Aleksander Slominski wrote:
> Martin Woolley wrote:
>
>> Hi
>>
>> I'm stress testing an application that uses XPP3. I'm seeing the
>> following stacktrace sporadically:
>>
>> java.io.EOFException: no more data available
>> Thread-5: at org.xmlpull.mxp1.MXParser.fillBuf(MXParser.java:2659)
>> Thread-5: at org.xmlpull.mxp1.MXParser.more(MXParser.java:2666)
>> Thread-5: at
>> org.xmlpull.mxp1.MXParser.parseProlog(MXParser.java:1385)
>> Thread-5: at org.xmlpull.mxp1.MXParser.nextImpl(MXParser.java:1370)
>> Thread-5: at org.xmlpull.mxp1.MXParser.next(MXParser.java:1111)
>> Thread-5: at
>>
com.nmss.ms.agent.api.XppStreamHandler.processDocument(XppStreamHandler.java:67)
>> Thread-5: at com.nmss.ms.agent.api.ApiWorker.run(ApiWorker.java:193)
>> Thread-5: at java.lang.Thread.run(Thread.java:534)
>>
>>
>> I have checked the XML input document and it looks fine.
>>
>> My instincts tell me I've hit some kind of synchronisation issue....
>> but I could be wildly wrong.
>>
>>
> are you using the same parser instance in multiple threads? this is not
> supported. as noted in docs and as it is usual for XML parser: XPP3 is
> not multi thread safe and each thread should have its own instance of
> parser. you can use parser pool (see xmlpull.org addons) if you want to
> recycle parser instances (run some perf test with and without pooling ot
> make sure you get optimal perf)
>
one more addition maybe it is obvious but ... you can use the same
instance in multiple threads *if* you will do all synchronization to
make sure no two threads are accessing parser (and its input) in the
same time.

best,

Alek
>> Are there are known issues that produce this Exception?
>>
>> Thank in anticipation
>>
>> Martin
>>
>>
>>
>>
>> Yahoo! Groups Links
>>
>>
>>
>>
>>
>
>
>


--
The best way to predict the future is to invent it - Alan Kay