[CM] R7RS in S7

[Iain Duncan]

Wow that's awesome to hear! (the benchmarking I mean).  I'm having great
success with my S7 in Max/MSP project so far. The ease of embedding, the
thread safety, the first-class environments, and the very straightforward
FFI are real winners in my context, so knowing that it's stacking up well
in terms of speed is fantastic to hear.

My only complaint, after spending a good month kicking the tires on schemes
and trying to figure out what path I'll take, is that it wasn't immediately
obvious how great S7 is. My first impression was that it probably wasn't
under active development  (I know now it is) and I couldn't tell how lively
the community using it is and so on, just because it was hard to see where
and what dev was happening, and the docs are all in one single html page.
If this is an area that I could help with, I would like to do so, as I feel
this work deserves better recognition and adoption. I expect to be using S7
in earnest for some time - the project is likely to be underpinning of a
very slow Masters. (I'm at school on very part time, so I mean on the order
of 5-10 years, haha) I expect to do a parallel implementation in a big
scheme too (either Gambit or Chez), but as I intend the whole thing to be
hackable by people new to scheme who want to muck with the C code (mine,
not S7s), I think S7 is likely to stay the default implementation language,
with the second there for folks who have a need for R6RS or R7RS stuff. I
have spent a long time trying to figure out which way of doing computer
music seriously I want to put effort into, and I'm really excited with the
possibilities of S7 on Max now.

Now that said, I also understand that in Bill's position, I'd be thinking
"well is this guy really going to be around or is this a fleeting
interest". So I would like to demonstrate interest and seriousness of
contribution by helping with documentation that can improve the S7 public
profile while inconveniencing Bill as little as possible. If Bill and those
more involved in this can let me know what might be useful, please do so.
If nothing springs to mind, then I will just start by slowly building a
users guide to Scripting Max in S7, and we can repurpose material from
there later as warranted. I have my (incredibly simple so far) project up
online now, and will make an accompanying repo for the user guide too.
(Really just my working private repo, but available for anyone to read).

https://github.com/iainctduncan/scheme-for-max

In the meantime, I'm earnestly awaiting my copy of Lisp In Small Pieces so
I can learn how to hack on scheme itself :-)

thanks
iain




On Sat, Jan 11, 2020 at 7:33 AM <bil at ccrma.stanford.edu> wrote:

> I can't resist responding to your comment that s7 is
> fast (which has me smiling from ear to ear): even though
> I've made little effort to "game" the standard benchmarks,
> if you use the ranking system the ecraven site uses,
> https://ecraven.github.io/r7rs-benchmarks/, s7 comes
> in 4th among all schemes, compiled or interpreted.
> Chez, Gambit, and Cyclone are faster.  Maybe it's
> tied with MIT-Scheme.
>
> _______________________________________________
> Cmdist mailing list
> Cmdist at ccrma.stanford.edu
> https://cm-mail.stanford.edu/mailman/listinfo/cmdist
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://cm-mail.stanford.edu/mailman/private/cmdist/attachments/20200111/5ddd3957/attachment.html>