Details
-
Type:
Bug Fix
-
Status:
Open
-
Priority:
Major
-
Resolution: Unresolved
-
Affects Version/s: None
-
Fix Version/s: Backlog
-
Component/s: Development
-
Security Level: Public (Public: Anyone can view)
-
Labels:
-
Similar issues:
KULRICE-6558Resolve remoting situation for string coercion service used for proposition validation KULRICE-4668Evaluate remote KIM services: implementation KULRICE-3323remote service invocations failing in load balanced scenario KULRICE-5224Convert RuleValidationAttribute to a remotable SOAP service KULRICE-5319Implement caching on remote soap services KULRICE-6509Synchronize with remote services on KSB startup KULRICE-9642Error message not being resolved KULRICE-6511Implement ability to load remote services lazily KULRICE-4666Evaluate "remote" KIM services KULRICE-12874Resolve issue with failing IT tests erroring with HTTP response: status code = 503, status message = [Service Unavailable] -
Rice Module:KRAD
-
KAI Review Status:Not Required
-
KTI Review Status:Not Required
Description
Currently the remote services pass back the error message key for remote attribute validations and then Rice resolves them. However, with truly remote service this will not be possible as the local rice instance won't be able to resolve remote service. Instead the actual error message that is being displayed to the user should be passed back from the remote service.
Claus: Back to my question, does the error builder of RemotableAttributeError support error message with arguments (ie "This error
{0}!")?
Eric: no it does not and it's not supposed to, the idea being that messages will be resolved from local resource bundles before passing them back across the wire
because say an application like kc was producing error messages, it would likely have some external message bundle for i18n, but it should resolve those outbound error messages prior to sending them to rice (in the case of this coming from something like an ActionTypeService, AgendaTypeService, whatever) because Rice has no idea about KC's local error message information and shouldn't need to
Claus: So you are saying the remote service will send the actual error message that is displayed to the user? Because that's not what's happening currently. KRMS is setup to accept an error message key (i.e. peopleFlow.peopleFlowId.require), not the actual error message. KRAD (in ErrorsFields.getMessages()) then looks up the actual message (i.e. "PeopleFlow ID is required.") from the appropriate ApplicationResources.properties file (in this case KEW's).
Eric: yeah, that's not the way it should be working, peopleflow comes from kew, so we should assume that KRMS should not even have access to kew's ApplicationResources.properties
in practice it does because our current modularity sucks 
but imagine that KC was creating a custom action type in KRMS
we won't have their ApplicationResources.properties available to us on the KRMS side
so if we are getting error messages from a RemotableAttributeError, they need to be pre-resolved
now, on the KC side in that case, they would probably do something similar to what you are doing in order to resolve the actual error message before sending it across the wire
does that make sense?
Claus: Uh, oh! We'll need to fix that then before the release. There is the KRMS part and the remote services we'll need to fix. I'll create a Jira for it.
It makes sense and I understand it.
Eric: cool, yeah we can just treat it as a bug
Issue Links
- discovered
-
KULRICE-5894
Improve documentation of AttributeError, especially for error messages
-
- is relied upon by
-
KULRICE-6948
KRMS is treating RemotableAttributeErrors as message keys instead of messages
-
Activity
DataDictionarySearchableAttribute, PeopleFlowActionTypeService and the sample app CampusAgendaTypeService have been updated to resolve error message keys from the KRAD ConfigurationService.
In KIM, IdentityMangementPersonDocumentRule and IdentityMangementRoleDocumentRule are using some interesting syntax to describe the error message which could possibly be some form of serialization for reconstitution in the consumer, so I didn't want to go ahead and update these without getting feedback on it. It looks like these would fit straight into a message format with arguments.
Integrated in rice-trunk-nightly #256 (See http://ci.rice.kuali.org/job/rice-trunk-nightly/256/)
KULRICE-5893 - fix, campusAgendaTypeService problem
Note on this: In order to translate the RemotableAttributeError to the MessageMap, we need to enhance the MessageMap to take in the actual error message (right now you can only add errors with the error key). KRADs ErrorsField will also need modified to check for the actual message in the MessageMap instead of the key.
Hopefully we can keep KNS the way it is, and when we convert KIM to KRAD it will use the new infrastructure.
Aaron,
I would recommend we leave the current KIM error handling the way it currently is. Once we convert to KRAD we can modify to use the RemotableAttributeError.
Jerry
Bulk update of incomplete 2.0.0-b2 issues to just a 2.0 fix version.
Changing the lead to Jeremy since this one is assigned to Aaron.
Moving the lead to Jerry and the module to KRAD. Could we change the behavior on the server side to first look for a message from the ApplicationResources.properties (so KIM will continue to work), and if it is not found, display the literal String? This would work in the interim for both KIM and KRMS, and will continue to work for applications & modules that do this the (new) correct way.
Integrated in rice-trunk-nightly #255 (See http://ci.rice.kuali.org/job/rice-trunk-nightly/255/)
KULRICE-5893 resolved some AttributeError messages via the ConfigurationService