tried to search 9fans through 9p (9fs 9front.org, look at /n/9front.org/lists/9fans) and it's slow as hell.
how would someone make a dataset exposed over 9p faster to search? provide an index file? how many index formats do you need for different kinds of data? or do you take the OOP route and have an RPC endpoint for searching that just runs stuff on the server?
this raises another question, how do you differentiate between files that are meant to be read as data and files that cause stuff to happen?
without networking, you can mostly just rely on some conventions, like /dev being device files, etc, but when you want to expose some weird service, what do you do? is there a good convention for saying "these are methods you can call, these are objects you can read/write"?
how much complexity can the conventions handle?
@grainloom This problem has been responsible for approximately infinity security holes in http server.
AIUI 9p has no concept similar to CGI. Files that aren't just static data are handled internally by the server and possibly delegated to another server, even another 9p server, so you just write a 9p server that exposes the namespace you want. Filenames don't have to show up in a listing, so you could have a directory that has infinity subdirectories each containing search results.
Cybrespace is an instance of Mastodon, a social network based on open web protocols and free, open-source software. It is decentralized like e-mail.