Am So., 6. Nov. 2022 um 18:05 Uhr schrieb Marc Nieper-Wißkirchen
<xxxxxx@nieper-wisskirchen.de>:
>
> Am So., 6. Nov. 2022 um 17:57 Uhr schrieb John Cowan <xxxxxx@ccil.org>:
> >
> > LGTM.
> >
> > On Sun, Nov 6, 2022 at 11:49 AM Marc Nieper-Wißkirchen <xxxxxx@nieper-wisskirchen.de> wrote:
> >>
> >> Am So., 6. Nov. 2022 um 16:57 Uhr schrieb Marc Nieper-Wißkirchen
> >> <xxxxxx@nieper-wisskirchen.de>:
> >> >
> >> > Am So., 6. Nov. 2022 um 16:54 Uhr schrieb John Cowan <xxxxxx@ccil.org>:
> >> > >
> >> > >
> >> > >
> >> > > On Sun, Nov 6, 2022 at 6:00 AM Marc Nieper-Wißkirchen <xxxxxx@nieper-wisskirchen.de> wrote:
> >> > >
> >> > >> Fluids have been added to the latest version in my personal repository.
> >> > >
> >> > >
> >> > > To improve interoperability, it would be good to add operations to make a fluid from an existing parameter object and to extract the parameter object from an existing fluid. I suggest the names "fluidify" (which is an unusual word, but standard) and "fluid-parameter".
> >>
> >> So,
> >>
> >> (fluid-parameter <fluid>)
> >>
> >> has to be syntax because a fluid is a syntactic entity.
> >>
> >> The same is valid for "fluidify" ("verflüssigen" in German). It has
> >> to bind a fluid because, at runtime, a fluid is not a value.
> >>
> >> What syntax did you have in mind?
> >>
> >> Akin to SRFI 212 ("alias"), we can have
> >>
> >> (fluidify <fluid> <param-expr>)
> >>
> >> which binds <fluid> to the (!) underlying fluid of <param-expr>.
> >> Therefore, no "define-" because there would be no new fluid created:
> >>
> >> (define p (make-parameter 'initial))
> >> (fluidify x p)
> >> (fluidify y p)
> >> (free-identifier=? x y) ; => #t
> >> (eqv? (fluid-parameter x) p))
> >> (eqv? (fluid-parameter y) p))
> >>
> >> Comments?
>
> When I just began to implement it, I noticed that it could not work.
> The problem is that there cannot be "the" underlying fluid of the
> value of <param-expr> (because the parameter could have been created
> anywhere). So a new fluid (but with the same underlying parameter)
> has to be created by "fluidify".
>
> So let me change the syntax to bind a fluid to
>
> (define-syntax x (fluidify p))
>
> and, analogously,
>
> (let-syntax ([x (fluidify p)])
> ...)
No, this is neither good as p should not be evaluated at expand-time.