Am Do., 21. Feb. 2019 um 05:01 Uhr schrieb Alex Shinn <xxxxxx@gmail.com>:
>
> 2019年2月21日(木) 6:26 Marc Nieper-Wißkirchen <xxxxxx@gmail.com>:
>>
>>
>> P.S.: I am currently working on an alternative implementation for SRFI
>> 159; that's why I have been finding all the minor issues I have been
>> reporting recently. (Actually, writing an independent implementation
>> for a SRFI looks like a great way to verify it.)
>
>
> That's fantastic. If not too late, I recommend using CPS so you don't need continuations.

For "trimmed/lazy", one has to jump out of a running formatter. If all
formatters were composed using the primitive "each", I could thread an
explicit continuation argument through the environment monad to
implement "trimmed/lazy" without call/cc.

However, I don't want to forbid formatters composed in a more general
fashion (for example, by using "make-computation" from SRFI 165).

In any case, "trimmed/lazy" would only need escape continuations, and
these should be supported by almost all implementations.

I haven't taken a close look at
"columnar"/"call-with-output-generator" in your implementation yet.
The continuations used there are not pure escape continuations,
however. Maybe one hide the use of first-class continuations by making
use of SRFI 121/158 (especially the "make-coroutine-generator").

Why do you suggest to refrain from first-class continuations, by the
way? They are a part of Scheme after all.

Marc

>
> --
> Alex