Steven Hawkins

@stianst moved the migration note

> @shawkins, whenever you find some time, could you please resolve those conflicts? The changes look good to me, despite the presence of Git conflicts. Thank you for reviewing. It's...

Updated to include an integration test to make sure a filter is picked up.

Can confirm the behavior described above. Will try to see if there's anything that can be done on the keycloak side, or if we'll need to open a quarkus issue.

@vmuzikar @pedroigor digging more, this is related as mentioned in the description to the usage of KeycloakApplication.getClasses. We are hitting - https://github.com/quarkusio/quarkus/blob/11c724f566946b1fa4550e019be24b34c93e4476/independent-projects/resteasy-reactive/common/processor/src/main/java/org/jboss/resteasy/reactive/common/processor/scanning/ApplicationScanningResult.java#L44 filterClasses is defaulting to true because we have...

> Is this something we can fix in 24? Not easily at first glance. This goes against usage assumptions when KeycloakApplication getClasses / getSingletons (singletons have also been deprecated). >...

> I would assume that if somebody actually uses filters, doing something with the KeycloakSession might not be an unusual use case. There seems to be some nuance here. The...

@geoand @pedroigor I'm not sure what exactly to open as the upstream issue yet. The situation we have here is we're allowing users to install their own jax rs ServerRequestFilter...

> You'll probably have to have some special build item to handle this, because this is definitely not standard behavior Thanks @geoand after reviewing the logic more, I've opened a...

@stianst @vmuzikar do we still want this back ported even if filters are not officially supported?