This release contains major performance improvements and a new experimental encoder based on the idea of Automerge's columnar encoding automerge/automerge#253 . Using the new encoder only decreases the size of the encoded document. Using real-world data, it reduces the amount of metadata by ~50% - which is quite impressive. But this is not an argument yet to deprecate the current encoder which is much easier to implement.
- implement search-marker prototype (limited usage for now) 6e3b708
- integration refactor with stackHead magic 6e8167f
- micro optimizations in struct reader 3449687
- choose rencoding version at random 3406247
- export YTextEvent - fixes #213 076d550
- remove bare for .. of iterations - fixes #220 bb45816
- yjs implements unpkg - implements #216 5414ac7
- add AbstractConnector interface - implements #215 0b8f032
- Merge branch 'relm-us-ydoctojson' into main dc136ff
- merge with upstream b73a720
- export decodeStateVector - fixes #218 cf420d6
- fix empty type name 859e169
- Implement experimental new encoder 🚀 6c2cf0f
- whitespace 1a942aa
- Add YDoc toJSON 368dc6b
- fix empty parentYKey issue 2151c51
- Remove tsc import doc because typescript is now natively supported bb25ce7
- fix node arraybuffer decoding e31e968
- add compatibility tests 1a49476
- merge upstream b434501
- update sponsors ❤️ d1d8627
- fix gc regression issue & add another breaking condition for the integration algorithm d7a11cc
- Added Sponsor ❤️ 4c48116
- reduce number of variables and sanity checks 😵 6dd26d3
- improve mem usage by conditional execution of the integration part (step throught the integration if there are conflicting items) 6b0154f