[Laszlo-dev] performance tuning http dataset

[Henry Minsky]

It depends which version of Laszlo you are using. If you are using a SOLO
(unproxied) data load, you are using the builtin flash XML parser. If you
are using a proxied load, it is being encoded to SWF on the server. But we
have been moving to use the Flash client XML parser exclusively (ie..,
dropping the SWF-encoding of XML on the server entirely)

In any case, if the  XML is sent as text to the client, it is parsed by the
Flash parser, which is pretty quick since Flash 6. However, we have to do a
deep copy of the DOM tree to convert it into Laszlo data objects, and that
takes some time. It might be as long as half a second I guess, depending on
if the dataset is large enough to have to have the copying task split across
frames.

We'd love to find a way to avoid copying the tree. We experimented with
splicing our  data element classes to be subclasses of the Flash XML node,
to avoid the need to copy them, but that caused the objects to stop getting
garbage collected.
We have another approach we've been meaning to try but haven't had a chance
yet.
Note: The same issue will arise in the DHTML runtime (yes we're working on
one), where again the native XML parser is pretty speedy, but we need a way
to coerce the DOM objects into obeying the LzDataElement API. In Firefox you
can add methods to the built-in DOM objects, but I have not been able to
figure out a way to do that in Internet Explorer yet. Any suggestions would
be warmly received.




On 3/3/06, Don Dwoske <don at loraxis.com> wrote:
>
> I'm seeing a delay and I can't figure out what's happening.
>
> My laszlo client is requesting a smallish XML dataset from a tomcat
> server.. as I watch the tomcat logs, the response is generated and
> returned very quickly, but from the time it's sent to when Laszlo
> calls my dataset ondata event, nearly a half second has passed.
>
> What's happening during this time period?
>
> I'm assuming that the XML data is being parsed, but could that really
> be taking so long?  The XML dataset is basically a list of 96 items,
> and it's basically flat.. in some of the branches the depth goes to
> about three, but that's it... it's nothing huge.
>
> Any ideas about how to track this down?
>
> (I'm posting to dev as well, as this could be relevant, I guess)
>
> --
>
> ---------------------------------------
> Donald Dwoske
> Software Journeyman
> http://www.loraxis.com/ddwoske
>
> _______________________________________________
> Laszlo-dev mailing list
> Laszlo-dev at openlaszlo.org
> http://www.openlaszlo.org/mailman/listinfo/laszlo-dev
>



--
Henry Minsky
Software Architect
hminsky at laszlosystems.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.openlaszlo.org/pipermail/laszlo-dev/attachments/20060303/5557d82c/attachment.html