Page MenuHomePhabricator

Investigate Multi Content Revisions
Closed, ResolvedPublic

Description

Investigate how MCR can be used to connect the different components of a recording. E,g, the recording an its annotations or even it's review.

This ensures we makes use of the built in MediaWiki version handling.

Related Objects

Event Timeline

kalle moved this task from Incoming to Sprint on the Wikispeech-Jobrunner board.
kalle set the point value for this task to 8.
Lokal_Profil removed the point value for this task.
Lokal_Profil set the point value for this task to 8.
Lokal_Profil moved this task from Unsorted to Recorder on the Wikispeech-Speech-Data-Collector board.

We need to find or implement some benchmarking on MCR before we do anything.

Is this to be the primary persistency layer for recording annotations, something that completely replace the annotations in the database?

If so, then we probably also need to use this as the primary persistency layer for recorded audio.

We need to implement hook onMultiContentSave to validate the content on edit to catch changes via other channels than the extension. How detailed should this validation be?

Consider the future possibility for storing recording dialect. Should other users be able to update that in the same manner as annotations? This might need already now require careful planning to allow for such a feature to minimize the impact on code and already stored data when introduced in the future. For example, what happens if someone records 1000 prompts with a terrible broken accent but truly think they have a perfect pronunciation? Should users be able to switch for all recordings? Or just the single one? Update the user dialect meta in the database? And in what sort of slot would we store this data? And how do we synchronize this back to the database? When do we trust such an edit? Should the recording user be told about this?

Lokal_Profil removed the point value for this task.Dec 3 2020, 8:48 AM