On Fri, Feb 14, 2020 at 11:16 PM Lassi Kortela <xxxxxx@lassi.io> wrote:
> What's the overall string situation in R7RS-large?

John made a good summary here (and he also talked about this problem
with disjoint string types)

https://groups.google.com/d/msg/scheme-reports-wg2/JYHXkDb-ng4/-zgMIk6YBQAJ

> The Haskell community
> has struggled with juggling different string types. (In their case, they
> had lazy strings which are basically linked lists of characters, which
> weren't fast enough so they added maybe 2 or 3 other string types.) SRFI
> 135 looks like the API is disjoint with the ordinary Scheme string API
> which would lead Scheme down a similar road.
>
> > I don't know if we could have a document cover ffi basics for
> > different schemes? I know they are all different, but having used some
> > I think the way they approach ffi is similar (and hopefully we could
> > standardize the common parts)
>
> Strongly in favor. Even more generally, An FFI cookbook would be very
> useful. In my experience, the basics are reasonably easy to grasp but
> the advanced features are complex and often undocumented. Perhaps the
> guide could start with simple stuff and move on to advanced concepts?

Yeah something as simple as wrapping C functions with only primitive
types (no pointers or structs), but for most schemes (except ones
based on "big VMs" like Kawa), would be a good starting point. Then we
can slowly build up (of point to relevant user guides)

> Have you encountered r6rs-pffi, a portable FFI covering several
> implementations? <https://github.com/ktakashi/r6rs-pffi>. IMHO it's a
> good start and would be nice to expand upon. Sagittarius Scheme's author
> maintains it, but a lot of his Scheme code is portable to other
> implementations.

Nope. I'm mostly on r7rs... Will have a look.
--
Duy