Started playing with a fully static search option.
I'm finding Meilisearch to be a pain over time, especially with the multiple moving parts involved (server, scraper, search libs) which would often change or be deprecated.
Would prefer something a bit more in our control, and ideally static.
This PR plays with webidx, which builds an SQLite search DB via a perl script, then allows front-end search via sql.js.
This does download the whole DB, but from testing it's only about 1MB total, and almost halves with gzip compression.
There is this option to stream via ranging across the DB object, which could be a later goal if worthwhile.
Download only occurs on search anyway.
Todo
- Mark webidx.js as modified from original source
- Update front-end with proper UI integration
- Sort/split results into docs/blog/hacks/other
- Remove redundant UI code in webidx.js (If any)
- Remove old search code, docs, files, guidance
Started playing with a fully static search option.
I'm finding Meilisearch to be a pain over time, especially with the multiple moving parts involved (server, scraper, search libs) which would often change or be deprecated.
Would prefer something a bit more in our control, and ideally static.
This PR plays with [webidx](https://github.com/gbxyz/webidx), which builds an SQLite search DB via a perl script, then allows front-end search via [sql.js](https://github.com/sql-js/sql.js).
This does download the whole DB, but from testing it's only about 1MB total, and almost halves with gzip compression.
There is [this option](https://phiresky.github.io/blog/2021/hosting-sqlite-databases-on-github-pages/) to stream via ranging across the DB object, which could be a later goal if worthwhile.
Download only occurs on search anyway.
## Todo
- [x] Mark webidx.js as modified from original source
- [x] Update front-end with proper UI integration
- [x] Sort/split results into docs/blog/hacks/other
- [x] Remove redundant UI code in webidx.js (If any)
- [x] Remove old search code, docs, files, guidance