Hi,

Although this may seem a very obvious question, none of the obvious answers are true. I think many Gambit users would benefit of really knowing how to use this, since there is barely any code that you can use as reference, no documentation and no google results:

https://github.com/search?l=Scheme&q=opaque%3A&ref=advsearch&type=Code
https://www.google.es/search?q=gambit+opaque&oq=gambit+opa&aqs=chrome.0.69i59j69i57j69i60j0j69i64.1665j0j1&sourceid=chrome&es_sm=91&ie=UTF-8

False hypothesis I could come up with:

* Opaque types make their fields not accessible to extending types so you can't access them using the parent type's procedures
* Opaque types make the fields unprintable by default
* Opaque types have R6RS semantics: [The opaque? flag must be a boolean. If true, the record type is opaque. If passed an instance of the record type, record? returns #f. Moreover, if record-rtd (see “Inspection” below) is called with an instance of the record type, an exception with condition type &assertion is raised. The record type is also opaque if an opaque parent is supplied. If opaque? is #f and an opaque parent is not supplied, the record is not opaque.] This doesn't hold true either for Gambit's types and you don't have such inspection procedures.
* Opaque makes the predicates yield #f for instances of the type always
* Opaque makes the predicates of the parent types yield #f for inherited types' instance
* Opaque disables the generation of a type exhibitor
* Opaque generates different code somehow. At least, I can't see a single difference.
* Opaque types can't be serialized and deserialized with object->u8vector

Is there any other obvious possibility that I'm missing out?

Thank you








On Tue, Apr 15, 2014 at 8:54 PM, Álvaro Castro-Castilla <alvaro.castro.castilla@gmail.com> wrote:
That's an awesome trick, Marc. Thanks!

But still.. I don't see any difference:

> (define-type context
  id: a0000000-0000-0000-0000-000000000000
  constructor: macro-make-context
  macros:                                                                                                                                                                                         
  ;; opaque: WITHOUT OPAQUE
  unprintable:
  a)

(begin
  (##define-macro (macro-make-context p1) (##list '(let () (##declare (extended-bindings)) ##structure) ''#<type #19 context> p1))
  (##define-macro (make-constant-context p1) (##define-type-construct-constant 'make-constant-context '#<type #19 context> p1))
  (##define-macro (context? obj) `((let () (##declare (extended-bindings)) ##structure-direct-instance-of?) ,obj ,''##type-1-a0000000-0000-0000-0000-000000000000))
  (##define-macro (context-a obj) (##list '(let () (##declare (extended-bindings)) ##direct-structure-ref) obj 1 ''#<type #19 context> #f))
  (##define-macro (context-a-set! obj val) (##list '(let () (##declare (extended-bindings)) ##direct-structure-set!) obj val 1 ''#<type #19 context> #f)))

> (define-type context
  id: a0000000-0000-0000-0000-000000000000
  constructor: macro-make-context
  macros:
  opaque: ;; WITH OPAQUE
  unprintable:
  a)

(begin
  (##define-macro (macro-make-context p1) (##list '(let () (##declare (extended-bindings)) ##structure) ''#<type #20 context> p1))
  (##define-macro (make-constant-context p1) (##define-type-construct-constant 'make-constant-context '#<type #20 context> p1))
  (##define-macro (context? obj) `((let () (##declare (extended-bindings)) ##structure-direct-instance-of?) ,obj ,''##type-1-a0000000-0000-0000-0000-000000000000))
  (##define-macro (context-a obj) (##list '(let () (##declare (extended-bindings)) ##direct-structure-ref) obj 1 ''#<type #20 context> #f))
  (##define-macro (context-a-set! obj val) (##list '(let () (##declare (extended-bindings)) ##direct-structure-set!) obj val 1 ''#<type #20 context> #f)))


There is no difference!
I noticed that some attributes produce no difference in the output, but will somehow set some internal flag of Gambit. That's the case of :unprintable, which makes a difference when pretty-printing the object but produces no difference in the generated function/macros. If this is the case, what does opaque: actually do?


Thanks a lot!








        Álvaro Castro-Castilla
          Digital Experiences.   --   visit The Blog





On Tue, Apr 15, 2014 at 4:57 PM, Marc Feeley <feeley@iro.umontreal.ca> wrote:

On Apr 15, 2014, at 9:59 AM, Álvaro Castro-Castilla <alvaro.castro.castilla@gmail.com> wrote:

> Hi!
>
> I've been figuring out all the attributes available with define-type in Gambit. Testing and thanks to:
>
> https://mercure.iro.umontreal.ca/pipermail/gambit-list/attachments/20090226/af2ee44c/attachment-0001.txt
> https://mercure.iro.umontreal.ca/pipermail/gambit-list/2004-December/000043.html
>
> I could understand all but one: opaque. The ways to "learn" how things work with macros producing code is generally the trick (pp (lambda () (macro...)) 'asdf). But what about macros that produce macros? In that case you get an empty form with 'asdf, so no way to know what has been generated.
>
> That of course applies to define-type. There is no difference in generated code with opaque: keyword, so I assume that it only affects non-generative types. Can I see the generated code in such case?
>
> Reading Gambit's code in nonstd.scm doesn't clarify things either.
>
>
> Thank you!
>

Try:

(set! ##define-type-expansion-show? #t)

Marc