s3 storage for the index files

[jippi]

Hey!

Would be fantastic if doctree serve could use s3 as filesystem backend - would allow me to use doctree index in CI and upload the files to s3 where they could be served, allowing a central highly-available doctree frontend for all projects

Maybe via a filesystem abstraction layer like https://github.com/usmanhalalit/gost , https://github.com/lytics/cloudstorage , https://github.com/graymeta/stow

Note: my use-case only covers frontend part, assuming i can s3 upload the indexed file to the right path and it would just work

[slimsag]

Thanks for filing this! I can't say for sure right now if this will make sense or not. The reason being: search.

I really need to figure out how to do good/nice search indexing for the docs, and that will likely involve leveraging local SSD performance (otherwise it'd require a lot of memory.)

I do think finding a way for this to be super simple to deploy, easily turned into an HA service, and (most importantly) easy to keep your projects indexed is extremely important though.

I'll keep this issue open and revisit once I have a better idea of how search in doctree will look / be implemented.