#2 also resolves this with Gin and Layout Builder. I'm bumping to Major is it makes LB un-usable. A test may be tricky since this seems to only be surfaced by Contrib modules...
bluegeek9 → credited jhedstrom → .
Moving to NR and adding a patch for use with composer.
Thanks all! I'll cut a new release shortly.
Since this is critical, do we want to commit the current fix (all tests are passing) and deal with the fallout/ideal fix in a follow-up?
Patch for composer. There's a branch in gitlab, but I'm not seeing the button to open a MR.
jhedstrom → created an issue.
jhedstrom → made their first commit to this issue’s fork.
Thanks all!
jhedstrom → made their first commit to this issue’s fork.
Thanks!
jhedstrom → made their first commit to this issue’s fork.
jhedstrom → made their first commit to this issue’s fork.
Thanks all!
jhedstrom → made their first commit to this issue’s fork.
Thanks all!
jhedstrom → made their first commit to this issue’s fork.
jhedstrom → made their first commit to this issue’s fork.
Done!
Thanks all!
jhedstrom → made their first commit to this issue’s fork.
jhedstrom → made their first commit to this issue’s fork.
smustgrave → credited jhedstrom → .
jhedstrom → created an issue.
jhedstrom → created an issue.
Thanks!
Thanks!
Thank you!
jhedstrom → created an issue.
It wasn't, but in order to commit, the failing unit test needs to be updated to pass with this change.
jhedstrom → created an issue.
jhedstrom → created an issue.
jhedstrom → created an issue.
jhedstrom → created an issue.
There should be a test to demonstrate this issue/fix.
These have been taken care of in other issues I think.
Thanks all!
jhedstrom → made their first commit to this issue’s fork.
jhedstrom → created an issue.
jhedstrom → created an issue.
This actually isn't reproducable on a vanilla D10.1 install.
jhedstrom → created an issue.
Sounds good. Here's a patch and a PR :)
jhedstrom → created an issue.
I think this makes sense to commit now if it fixes the issue, since this is critical. It is a bit disconcerting though that neither the issue here, or the change over in 🐛 Invalid context for call to FieldDefinitionInterface->isDisplayConfigurable() Fixed were caught by tests. I'd recommend we do a follow-up to add test coverage for this so we don't regress it again in the future.
#6 appears to have been addressed already in the module. I've merged the other branch above into the issue branch, and am attaching a patch for use with composer.
jhedstrom → made their first commit to this issue’s fork.
jhedstrom → created an issue.
jhedstrom → created an issue.
That code is from a very early port of the module to 8.x (commit 68bbb58
).
There's one more use where the module hard codes display
to that method call in field_permissions_jsonapi_entity_field_filter_access
that should be updated to view
I guess.
(cross post!)
Wow, that's bizarre. I wonder how this was ever working since BaseFieldDefinition
is just checking the definition array (which is presumably also just using view
and edit
):
public function isDisplayConfigurable($display_context) {
return $this->definition['display'][$display_context]['configurable'] ?? FALSE;
}
Hmm, these aren't failing locally for me.
As noted above, the failing test still needs to be addressed in 8.2.
This fixes the error.
This PR (and attached patch for use with composer) adds this option to the field formatter. The UI might be confusing, so we could perhaps use the form states api to hide or show the options as needed...
jhedstrom → created an issue.
Adding a patch for use with composer.
jhedstrom → created an issue.
I've made an attempt to implement the direction in #2. Attaching a patch for use with composer.
jhedstrom → made their first commit to this issue’s fork.
Adding a patch file for use with composer.
jhedstrom → created an issue.
I've fixed the error in #11 (it was due to final
classes not being mock-able in unit tests, so the fix was to mock the interface instead.)
The fatal error noted in #8 has been resolved in the Webform module since 2021 (#3213630), so perhaps tests are being run with a very old version of that module somehow?
I've tested #3 and created an MR with the same code. In manual testing it seems to work with D10.
jhedstrom → made their first commit to this issue’s fork.
Thanks all!