> Le 20 sept. 2022 à 23:44, Jacob Hrbek <krey...@rixotstudio.cz> a écrit :
>
>
> In what world is this considered a strongly typed language when I need to do
> these checks like it's a weakly typed one?
>
> (define* (lazy-assign key #:optional (val ""))
> "Assign environmental variable KEY with an optional value VAL, both
> must be a string or a thunk that evaluates to a string
>
> This procedure sets an entry in the @{%makevars} hash table"
>
> (cond ((procedure? key)
> (set! key (key)))
> ((string? key)
> ;; FIXME-QA(Krey): Seems like a wasteful @{format}
> (set! key (format #f "~a" key)))
> ;; FIXME-QA(Krey): Make sure that the error here is clear and descriptive
> (else (make-non-continuable-error)))
>
> ;; FIXME-QA(Krey): Add check for sanity of VAL
>
> (makevars-set key (delay val)))
>
> Instead of something like:
>
> (define* (lazy-assign (string-type key) #:optional (val ""))
> "Assign environmental variable KEY with an optional value VAL, both must be
> a string or a thunk that evaluates to a string
>
> This procedure sets an entry in the @{%makevars} hash table"
>
> (makevars-set key (delay val)))
>
> Notice the (string-type key) meant to declare that it's only expecting an
> input that is a string or evaluates into a string which mitigates the need to
> include sanity checking in every procedure..
>
> or even something like:
>
> (define* (lazy-assign key:string #:optional (var:string "")) ...)
>
> -- Jacob "Kreyren" Hrbek
https://www.gnu.org/software/guile/manual/html_node/GOOPS.html
Writing that denigrates is unlikely to attract helpful responses. I won’t reply
further.
