a compiler that generates the
> information ASIS uses. Why build another compiler on top of that?

There is Gela: http://www.ten15.org/wiki/Ada/Docs

At the moment, it includes (partial) compiler/platform independent
ASIS implementation and I think there is a plan to use ASIS as
an intermediate representation for the compiler.

-- 
Tero Koskinen - http://iki.fi/tero.koskinen/

Re: ASIS : questions about the ASIS status against the Ada reference

[Stephen Leake]

Tero Koskinen <tero.koskinen@iki.fi> writes:

> On Sat, 06 Jun 2009 02:25:45 -0400 Stephen Leake wrote:
>> "Hibou57 (Yannick Duch�ne)" <yannick_duchene@yahoo.fr> writes:
> ...
>> > not practicable), is it theorically possible to imagine a compiler
>> > builded around ASIS ? 
>> 
>> I don't see how; ASIS presumes a compiler that generates the
>> information ASIS uses. Why build another compiler on top of that?
>
> There is Gela: http://www.ten15.org/wiki/Ada/Docs
>
> At the moment, it includes (partial) compiler/platform independent
> ASIS implementation and I think there is a plan to use ASIS as
> an intermediate representation for the compiler.

I stand corrected; ASIS presumes a parser plus name and overload
resolution, not a full compiler.

Interesting project.

-- 
-- Stephe

Re: ASIS : questions about the ASIS status against the Ada reference

[Randy Brukardt]

"Stephen Leake" <stephen_leake@stephe-leake.org> wrote in message 
news:umy8kgys1.fsf@stephe-leake.org...
...
> I stand corrected; ASIS presumes a parser plus name and overload
> resolution, not a full compiler.

Of course, name and overload resolution is the hard part. Most of the rest 
is just trivial details. ;-)

                                   Randy.