Trouble creating SRFI-9 Record in C

2021-09-10 Thread paul

Good day guile-users,

I am having a struggle with SRFI-9 records.  They look very 
convenient, so i'd like to use them in my Guile scripts.  However, 
i'm not sure how to correctly construct them from C-land.  I have 
something like the following:


```
(define-record-type 
 (make-foo a b)
 foo?
 (a foo-a)
 (b foo-b))
```

In Guile land, that works great.  Now, i want to create a foo in C 
and pass it to a function in the Guile script.  I do something 
like the following:


```
scm_c_primitive_load("foo.scm");
scm_call_5(scm_variable_ref(scm_c_lookup("make-foo")),
  scm_from_utf8_string("blah"),
  scm_from_int32(Int32(42)))
```

However, this results in an error:

guile: uncaught exception:
Wrong type to apply: #

I've tried with and without (define-module foo) at the top of the 
file, that doesn't seem to make a difference.  I've been able to 
work around the issue by defining a wrapper (define (foo-prime a 
b) (make-foo a b)) and using that in C as shown above, but that 
feels ugly.  I'm probably missing something obvious, but trawling 
the mailing list didn't turn up anything i could understand.


Does anyone see what i'm doing wrong, or can i simply not use 
SRFI-9 records in this way?


Thanks, 🙌
p.



Re: Trouble creating SRFI-9 Record in C

2021-09-10 Thread Olivier Dion via General Guile related discussions
On Sat, 11 Sep 2021, paul  wrote:

> In Guile land, that works great.  Now, i want to create a foo in C 
> and pass it to a function in the Guile script.  I do something 
> like the following:
>
> ```
> scm_c_primitive_load("foo.scm");
> scm_call_5(scm_variable_ref(scm_c_lookup("make-foo")),
>scm_from_utf8_string("blah"),
>scm_from_int32(Int32(42)))
> ```
>
> However, this results in an error:
>
> guile: uncaught exception:
> Wrong type to apply: #

Seems like `make-foo` is a syntax-transformer and not a procedure.  You
can not call a syntax-transformer.  I don't think you can do much with a
syntax-transformer in C.

>
> I've tried with and without (define-module foo) at the top of the 
> file, that doesn't seem to make a difference.  I've been able to 
> work around the issue by defining a wrapper (define (foo-prime a 
> b) (make-foo a b)) and using that in C as shown above, but that 
> feels ugly.  I'm probably missing something obvious, but trawling 
> the mailing list didn't turn up anything i could understand.

By making a wrapper, you're effectively creating a procedure that can
use the `make-foo` syntax because it's in Scheme and it's solved a
expansion time.

-- 
Olivier Dion
Polymtl