I've had a few people email me asking about tree2 status so I'm bumping this message to the top of the list for people to respond to. Please see the archives for the original part of this thread.
sean On 7/26/05, Sean Schofield <[EMAIL PROTECTED]> wrote: > > Some of my larger tree2 forms are incredibly slow - even slower than the > > JIRA. As many other users, I approached the problem with a lazy load > > mechanism. > > > > It turned out the bottleneck was bandwidth. One page was a 1.7 MB download. > > I am now extending the renderer and using CSS instead of HTML attributes > > for the many elements generated by tree2 . Using a few sed scripts, I > > estimate this'll cut bandwidth by over 80%. Another idea I had was to just > > make the renderer send an XML document (it is hierarchical data) and tell > > the client to use XSLT. This would cut bandwidth and push the processing > > task to the client. > > If you have a huge amount of data in a tree its going to take a long > time to render that tree no matter what kind of mechanism you use to > generate the HTML. A 10,000 row table with nested tables is going to > be slow no matter what! > > For such large trees I think client-side toggle (CST) does not make > sense. Just accept the slightly slower response time of the SST > version. CST is the only reason why you would have such a large tree > as you are describing (b/c all of the nodes are needed at once.) Why > else would you be pushing down such a big tree? > > > This does not mean the lazy load issue is not important, I'm saying that > > there are other parts to the puzzle. As far as your ideas about being able > > to specify the a model vendor via the tag attributes, I agree - this would > > be great. But I have always feared that a post like that would get me > > smacked down by any few spec authors who might be listening - after all, > > that literally puts "model" inside a tag called "view" . > > All you are doing is allowing the user the ability to plug in their > own model along with the data. Swing already allows for this with its > tree. Its just MVC. You are not really mixing M + V, you are just > specifying the M that the V should use. > > > Dennis Byrne > > sean >
