axis-java-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rick McGuire <rick...@gmail.com>
Subject Re: Axis 1.X HEAD broken
Date Fri, 10 Feb 2006 15:24:02 GMT
David Blevins wrote:
> Hey Rick, see below.  Smushing some of your replies into this email 
> (man, you work fast :)
>
> David Blevins wrote:
>
>> Hey Rick, I know you are working on the javamail code.  Is this 
>> something you can help with?  See below....
>>
>>
>>> On Feb 7, 2006, at 6:03 PM, Changshin Lee wrote:
>>>
>>>
>>>> Currently ant clean all-tests fails at test.saaj package test:
>>>>
>>>> TestAttachment
>>>>
>>>> - java.io.IOException:
>>>> javax.activation.UnsupportedDataTypeException: multipart/related; 
>>>> type="text/xml"; 
>>>> start="<DCD809ADF0929061EEBD95D393FAA38E>";;boundary=----=_Part_0.1139364018621

>>>>
>>>>     at javax.activation.DataHandler.writeTo(DataHandler.java:76)
>>>>     at javax.mail.internet.MimeMessage.writeTo(MimeMessage.java:513)
>>>>     at 
>>>> org.apache.axis.attachments.MimeUtils.writeToMultiPartStream(MimeUtils.java:202)

>>>>
>>>>     at 
>>>> org.apache.axis.attachments.AttachmentsImpl.writeContentToStream(AttachmentsImpl.java:493)

>>>>
>>>>     at org.apache.axis.Message.writeTo(Message.java:545)
>>>>     at 
>>>> test.saaj.TestAttachment.testStringAttachment(TestAttachment.java:55)
>>>>
>>> [...]
>>>
>
> http://svn.apache.org/repos/asf/webservices/axis/trunk/java/test/saaj/TestAttachment.java

>
>
>
>>>> TestAttachmentSerialization
>>>>
>>>> - java.io.IOException:
>>>> javax.activation.UnsupportedDataTypeException: multipart/related; 
>>>> type="text/xml"; 
>>>> start="<BB649A1F2EB5FA531EA2D7CDC8E1063F>";;boundary=----=_Part_0.1139364114325

>>>>
>>>>     at javax.activation.DataHandler.writeTo(DataHandler.java:76)
>>>>     at javax.mail.internet.MimeMessage.writeTo(MimeMessage.java:513)
>>>>     at 
>>>> org.apache.axis.attachments.MimeUtils.writeToMultiPartStream(MimeUtils.java:202)

>>>>
>>>>     at 
>>>> org.apache.axis.attachments.AttachmentsImpl.writeContentToStream(AttachmentsImpl.java:493)

>>>>
>>>>     at org.apache.axis.Message.writeTo(Message.java:545)
>>>>     at 
>>>> test.saaj.TestAttachmentSerialization.saveMsgWithAttachments(TestAttachmentSerialization.java:113)

>>>>
>>>>
>>> [...]
>>>
>
> http://svn.apache.org/repos/asf/webservices/axis/trunk/java/test/saaj/TestAttachmentSerialization.java

>
>
>
>>>> Note that activation.jar and mail.jar from Sun solves these problems.
>>>>
>>>
>>>
>>> So I added the javamail and activation jars from geronimo specs 
>>> assuming they would work as well (or as bad) as the sun ones for 
>>> Axis' purpose, but it looks like we were lucky enough to uncover a 
>>> really great test case :)
>>>
>>> You know what's going on here?
>>>
>
> On Feb 8, 2006, at 2:36 AM, Rick McGuire wrote:
>
>> I'm a little confused at the situation here.  Does this work, or fail 
>> using the Sun javamail version?  The stack traces above are clearly 
>> from the Geronimo one.  Without a little debugging, I'm not sure I 
>> understand why it's failing.  The Geronimo javamail version only has 
>> JAF handlers for two MIME types, text and XML...but this IS XML, so I 
>> don't understand why it's not locating the handler.
>> If you can get me a test case to reproduce this, I'll move chasing 
>> this down to the top of my javamail priority list.
>>
>
> Right, it works for the Sun javamail, but not for the Geronimo 
> javamail version.  The failures above are caused by the Geronimo version.
>
> On Feb 8, 2006, at 3:16 AM, Rick McGuire wrote:
>
>> I think I have an idea on what's going on here.  Comparing the 
>> handlers registered in the Sun mailcap file vs. the ones registered 
>> in the Geronimo file, it appears the problem is not with locating a 
>> handler for the "text/xml" datatype, but rather with 
>> "multipart/related".  The Sun version registers a handler for
>> "multipart/*", while the G. version only has handlers for 
>> "text/plain" and "text/xml" (it's missing other types registered by 
>> the Sun version too, but this problem, I believe, is caused by the 
>> missing multipart bit).
>> I just started working on filling on some of the gaps in the 
>> MimeMessage class, and this appears related.  I'll see if I can whip 
>> up a MimeMessage handler.  It may take a day or too, as I need to do 
>> a little research to figure out what's required.  I'd appreciate the 
>> test case, since that will help me figure out if I've got it right or 
>> not.
>>
>
> Ok, so we're a little short on handlers for multipart/*, specifically 
> multipart/related.  Got it.
>
> On Feb 8, 2006, at 5:44 AM, Rick McGuire wrote:
>
>> Writing handlers looks easier than I originally suspected it would 
>> be.  I think I have a handler now that will do the job, but I still 
>> need a test case to verify this fixes the problem (and also debug my 
>> code).
>>
>
> Excellent -- that was a quick "day or two" :)
>
> How much coffee have you been drinking this morning?  This one is a 
> record, i've clocked you at going from confused to having a solution 
> in three emails and 3hrs and 8min.
>
> I've linked the test cases by name above.  This is the source you'll 
> need to grab to run them:
>
> http://svn.apache.org/repos/asf/webservices/axis/trunk/java
>
> It's basically 'ant all-tests' to run them.  You can of course dig 
> through and run just the saaj tests too.
>
> You the man!
I've just opened Jira GERONIMO-1619 with the fix (fixes, actually, as it 
required a chain of at least 6 things to be fixed to clear up the 
problem).  There's also another Jira (1618) I opened to fix a build 
problem that will require resolution for this to be completely operational.

I'm able to cleanly run the tests up to some of the wsdl/interop tests, 
which fail because of compile errors.  The specific identified tests 
appear to be working now.

Rick

>
> -David
>
>
>


Mime
View raw message