Re: [Axiom-developer] Re: interpreter and with-expression

[Gabriel Dos Reis]

On Wed, 17 Jan 2007, Waldek Hebisch wrote:

| > On Wed, 17 Jan 2007 address@hidden wrote:
| >
| > | > The interpreter (in particular pf2Sex) does not handle category
| > | > definitions (with-expression). Is that by design or an "unfinished part"
| > |
| > | I'm not sure that it is possible to create categories at the
| > | interpreter level. Well, in theory it is since a category is
| > | simply a lisp expression, but in practice I don't think it works.
| > | Categories are a compile-time concept.
| >
| > I'm unsure about the meaning you give to "compile-time" here.
| >
| > Types are compile-time concepts too.  Yet, the interpreter has no
| > problem handing
| >
| >    Foo == Record(f: Integer, b: Boolean)
| >
| > or
| >
| >    Bar := Record(f: Integer, b: Boolean)
| >
| >
| > Furthermore within an axiom session, it is possible to )compile a file
| > and have the result available in the working frame.
| >
| > I tripped over this while working on new packages for Axiom, and
| > it really is confusing.  I would like to know whether there are
| > deep reasons for this, or whether it is because it was not
| > finished.
| >
|
| In Axiom getting a "new" instance of an existing type is easy,
| you just call the constructor.

I'm sorry, that is bogus argument.  Try )compile on a file that has
category and domain definitions.

| But to create a new type you
| have to provide the constructor function.

Of course.  So?

|  Effectively you have
| to "compile" the type.  In principle interpreter could transparently
| pass the definition to the compiler, but IMHO differences in
| interpreter and compiler language would destroy the illusion.

Now, please do nail down those differences and and let see why they
are any useful in the specific issue at hand.

-- Gaby