Skip to main content

"Pin compatible" with Xerces

3 replies [Last post]
kalnayak
Offline
Joined: 2005-06-06
Points: 0

In my perspective, Fast Infoset will get wide adoption only when someone develops an engine that is "pin compatible" with Xerces (or should I say JAXP complaint!).

In fact, this should be able to transparently handle XML or binary XML without the application knowing the format and reuse the same, say, SAX events handlers.

Can anyone point me to some place (other than the code) where I can get some idea, how far we are from this fantasy. More specifically, I wanted know if there is a Status page or a TODO feature list .

(The comment "FastInfoset SAX package does not implement factory, a document parser is instantiated directly" in one of the SAX samples, triggered this question.)

-Sridhar Guthula

Reply viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.
kalnayak
Offline
Joined: 2005-06-06
Points: 0

Thanks for the info.

I will keep an eye for the JAXP layer for FI. My situation is such that I cannot touch the application, yet give all the benefits provided by binary XML.

-Sridhar

sandoz
Offline
Joined: 2003-06-20
Points: 0

> I will keep an eye for the JAXP layer for FI. My
> situation is such that I cannot touch the
> application, yet give all the benefits provided by
> binary XML.
>

OK.

JAXP is quite plugable in terms of the factories that prototypes/development could be done separate from any JDK JAXP implementation.

Would you be interested on working on such a solution at java.net?

Paul.

sandoz
Offline
Joined: 2003-06-20
Points: 0

Hi Sridhar,

We have added a number of tests to the project that test
round trip conformance. Part of these tests check that
the same SAX events are produced from an XML document or
an equivalent fast infoset document. The process is as follows:

1) Parse an XML document to produce SAX events
2) Convert the XML document to an FI document using the
FI SAX serializer
3) Parse the FI document to produce SAX events
4) Compare the SAX events produced from 1) and 3)

The SAX events from 1 and 3 should be equivalent.

All the XML documents we have tested so far pass the SAX-based tests.

The project area:

https://fi.dev.java.net/source/browse/fi/RoundTripTests/

contains a large set of XML files to test against and run to produce a test report.

> Can anyone point me to some place (other than the
> code) where I can get some idea, how far we are from
> this fantasy. More specifically, I wanted know if
> there is a Status page or a TODO feature list .
>

I need to update the web page with the latest information and I will also add a TODO list.

Currently we have been concentrating on getting all features of the FI specifiction implemented.

> (The comment "FastInfoset SAX package does not
> implement factory, a document parser is instantiated
> directly" in one of the SAX samples, triggered this
> question.)
>

There has been no work to integrate FI into JAXP and we need to think carefully about this.

Fast infoset documents are designed such that the first couple of bytes are different from XML documents (encoded using the most common character encoding schemes). Thus it could be possible to "plug-in" to JAXP and enable dynamic choosing of parsers.

So currently a developer does have to use two different ways to initiate parsing. But once the parser is instantiated the the same XMLReader interface can is used.

Hope this helps,
Paul.