[
https://issues.apache.org/activemq/browse/CAMEL-1867?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Mark White updated CAMEL-1867:
------------------------------
Attachment: JmsPassThroughtJmsKeyFormatStrategyTest.java
I tried running the tests against M3 and indeed the 'HEADER.1' key is added to
the message in M3 but not in trunk, so that appears to be an improvement. But
there is still a problem on the getHeader() in trunk.
The unit test doesn't do enough to show - instead of testing the new header
value, I changed it to test the change of the input header value, this then
causes the error where getHeader(key) is returning a different value from
getHeaders().get(key) and so test fails - code is attached. I also added some
output of the headers in my copy and you can see difference in the two values:
--Start Message Headers (from getHeaders().get(key)) --
HEADER_1=VALUE_2
--End Message Headers--
--Start Message Headers (from getHeader(key)) --
HEADER_1=VALUE_1
--End Message Headers--
> JmsKeyFormatStrategy not used consistently and JmsMessage.getHeader() not
> returning correct value
> -------------------------------------------------------------------------------------------------
>
> Key: CAMEL-1867
> URL: https://issues.apache.org/activemq/browse/CAMEL-1867
> Project: Apache Camel
> Issue Type: Bug
> Components: camel-jms
> Affects Versions: 2.0-M3
> Environment: 2.0-M3
> Reporter: Mark White
> Attachments: JmsPassThroughtJmsKeyFormatStrategyTest.java, log.out,
> MessageTester.java
>
>
> On entry to a route from a JMS endpoint configured with
> PassthroughJmsKeyFormatStrategy the headers with underscores (_) are passed
> through correctly.
> However, when any header value in the message is then changed, the original
> headers are added back into the message using the
> DefaultJmsKeyFormatStrategy. For example, sending a JMS message with headers;
> HEADER_1=VALUE_1
> HEADER_2=VALUE_1
> we do see just these two headers in the message received into the route.
> However, if we then set a header value in the route, say HEADER_1=VALUE_2, we
> now see the headers;
> HEADER.1=VALUE_1
> HEADER.2=VALUE_1
> HEADER_1=VALUE_2
> HEADER_2=VALUE_1
> For some reason in the set header process the original message headers and
> values get added into the message after being passed through the
> DefaultJmsKeyFormatStrategy. This has the implication of unnecessary
> duplication of the headers.
> Also, when a message header is accessed using the getHeader(key) method it
> also goes through the DefaultJmsKeyFormatStrategy to decode the key that is
> passed.
> In the example above, if after we have updated 'HEADER_1' to 'VALUE_2', we do;
> message.getHeader("HEADER_1") this goes through the
> DefaultJmsKeyFormatStrategy and actually gets executed as
> message.getHeader("HEADER.1") which returns VALUE_1 (wrong!).
> Note: if instead we lookup the header through message.getHeaders() which
> returns the all headers in a Map, we can correctly access the 'HEADER_1' key.
> (Why is getHeader(key) even going through the KeyFormatStrategy anyway if
> setHeader() isn't?)
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
