Conversation
- Updated .gitignore to correct the entry for hs_err_pid. - Added "version": "6.1.3-SNAPSHOT" to package-lock.json and package.json for consistency. chore: refactor Gulpfile and update dependencies for improved build process - Removed unused script-related settings and tasks from Gulpfile.js. - Updated Gulp tasks to use async/await for better handling of asynchronous operations. - Replaced deprecated gulp plugins with modern alternatives (e.g., gulp-postcss, autoprefixer, cssnano). - Enhanced SVG optimization process using svgo with dynamic import. - Updated package.json and package-lock.json to reflect new versions and dependencies. - Bumped project version to 7.0.0-SNAPSHOT in package files and pom.xml.
- Updated Lucene version references in documentation from 4.10.4 to 10.3.0. - Introduced new vector field configuration in `listing-90.xml` for KNN and semantic search. - Added example queries for vector fields in `listing-91.txt`. - Included reindexing commands for vector fields in `listing-92.txt`. - Minor formatting adjustments in existing XML listings.
…registry examples - Added details about on-disk structures related to Lucene vector indexing in backup and configuration documentation. - Introduced a new section for vector model registry in configuration, including examples for registering embedding models. - Updated indexing documentation to clarify additional files created by vector indexing. - Added two new XML listings demonstrating model registration and configuration.
48c41de to
f3b30c7
Compare
line-o
left a comment
There was a problem hiding this comment.
Thanks for providing the documentation for this large change. That definitely helped me to get up to speed.
I would love to see this PR split up into a) a PR that updates the build and b) the Lucene Vector documentation. I offer my help to achieve that.
For the documentation itself, I find it comprehensive and covering the important settings. I think it would help to explain that the docbook articles of the documentation can be used.
The examples would have to be slightly adapted to include the docbook namespace but it makes it a whole lot easier to come to a local working example.
The question I have on the configuration is actually something to discuss in the implementation PR. It is just that I saw it here first.
|
Since the PR combines the two, I think it ll make our live easier to keep it that way here too. It already allows me to have up to date docs with the test containers, if we split this I wouldn't be able to do that. Also the inevitable rebases until we an merge will be easier to handle if we have just one branch / PR to keep in sync. |
Very much targeting 7.0.0
TODO
see eXist-db/exist#6146