[
https://issues.apache.org/jira/browse/UIMA-5078?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jerry Cwiklik updated UIMA-5078:
--------------------------------
Description:
The BaseHttpDispatcher.dispatchAndWaitForDuccReply() "swallows" exceptions in
try .. catch(). On exception it dumps the trace using e.printStackTrace() never
re-throwing it up to the caller and instead returning null.
Modify to rethrow an exception and also modify all code that uses the method in
question. This method is used by the CLI apparently. Modify every instance of
code that calls the API and add try .. catch to handle an exception in a
specific way. It looks like the method returning null is an error condition the
client code considers as an error. Seems like a bad practice to me to use null
as an error condition if it is caused by an exception in a generic code which
dispatchAndWaitForDuccReply() in fact is. This code is also called from the PM
and perhaps other places and null as a return value obfuscates the real cause.
Each client should handle an exception as appropriate.
was:
The BaseHttpDispatcher.dispatchAndWaitForDuccReply() "swallows" exceptions in
try .. catch(). On exception it dumps the trace using e.printStackTrace() never
re-throwing it up to the caller and instead returning null.
Modify to rethrow so that the caller can handle the problem.
> UIMA DUCC: http dispatch code doesn't rethrow exception
> -------------------------------------------------------
>
> Key: UIMA-5078
> URL: https://issues.apache.org/jira/browse/UIMA-5078
> Project: UIMA
> Issue Type: Bug
> Components: DUCC
> Reporter: Jerry Cwiklik
> Assignee: Jerry Cwiklik
> Fix For: 2.1.0-Ducc
>
>
> The BaseHttpDispatcher.dispatchAndWaitForDuccReply() "swallows" exceptions in
> try .. catch(). On exception it dumps the trace using e.printStackTrace()
> never re-throwing it up to the caller and instead returning null.
> Modify to rethrow an exception and also modify all code that uses the method
> in question. This method is used by the CLI apparently. Modify every instance
> of code that calls the API and add try .. catch to handle an exception in a
> specific way. It looks like the method returning null is an error condition
> the client code considers as an error. Seems like a bad practice to me to use
> null as an error condition if it is caused by an exception in a generic code
> which dispatchAndWaitForDuccReply() in fact is. This code is also called from
> the PM and perhaps other places and null as a return value obfuscates the
> real cause. Each client should handle an exception as appropriate.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
