Page MenuHomePhabricator

Timebox dev unhappy path collection [2h]
Closed, ResolvedPublic

Description

Prepare input for T219886: Unhappy Editing Paths

AC

  • collect unhappy paths
  • think about how these should be prioritized from dev point of view
  • think about how these could be clustered (warning vs error) -> to be able to discuss possible ways of handling different levels of "severity" of unhappy situations
  • (think about options to proactively avoid them, e.g. disable save button if there is no internet connection, ...)

Related Objects

Event Timeline

Some super categories of unhappy paths to consider:

  • API responding with error
    • validation error
    • merge conflicts
    • permission issues
    • storage error
    • generic errors
  • API not responding
  • malformed API response
  • time based problems
  • client side validation
  • network problems
  • User edits a language that doesn't exist on Wikibase (case toki pona)

...

WMDE-leszek renamed this task from Timebox dev unhappy path collection to Timebox dev unhappy path collection [2h].Apr 24 2019, 2:27 PM

New tickets created as children of the story.

I think the priority should probably be:

  • lock save button from firing another save event until either clear success or failure (including timeout). Hint that the button is locked / work is going on in the background
  • lock text fields until saving success Hint to user with style change
  • handle all obvious failures generically with a localised "something went wrong; maybe you could try clicking save again" message. Position, size wording etc.. of message
  • Increase clarity of user displayed messages by inferring them from the error returned by either the API or the HTTP client.

I'd love some feedback. There is about 30mins of this timebox remaining given I spent some time chasing unrelated bugs.

Tarrow subscribed.
Tarrow claimed this task.
Tarrow moved this task from Peer Review to Done on the Wikidata-Termbox-Iteration-16 board.

Seems this is done since we have a selection of new tickets and this has been sat unmoved in peer review for a whole iteration.