Re: Large content rendering in XSLT

> We would like to transform the XML data into HTML and TEXT output and
> construct a Mimemessage using java. One of the element would have more
> of Text data like something run into 3-5 page of text. We would like

Depending on how complicated your logic for the transformation needs to
be, your description so far appears to be reasonable for an application
of XSLT.

Do you need to simply output the text portions of the document, or do you
need to actually evaluate the text?  I ask because heavy manipulation
of large amounts of xs:string data might not be the best thing to do
in XSLT alone, depending on the complexity (e.g., if you're having to
perform decoding operations).

> to format the text into HTML and text and tranform the same .
> I have taken refund policy  as example ,cause it is easy to understand
> the context.
> Application upstream does not want send text data for every request.
> It should be presisted and ready to use .

I'm not sure what you mean when you say an upstream application does not
want to send text data for every request, or how you think this may have
an impact on the question you are asking.  If you could clarify by
perhaps sharing some example XML and brief descriptions of the kinds of
manipulations you will need to perform, that might help people here give
you some advice.

You might also want to give us an idea of what your performance and
resource requirements are (E.g., do you require response times in seconds,
milliseconds, or microseconds? Are there are limits on memory usage?)

I'm not sure how helpful this is, but I can describe a fairly simple
XSLT transformation we are running here which might give you a sense of
the size of documents which can be reasonably manipulated with XSLT.

In this simple system we have a number of documents which have a
hierarchical relationship.  They are Issues which have child Article
documents.  We have written an XSLT stylesheet which can dynamically
retrieve a given Issue and all its child articles, returning a composite
document.  The XSLT actually retrieves all of its documents over HTTP
from a secondary server (so it isn't reading the files off of local disk),
running each through a fairly light transformation.

Using Saxon-SA 8.x on a dual-core AMD Opteron 1210 machine w/ 8 gb
memory, it is possible on the first request to transfer a 4-megabyte
compound issue document (composed of about 335 individual documents)
in a little under 5 seconds, with an initial response (meaning when the
first part of the document starts coming back), in .03 seconds.

After the first request the individual documents have been pulled into
a cache (they've been converted into the native Saxon TinyTree format),
and the same request takes, on average, around 2.1 seconds to transfer.
As with the initial request, Saxon is able to start feeding back the
response within .03 seconds.


Jim

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
James A. Robinson                       jim.robinson@xxxxxxxxxxxx
Stanford University HighWire Press      http://highwire.stanford.edu/
+1 650 7237294 (Work)                   +1 650 7259335 (Fax)