encourage users, especially new users to save early and save often

[matkoniecz]

Description

especially new users making first edit(s) are likely to be impacted by rate limiting

in bad cases rate limit may make impossible to save their first edit

maybe tutorial can be more insistent on it? maybe iD can guess when there is risk of breaching rate limit and demand saving?

See https://community.openstreetmap.org/t/rate-limiting-for-lines-with-far-too-many-nodes-points/141852

Screenshots

No response

[systemed]

In case it helps, Potlatch traditionally had a timer on the Save button that started when you made an unsaved edit, with the elapsed time displayed from five minutes onwards.

So after five minutes of unsaved work, the button would read "Save (5m)", then a minute later it would change to "Save (6m)", and so on.

Additionally, after 20 minutes, a toast would appear to remind the user to save their work.

(This wasn't to defeat rate limiting, but generally to encourage "save early and save often" - any application can crash, but particularly one running in Flash Player ;) )

[1ec5]

A timer is not a terrible idea. Long-lived changesets can also suffer from merge conflicts. Even apart from the known issues with iD’s automatic conflict resolver, the occasional required manual resolution can be inconvenient too.

[matkoniecz]

still, failure story of "newbie maps several round buildings, goes over quote for the first edits, get stuck" happened few times and I worry that you can manage that in less than 5 minutes

so even if it would take some hardcoding (or extending info provided by capabilities call) I think that hard stop after making X changes would be useful

[1ec5]

#11902 includes a mechanism to break up a changeset geographically to prevent the OSM API from getting indigestion. It’s overkill for the issue it was trying to solve (sparse changesets with huge bboxes) but might be more appropriate as a contingency for a rate limit that we can’t necessarily predict ahead of time.

[matkoniecz]

attempting to autosplit (and deal with edit conflicts happening in the middle) seems to be a horrific headache of much greater complexity

StreetCoplete mostly gets away with doing it but even there it is annoying and it could make some assumptions not viable for iD