We need to research how performant our real-time preview work is, once the biggest part is implemented e.g. measuring how long it takes to fetch each preview, and if the time is increasing or is over some threshold.
Great Slack thread where the Editing team brought our attention to this:
https://wikimedia.slack.com/archives/C02CDL76HEH/p1645485371745649
When we hit the threshold there seem to be different options, for example:
- making the preview non-realtime (and adding a button to trigger it)
- send notification to inform about reduced render times ( because it seems render request can not be cancelled)
- only fire off the Nth request when the (N-1)th has finished, once we know how many requests the API can handle
... what else?
//Acceptance criteria://
- [] test feature on test wiki ( which is most production like, while betawiki isn't)
- [] plan out how to inform the user of slow render times through UI
- [] implement UI additions
- [] find specific treshold for when render time is considered ok or too bad