Performance

[Peter Murray-Rust]

The question of caching documents (Trees) has been raised on XML-WG and whilst
I'm a believer in leaving performance until late in the day, it's useful to
think about it now. (JUMBO is 
presently well-named as large, slow, and dumps on you at regular intervals).
With AUTO/NEW one you start generating a lot of Trees, especially if you have
a 'Back' button.  Leaving them all around will (presumably) eat up memory
quite quickly.  Perhaps a PI could give hints that a tree was likely to be 
re-used?

JUMBO is slow partly because there is 500 Kbyte of *.class (it covers everything
from matrix algebra through stats, graphics, molecules as well as having a 
general XML GUI.)  In cases where you know that there is only likely to be one
tree the parser could be garbage collected, for example after it had been used.
Again this could be done through a PI?  Another problem is when you load new
class libraries for a new DTD.  E.g. when I have finished with molecules and
go on to PLAY, I'd like to get rid of the molecule *.class.  *** is there a 
way of telling Java to collect *classes* as well as objects?***  (Or will this
only happen when the last object of that class has been garbaged??

	P.

-- 
Peter Murray-Rust, domestic net connection
Virtual School of Molecular Sciences
http://www.vsms.nottingham.ac.uk/

xml-dev: A list for W3C XML Developers
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/
To unsubscribe, send to majordomo at ic.ac.uk the following message;
unsubscribe xml-dev
List coordinator, Henry Rzepa (rzepa at ic.ac.uk)