Scheme in the browser, comparator and implementations metadata Amirouche Boubekki 04 Oct 2019 10:01 UTC
Re: Scheme in the browser, comparator and implementations metadata Lassi Kortela 04 Oct 2019 14:03 UTC

Re: Scheme in the browser, comparator and implementations metadata Lassi Kortela 04 Oct 2019 14:03 UTC

> I have good news.
>
> 1) My prototype comparator finally works under Linux Chrome and
> Chromium, see https://scheme-live.github.io/ff.scm/ It was only tested
> on Ubuntu 19.04. (the problem what with chrome, not Chibi)
>
> 2) I tried Gambit JavaScript backend, it works very well, See
> https://scheme-live.github.io/scheme-fuss/todomvc/ That said, it does
> not support R7RS as of yet. I am waiting 12th or 13th October Gambit
> 30 meeting [0] to decide whether I will continue with Chibi or Gambit.
>
> [0] https://mailman.iro.umontreal.ca/pipermail/gambit-list/2019-September/009152.html

Very nice! Thanks for reporting. Both are responsive, and the TodoMVC
also loads fast.

helpers.js is still one megabyte, but it loads quickly anyway. Do you
know what stage Gambit's tree-shaker is at? That should bring down the
file size a lot.

> 3) I want to change the format used in implementation metadata
> repository [1] to rely on chunks that is n-tuple or lists of n items
> where n is fixed. I am thinking about using n=4. This could help (me)
> demonstrate SRFI-168 capabilities. What do you think? Triples or Quads
> are the defacto standard for representing data (e.g. wikidata,
> dbpedia). I think it will help to rely on a similar formalism.
>
> [1] https://github.com/schemedoc/implementation-metadata/

I would prefer a structure similar to JSON (but using association lists
instead of JSON hashes). Arthur recently converted the official SRFI
metadata to that kind of format, and it's now easy to read and edit:
<https://github.com/scheme-requests-for-implementation/srfi-common/blob/master/admin/srfi-data.scm>.
What do you think of that format?

If you want the implementation metadata in a n-tuple format as well, I
can write a converter and ensure the n-tuple files are kept up to date
:) We should have copies in every format that people find useful.