Getting started is a standard phoenix/ash/postgres application.
To get data imported from the various projects that are currently seeded (see the seeds.ex file), once you've set up and created the data base, go into iex and run AshHq.Docs.Importer.import()
.
The liveview part is not very conventional, I was focused on speed when I first wrote it, and it (as things always do) evolved from something far more simple. The truly complicated part is that the docs are not static content like you would typically see for documentation. They are all stored in a database, because they can all be full-text searched using postgres. Eventually it would make sense to serve the individual doc pages from a CDN or something like that. The interesting pages in that regard is the Docs
page. We do quite a bit of work to make sure that we are only loading the html that will be served for the exact document we are seeing.
The magic of search is done via AshHq.Docs.Extensions.Search
, which modifies resources to make them full text searchable using postgres, amongst other things. See lib/ash_hq/docs/extensions/search/transformers/add_search_structure.ex
for more.
There is also a usage of Ash.Flow
for searching, which can be found in lib/ash_hq/docs/search/search.ex
. It is actually a bit un-ideal, as the better thing to do would be to use a postgres UNION. There are ways to do that with Ash by extracting the resource queries and then making a union, but it was much easier to just put a search action in a flow. Eventually, Ash will support unions.