Am So., 28. Juni 2020 um 22:45 Uhr schrieb Shiro Kawai <xxxxxx@gmail.com>:
>
> If you want such an extended case, another design choice(*) would be:
>
>  (case key
>     (`(a ,x) #t)
>     ...)

While this syntax is possible, it is not a natural extension to the
syntax of case, I think, because the R7RS version case already quotes
without any explicit (quasi)quote.

> Suppose if you want to generate
>    (case key ((a ,x) `(,x)))
> or
>    (case key (`(a ,x) `(,x)))
>
> With implicit quasiquote
>     `(case key ((a ,',x) `(,,'x)))     ;; need to use different order of unquote-quote before x
> explicit quasiquote
>     `(case key (`(a ,,'x) `(,,'x)))    ;; can consistently use unquote-unquote-quote
>
> I think the former is annoying, at least.   It's up to the macro designer, but if I'm the one to decide, I'd pair up quasiquote and unquote.

I agree that the qq engine is better tailored to outputting the second
syntax. But this is only relevant for macros that want to expand into
this hypothetical version of case. And then only for macros using qq
for output, which is not the best choice anyway.

> Footnote: (case key (`x #t)) would be confusing but should be interpreted as key being checked with quasiquote or x.  The same confusing pattern arises in the implicit version as well, namely (case key (,x #t)).

Yes, one would have to look carefully.