Change Details

This is a part of schema change of wb_terms table (custom table of Wikibase extension). We would like to know if it would be possible to add empty (all NULLs) column to wb_terms without needing to make Wikidata DB read-only. https://gerrit.wikimedia.org/r/#/c/341322/1/repo/sql/AddTermsFullEntityId.sql is a draft of the change we want to do. There will be no code writing to and reading from the new column until the process of adding the column is complete. **Background:** General plan for the whole schema change is: 1. Add term_entity_id_s VARBINARY(32) to wb_terms. All values NULL 2. Populate term_entity_id_s based on existing data in wb_terms 3. Create index on term_entity_s 4. Create index on term_language, term_entity_id_s, term_type, term_search_key(16) 5. Drop old column term_entity_id from wb_terms. 6. Rename term_entity_id_s to term_entity_id

This is a part of schema change of wb_terms table (custom table of Wikibase extension). We would like to know if it would be possible to add empty (all NULLs) column to wb_terms without needing to make Wikidata DB read-only. https://gerrit.wikimedia.org/r/#/c/341322/1/repo/sql/AddTermsFullEntityId.sql is a draft of the change we want to do. There will be no code writing to and reading from the new column until the process of adding the column is complete. **Background:** Current schema change plan (as suggested by Jaime) 1. Create a new column with new indexes 2. Populate new column at the same time than the current one (ideally with code, but if you do not want to touch code, it can even be made with a trigger) + populate old values with maintenance 3. Deploy the new code using only the new column 4. At some point in the future, drop the old column (it can be made all null and it will save space already) -low priority

This is a part of schema change of wb_terms table (custom table of Wikibase extension). We would like to know if it would be possible to add empty (all NULLs) column to wb_terms without needing to make Wikidata DB read-only. https://gerrit.wikimedia.org/r/#/c/341322/1/repo/sql/AddTermsFullEntityId.sql is a draft of the change we want to do. There will be no code writing to and reading from the new column until the process of adding the column is complete. **Background:** General plan for the whole schema change is:Current schema change plan (as suggested by Jaime) 1. Add term_entity_id_s VARBINARY(32) to wb_terms. All values NULLCreate a new column with new indexes 2. Populate term_entity_id_s based on existing data in wb_termsnew column at the same time than the current one (ideally with code, but if you do not want to touch code, it can even be made with a trigger) + populate old values with maintenance 3. Create index on term_entity_sDeploy the new code using only the new column 4. Create index on term_language, term_entity_id_s, term_type, term_search_key(16) 5. Drop old column term_entity_id from wb_terms. 6.At some point in the future, Rename term_entity_id_s to term_entity_iddrop the old column (it can be made all null and it will save space already) -low priority