Page MenuHomePhabricator

Atomization and matching messages send back to clients
Open, MediumPublic


Meta: The Issue may be too vague or may not use the "right" terms. Feel free to improve it and/or ask

Need: It should be possible to identify which error message (or any other data send back from the API) concerns which data in different granularities.
This might be particularly relevant when Wikibase is used in Projects aside of Wikidata since they might need e.g. a different UI or different specialized API requests compared to Wikidata.

Problem: When data is entered and send and the API reacts to this data (e.g. with an error message, a success message etc.) the data send back as reaction may concern a specific part of the send data (represented in the UI e.g. via an input field) – e.g. a specific value in a is illicit and that value was part of a larger piece of data which was send to the backend.
It is currently not possible to expose to the user which part of the data or which input field contains the problematic data.

Solution: We should aim for (optionally) atomizing/decoupling UI, data that is send and data which is send back. This would enable to implement more useful error messages, notifications and potentially other functionality.

Related: For error messages: T138748, T138312. For atomization of send data and edit summaries: T153857