What's Changed * restructure disk management by hermeGarcia in https://github.com/nuclia/nucliadb/pull/1049 * Fix grpc node binding by lferran in https://github.com/nuclia/nucliadb/pull/1054 * Fix data path by jotare in https://github.com/nuclia/nucliadb/pull/1057 * migrations sequence numbers by vangheem in https://github.com/nuclia/nucliadb/pull/1055 * Add annotation related metadata to the API by jotare in https://github.com/nuclia/nucliadb/pull/1053 * Add user id on ask document by lferran in https://github.com/nuclia/nucliadb/pull/1059 * remove unused code from texts and paragraphs by hermeGarcia in https://github.com/nuclia/nucliadb/pull/1058 * Remove reload param by lferran in https://github.com/nuclia/nucliadb/pull/1052 * Fix selection labelsets and expose annotations through metadata by jotare in https://github.com/nuclia/nucliadb/pull/1063 * Validate facets by lferran in https://github.com/nuclia/nucliadb/pull/1061 * Use stored model min score by lferran in https://github.com/nuclia/nucliadb/pull/1064 * Feature flag min score by lferran in https://github.com/nuclia/nucliadb/pull/1065 * fix pydantic version req by vangheem in https://github.com/nuclia/nucliadb/pull/1066 * remove stop and reload by hermeGarcia in https://github.com/nuclia/nucliadb/pull/1067 * Fix chat context token counting by lferran in https://github.com/nuclia/nucliadb/pull/1069 * Doc fixes by lferran in https://github.com/nuclia/nucliadb/pull/1068 * Implement new training: image classifier by jotare in https://github.com/nuclia/nucliadb/pull/1060 * Add truncate option to chat request by lferran in https://github.com/nuclia/nucliadb/pull/1070 * New dataset image classification by jotare in https://github.com/nuclia/nucliadb/pull/1071 * Fix purge: actually delete shards by lferran in https://github.com/nuclia/nucliadb/pull/1072 * Prevent icon to be recomputed after a patch by lferran in https://github.com/nuclia/nucliadb/pull/1074 * Fix faceted param validation by lferran in https://github.com/nuclia/nucliadb/pull/1076 * Remove limits on chat prompt context in nucliadb by lferran in https://github.com/nuclia/nucliadb/pull/1075 * fix cached dimension by hermeGarcia in https://github.com/nuclia/nucliadb/pull/1077 * Implement shard rollover migration by vangheem in https://github.com/nuclia/nucliadb/pull/1062 * Fix suggestion on link field titles by lferran in https://github.com/nuclia/nucliadb/pull/1078 * better shutdown handling for cluster manager by vangheem in https://github.com/nuclia/nucliadb/pull/1083 * wait for index nodes before attempting to migrate by vangheem in https://github.com/nuclia/nucliadb/pull/1084 * get raw js client during migration by vangheem in https://github.com/nuclia/nucliadb/pull/1085 * Fix python 3.8 support on nucliadb_models by lferran in https://github.com/nuclia/nucliadb/pull/1086 * Migrations should not disrupt the release process by vangheem in https://github.com/nuclia/nucliadb/pull/1087 * delay merge data points deletion by hermeGarcia in https://github.com/nuclia/nucliadb/pull/1088 * retry tikv failures during rollover by vangheem in https://github.com/nuclia/nucliadb/pull/1089 * Sdk feedback + fix bug in find by lferran in https://github.com/nuclia/nucliadb/pull/1090 * Fix at least some flaky tests on search by jotare in https://github.com/nuclia/nucliadb/pull/1091 * Address github action deprecation warnings by lferran in https://github.com/nuclia/nucliadb/pull/1092 * Vectors 2 by hermeGarcia in https://github.com/nuclia/nucliadb/pull/1093 * add vectors 2 to dockers by hermeGarcia in https://github.com/nuclia/nucliadb/pull/1094 * add vectors2 to missing places by hermeGarcia in https://github.com/nuclia/nucliadb/pull/1097 * Fix blocking threads in the node (writer) by jotare in https://github.com/nuclia/nucliadb/pull/1000 * Make choose_nodes not too strict by lferran in https://github.com/nuclia/nucliadb/pull/1099 * Do not return semantic results on exact match only queries by lferran in https://github.com/nuclia/nucliadb/pull/1096 * Cache getting field extracted text in memory by lferran in https://github.com/nuclia/nucliadb/pull/1098 * remove vectors2 by hermeGarcia in https://github.com/nuclia/nucliadb/pull/1103 * Auto load kubernetes cluster index nodes by vangheem in https://github.com/nuclia/nucliadb/pull/1100 * bump docker pip version by vangheem in https://github.com/nuclia/nucliadb/pull/1044 * python refactoring proposal by vangheem in https://github.com/nuclia/nucliadb/pull/1056 * Audit chat by lferran in https://github.com/nuclia/nucliadb/pull/1104 * fix metrics wiring by vangheem in https://github.com/nuclia/nucliadb/pull/1106 * add node metrics back by vangheem in https://github.com/nuclia/nucliadb/pull/1107 * use both service and pod metrics by vangheem in https://github.com/nuclia/nucliadb/pull/1108 * log exceptions when gathering result data by vangheem in https://github.com/nuclia/nucliadb/pull/1111 * fix in cluster config and setup for k8s discovery by vangheem in https://github.com/nuclia/nucliadb/pull/1112 * Implement a better settings module index by jotare in https://github.com/nuclia/nucliadb/pull/1105 * update with current interface by hermeGarcia in https://github.com/nuclia/nucliadb/pull/1114 * Chat: do not query predict chat if no context is found by lferran in https://github.com/nuclia/nucliadb/pull/1116 * use node ip instead of dns by vangheem in https://github.com/nuclia/nucliadb/pull/1115 * Support uppercase characters slugs by lferran in https://github.com/nuclia/nucliadb/pull/1117 * Do not audit processor messages by lferran in https://github.com/nuclia/nucliadb/pull/1119 * Fix blocking threads in the node - Bindings by jotare in https://github.com/nuclia/nucliadb/pull/1113 * Moving and cleaning things around nucliadb_node by jotare in https://github.com/nuclia/nucliadb/pull/1120 * Add migration to populate all fields key in maindb by lferran in https://github.com/nuclia/nucliadb/pull/1101 * remove advance query from external API by hermeGarcia in https://github.com/nuclia/nucliadb/pull/1124
What's Changed * Always install asyncpg as Postgres is the default driver by jotare in https://github.com/nuclia/nucliadb/pull/1015 * "pip install nucliadb" is broken by tarekziade in https://github.com/nuclia/nucliadb/pull/1013 * Instrument metrics in some index node writer tasks by jotare in https://github.com/nuclia/nucliadb/pull/1014 * Safer filenames on HTTP response headers by lferran in https://github.com/nuclia/nucliadb/pull/1017 * Protos modifications for Donut PAWLS by jotare in https://github.com/nuclia/nucliadb/pull/1016 * Add create_kb metric in ingest processor by lferran in https://github.com/nuclia/nucliadb/pull/1019 * Hnsw minscore by lferran in https://github.com/nuclia/nucliadb/pull/1023 * Fix find incomplete results logic by lferran in https://github.com/nuclia/nucliadb/pull/1022 * min score can only be used on the final filtering by hermeGarcia in https://github.com/nuclia/nucliadb/pull/1025 * Improve logging on node errors by lferran in https://github.com/nuclia/nucliadb/pull/1024 * Fixes typo "KnowlwedgeBox". Closes nuclia1012. by AndrewCloete in https://github.com/nuclia/nucliadb/pull/1021 * Chat on a resource (with find) by lferran in https://github.com/nuclia/nucliadb/pull/1009 * Return trace ids on response headers by lferran in https://github.com/nuclia/nucliadb/pull/1026 * optimise suggest by hermeGarcia in https://github.com/nuclia/nucliadb/pull/1027 * Fix cluster teardown by lferran in https://github.com/nuclia/nucliadb/pull/1030 * Chat on docs tweaks by lferran in https://github.com/nuclia/nucliadb/pull/1029 * Fix middleware usage by lferran in https://github.com/nuclia/nucliadb/pull/1031 * Text index indexing optimizations by jotare in https://github.com/nuclia/nucliadb/pull/1032 * Access Control Expose Trace Id header by lferran in https://github.com/nuclia/nucliadb/pull/1033 * Fix SDK by increasing pydantic min version by jotare in https://github.com/nuclia/nucliadb/pull/1034 * More docs for resource operations by lferran in https://github.com/nuclia/nucliadb/pull/1035 * version 1.70 by hermeGarcia in https://github.com/nuclia/nucliadb/pull/1038 * fix trace id format by vangheem in https://github.com/nuclia/nucliadb/pull/1041 * Migration framework implementation by vangheem in https://github.com/nuclia/nucliadb/pull/1011 * bump telemetry by vangheem in https://github.com/nuclia/nucliadb/pull/1042 * fix missing kb during migration by vangheem in https://github.com/nuclia/nucliadb/pull/1045 * More resilient KB creation by lferran in https://github.com/nuclia/nucliadb/pull/1018 * Store model vector dimension and min score on KB creation by lferran in https://github.com/nuclia/nucliadb/pull/1039 * upgrade bindings by hermeGarcia in https://github.com/nuclia/nucliadb/pull/1040 * More docs for standalone config options by lferran in https://github.com/nuclia/nucliadb/pull/1036 * Migration framework updates by vangheem in https://github.com/nuclia/nucliadb/pull/1043 * better word counting in chat by vangheem in https://github.com/nuclia/nucliadb/pull/1050 * Improve slug error message by jotare in https://github.com/nuclia/nucliadb/pull/1051 * Ask a document by lferran in https://github.com/nuclia/nucliadb/pull/1048 * Fix audit trace id formattings by lferran in https://github.com/nuclia/nucliadb/pull/1047
New Contributors * tarekziade made their first contribution in https://github.com/nuclia/nucliadb/pull/1013 * AndrewCloete made their first contribution in https://github.com/nuclia/nucliadb/pull/1021
What's Changed * Consider pg as local driver (from processing PoV) by lferran in https://github.com/nuclia/nucliadb/pull/1007 * Improve ingest grpc service by lferran in https://github.com/nuclia/nucliadb/pull/1001 * fix use of alias fields in sdk by vangheem in https://github.com/nuclia/nucliadb/pull/1005 * Update minimum required Python version by jotare in https://github.com/nuclia/nucliadb/pull/998 * fix TUS PG Support by vangheem in https://github.com/nuclia/nucliadb/pull/1008 * add bump action and tag stable by vangheem in https://github.com/nuclia/nucliadb/pull/1010
- SDK v2, docs improvements - Improved conversation field support and chat integration - Tracing middleware for rust components - Autofilter support - Shard by number of paragraphs/vectors instead of fields - Improve iterating over resource fields - min_score support for vectors - Log context injections