Also the category/categories for each SRFI need to be specified in the
manual metadata, but I recall you had a fairly complete list of those.
Yes, they're all in "srfi-data.scm".
I'm using Heroku to host the server because they make it really easy and
free, but the code is not really tied to Heroku. Any server where we can
run Racket and PostgreSQL/MySQL should work with some setup. We could
make a Docker container but I don't know if it's worth the effort. I
gave up on the serverless (AWS Lambda / Google Cloud Functions) approach
because it's so easy to run Racket on Heroku. Serverless could still be
made to work but I don't think it would be any easier or cheaper.
Heroku can be set up to serve sites under a custom domain (e.g.
schemers.org) but I don't know if it matters. The current domain is
<our-server>.herokuapp.com which is what they give by default. On a
custom domain we'd need to figure out how to set up SSL so that https://
URLs work.
I use a Linode instance for
srfi.schemers.org. It's a virtual Linux instance. I use MIT Scheme for my own work, but I'd be happy to try Racket for this purpose. We could probably run it all there.
Another possibility would be to convert everything to statically served files so that Nginx could work its speed magic on them. Since SRFIs change rarely, that would be practical. It also reduces the attack surface and maintenance burden of the server.