Page MenuHomePhabricator

transform: Add warnings to metadata when statements are omitted from Wikidata import
Open, In Progress, MediumPublic

Description

Description

When the orchestrator imports Wikidata content, by calling one of the Wikidata fetch functions, there are a few situations in which we omit statements from the import, because we are not fully prepared to handle their values. (For example, we cannot handle values that specify Julian calendar dates until that type has been created.) In these cases, we should insert a warning in the metadata that comes back with the results of the function call.

Because the Z22/evaluation result object is not created in transform.js, some re-factoring will be needed to make this possible.


Remove all the non-applicable tags from the "Tags" field, leave only the tags of the projects/repositories related to this task


Completion checklist

Details

Related Changes in GitLab:
TitleReferenceAuthorSource BranchDest Branch
definitions: Add Z594–Z599 orchestrator warningsrepos/abstract-wiki/wikifunctions/function-schemata!399jforresterorchestrator-warningsmain
Customize query in GitLab

Event Timeline

Jdforrester-WMF lowered the priority of this task from High to Medium.May 13 2026, 8:13 PM
Jdforrester-WMF subscribed.

Re-prioritising to Medium during an Engineering Backlog triage. The silent-data-loss concern is real (orchestrator drops Wikidata statements it can't represent — e.g. Julian dates — without surfacing that to the caller), and the deliverable is concrete (emit a warning into the Z22 evaluation-result metadata, modulo the transform.js / Z22-construction refactor the description notes). But 10 months without movement says High wasn't actually driving action; Medium is the honest signal. Related Wikidata-fetch reliability cards for context when this is picked up: T379441 (now Medium — large entities tripping rate limits) and T407827 (Medium — too many statements exceeding orchestrator time limits). All three are different angles on Wikidata-fetch robustness; a person picking up this card should consider whether a shared metadata-warning channel can serve all three rather than building three separate paths.

We will get Julian dates eventually. IMO this is covered by "Wikifunctions is under development."