Re: [Axiom-developer] Axiom meeting at ISSAC

[Bill Page]

Perviously Stephen Wilson wrote:

 | There is also the question of being able to provide the compiler
 | components in the form of a library which can be used by other aspects
 | Axiom.  For example, one would be able to use the new parser and
 | (forthcoming) type checker to support an IDE.

Previously Gabriel Dos Reis wrote:
> People may note that, in Axiom as of today, one can already get the
> syntax tree from both the old and the new parsers, and in fact use the
> parser as a library -- as has been done the algorithmic
> differentiation project.  You don't need to wait for a
> functional new compiler to do that.

On 01 Aug 2007 15:57:14 -0400, Stephen Wilson wrote:

> Can it incrementally parse input as the user is typing, thus making it
> a usable component in an IDE?  Is the AST itself specified?  Does it
> the representation past the lexical analysis phase contain line and
> column number information?  Is the current implementation easily
> modifiable and extensible?
>

On 8/1/07, Gabriel Dos Reis <address@hidden> wrote:
>
> Stephen Wilson asked the following questions:
> ...
>   (iii) Does it the representation past the lexical analysis phase contain
>        line and column number information?
> ...
> (iii) of course, the answer is yes.  As I mentioned earlier, there are two
>     parser in Axiom, an old and a new.  You're looking at the old compiler.

On 01 Aug 2007 18:03:08 -0400, Stephen Wilson wrote:

> Of course I am looking at the old compiler, it is the only compiler.

Just so I am clear: Are you saying that there is only one Spad
compiler in Axiom but there are two parsers?

> It accepts one AST, not two.

Do you mean that only one of the two existing parsers can be used with SPAD?

> That AST does not contain line/column information.  The AST which
> Axiom uses does not contain line/column information.

Is there a good reason why it does not use the 2nd parser?

> That information is intentionally erased at one point of parsing or another,
> old or new, it does not matter.
> ...

I hope this is a sufficiently technical issue that it is possible to
reach an agreement. Is it possible now to use "the compiler components
in the form of a library which can be used by other aspects Axiom"? Or
not?

If the answer is yes, then some documentation on how to do this would
be much more useful than having to refer to the source code.

Regards,
Bill Page.