users@fi.java.net

Re: Difference betweenJWSDP FastInfoSet.jar and fi.dev.java.net stable release

From: Paul Sandoz <Paul.Sandoz_at_Sun.COM>
Date: Mon, 16 Jan 2006 11:16:58 +0100

Tony Seebregts wrote:
> Yes - I'm just using the FI.jar, along with jsr173-api.jar and
> sjsxp.jar. (It works like a dream BTW - solved a major problem for me ,
> so *thanks*).

Great, that makes me really happy :-)


  If the source for the JWSDP 1.6 version is available
> somewhere I can try and isolate the error for you.
>

Thanks for the offer, but i think there is no need, because of the way
JWSDP 1.6 buffers SOAP messages this error will never occur (phew!).

Paul.

> regards
>
> Tony
>
>
>
> Paul Sandoz wrote:
>
>
>>Hi Tony,
>>
>>Tony Seebregts wrote:
>>
>>
>>>Hi,
>>>
>>>Don't know if its relevant/important - the FastInfoset.jar file from the
>>>Java Web Service Developers Pack 1.6 throws the following exception on
>>>code that works perfectly with the stable release from fi.dev.java.net:
>>>
>>
>>I presume you are using just the FI jar in JWSDP 1.6? and you have not
>>seen this error occur when using JAX-RPC?
>>
>>
>>
>>>java.io.IOException: Full bytes not read
>>> at
>>>com.sun.xml.fastinfoset.Decoder.ensureOctetBufferSize(Decoder.java:1248)
>>> at
>>>com.sun.xml.fastinfoset.Decoder.decodeOctetsOnSeventhBitOfNonIdentifyingStringOnThirdBit(Decoder.java:723)
>>>
>>> at
>>>com.sun.xml.fastinfoset.stax.StAXDocumentParser.next(StAXDocumentParser.java:402)
>>>
>>>
>>
>>Between 1.0 and 1.0.1 i did fix some code in ensureOctetBufferSize.
>>Previously the read would fail early if it could not get all the bytes
>>it required, and this was causing issues in certain streaming cases. I
>>updated this code to perform repeated reads if necessary and throw an
>>EOF exception instead.
>>
>>When using JWSDP 1.6 i do not think this error can occur because the
>>whole the message is buffered as byte[] before being parsed (not very
>>efficient i know!). So ByteArrayInputStream.read() will always return
>>the correct number of bytes or if not it will be an error because the
>>messages is terminated incorrectly.
>>
>>Thanks for the update, this is very useful to know.
>>Paul.
>>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe_at_fi.dev.java.net
> For additional commands, e-mail: users-help_at_fi.dev.java.net
>

-- 
| ? + ? = To question
----------------\
    Paul Sandoz
         x38109
+33-4-76188109