- Issue created by @larowlan
- ð§ðŠBelgium wim leers Ghent ð§ðŠðŠðš
AFAICT this issue queue component is a more precise match?
- ðĶðšAustralia larowlan ðĶðšð.au GMT+10
This is most obvious after ð Add e2e tests for multi-value textfield widget in page data form Active because if you have added multiple items via ajax, they're missing if you reload the page
- ðĶðšAustralia larowlan ðĶðšð.au GMT+10
Postponed on ð Add e2e tests for multi-value textfield widget in page data form Active
Bumping to critical because there is data loss here.
Updated issue summary.
This can be based on top of ð Add e2e tests for multi-value textfield widget in page data form Active if someone wants to work on it in the meantime. - ðšðļUnited States phenaproxima Massachusetts
Will start this tomorrow based on the very clear suggestions in the issue summary. Thanks @larowlan!
- ðĶðšAustralia larowlan ðĶðšð.au GMT+10
If you're debugging, this hunk in FormBuilder is where it should pick up the form cache based on the autosave entry.
- Merge request !977Resolve #3514900 EntityFormController should load auto-save state if it exists â (Merged) created by phenaproxima
- ðšðļUnited States phenaproxima Massachusetts
I'm a little puzzled. End-to-end tests are failing because there are several places where they assert that the form build ID has changed. That, apparently, is now inverted by what we're doing here. What's the correct approach here -- do we expect the same form build ID to persist (in which case these assertions need to be changed), or do we expect a new build of the form, populated by data from the auto-save manager?
- ðĶðšAustralia larowlan ðĶðšð.au GMT+10
I think what is happening is the converse of what we're fixing in the issue this is blocked on
IE there's a cached programmed form state, which prevents Ajax rebuilding the form
It felt too easy that this would just work ð
The changes the other issue make to ensure the form state is programmed even if cached. We will need the controller to ensure the form state is not programmed even if cached
Can pick it up Tues morn my time
- ðĶðšAustralia larowlan ðĶðšð.au GMT+10
ðĪĶI'm overthinking it
We can't set input on post requests as it ignores the submitted values
So we just need to wrap the new changes in a check for GET method on the request
Hopefully that's the sauce - Assigned to larowlan
- ðĶðšAustralia larowlan ðĶðšð.au GMT+10
Should this be a beta blocker - there's data loss in HEAD with this - steps to reproduce
* add a multi-value field
* add three values and save a draft to auto-save
* reload page and see the second and third values missing - ðĶðšAustralia larowlan ðĶðšð.au GMT+10
Rebased on top of ð Page status changes from "Published" to "Changed" even when no actual changes are made Active as that makes this much simpler, back to postponed until that is in.
- ðšðļUnited States tedbow Ithaca, NY, USA
- ðĶðšAustralia larowlan ðĶðšð.au GMT+10
This is ready for reviews
- ðĶðšAustralia larowlan ðĶðšð.au GMT+10
This ended up being really straight forward in the end, much simpler than the previous iterations - because of ð Page status changes from "Published" to "Changed" even when no actual changes are made Active
- ð§ðŠBelgium wim leers Ghent ð§ðŠðŠðš
#18++ â nice!
Merged in upstream now that ð PHPUnit Next Major tests failing Active is in, hopefully still green ðĪ
-
wim leers â
committed e56297c9 on 0.x authored by
phenaproxima â
Issue #3514900 by phenaproxima, larowlan: EntityFormController should...
-
wim leers â
committed e56297c9 on 0.x authored by
phenaproxima â
Automatically closed - issue fixed for 2 weeks with no activity.