AQuAMan

We just updated to 21.2 and have started having issues when trying to "Analyze Patch"  We are getting the following;

:  An error occurred while evaluating expression: =rule!AQA_ER_getObjectAttributeValueByAttributeName( object: pv!Object, attributeName: cons!AQA_TEXT_OBJECT_ATTRIBUTE_NAME, objectAttributes: pv!objectAttributes ) (Expression evaluation error in rule 'aqa_er_getobjectattributevaluebyattributename': Invalid index (1) for list: valid range is empty) (Data Inputs) 

Hi Ed - I am looking into this issue. Initially I also thought it was because the was related to the patch now containing the 'patches.xml' file (which isn't an Appian design object) but even excluding this (as you previously mentioned) I am still getting failures. The symptom appears to be that the analysis is incorrectly identifying some objects. Since this was working previously my current hypothesis is that one (or more) object type has changed its structure that the XPath is no longer correct.

As an interim you can (as I have just done) cancel any failed process instances as this then allows the analysis process to proceed on the remaining objects. In my example I am cancelling 34 out of 2000 objects, which means I can still successfully analyze >98% of the objects. Given that it's all about risk, then if you're prepared to accept that you don't have automated insight into those that are failing you could still manually examine those if you are risk averse.

I'll come back with an update when I determine the root cause of the error.

Stewart

Hi Ed - I am looking into this issue. Initially I also thought it was because the was related to the patch now containing the 'patches.xml' file (which isn't an Appian design object) but even excluding this (as you previously mentioned) I am still getting failures. The symptom appears to be that the analysis is incorrectly identifying some objects. Since this was working previously my current hypothesis is that one (or more) object type has changed its structure that the XPath is no longer correct.

As an interim you can (as I have just done) cancel any failed process instances as this then allows the analysis process to proceed on the remaining objects. In my example I am cancelling 34 out of 2000 objects, which means I can still successfully analyze >98% of the objects. Given that it's all about risk, then if you're prepared to accept that you don't have automated insight into those that are failing you could still manually examine those if you are risk averse.

I'll come back with an update when I determine the root cause of the error.

Stewart

Any update on this issue?  We tried the fix mentioned by edn0001 but still see this error for a subset of the objects.

Thanks

Steve Z