[plt-scheme] Style question

Tomasz wrote:
> Hello.
> 
> I'm writing simple PostgreSQL connection pool (spgsql based). I've
> little scheme experience and I have style question. I've defined pool
> as structure (basically opaque type from user perspective):
> 
> (define-struct pool (max-size queue connection-producer connection-
> closer custodian))
> 
> and i've written constructor-like function (provided from module):
> 
> (define (create-pool max-size connection-producer connection-closer)
>   (let ((pool (make-pool max-size (make-queue max-size) connection-
> producer connection-closer (make-custodian))))
>     ;; some more initialization snipped for clarity
>     pool))
> 
> I was wondering if this is "proper" style. How is constructor-like-
> function idiom typically implemented in Scheme? At first it seemed to
> me that "make-pool" would be more appropriate (maybe with some magic
> prop:constructor function property to initialize) but on the other
> hand I'm not entirely sure that such overloading of standard make-
> <type> meaning is apropriate...

I think exporting the function above as 'make-pool' is entirely 
reasonable, even though it isn't a "pure constructor". If you look in 
the reference, there are a lot of 'make-' procedures that do something 
other than just create a struct instance containing the procedure's 
arguments.

How to do it? Define it as above and export it using 'rename-out':

   (provide pool?
            (rename-out [create-pool make-pool]))

There are other small tricks you can play to get the procedure to print 
with the external name instead of the internal name. Something like the 
following:

   (define create-pool (let ([make-pool (lambda ___)]) make-pool)

or just use 'procedure-rename'.

Ryan