Feature request: Incremental and on demand data loading

[GoogleCodeExporter]

It would be great if I could dynamically load single or multi-resolution data 
for any interval + resolution that comes into view. This data would then be 
added to the original dataset and updated into the view, and would never be 
requested again.
(Bonus points if it were possible to get data for several datasets in a single 
request.)

The main motivation is to minimize loading time by loading only the resolution 
and data needed for the initial view.
Only when the user chooses to navigate (zoom/pan) will additional data be 
fetched. And if most users want to investigate last week for example, it's 
wasteful to load the entire year.

Original issue reported on code.google.com by tom...@gmail.com on 27 Sep 2010 at 5:42

[GoogleCodeExporter]

The data is automatically mip-mapped and can be loaded incrementally (cf 
incrementalHandler) but keep in mind that in most cases this is probably only 
worth it if you're sending thousands of points.  Of course it depends on the 
browsers and network, etc but usually there's an easier way to get the 
bandwidth savings if under 10k points (a more compact date encoding, for 
instance).

Original comment by timepedia@gmail.com on 5 Oct 2010 at 7:47

[GoogleCodeExporter]

I've tried using incremental handler without success.
From what I read in the code, the data loaded by the incremental handler is 
used alongside the original dataset and does not update it. It is also 
overwritten by consecutive incremental requests. This also means that 
incremental data isn't cached.
I also asked about this in the user group: 
http://groups.google.com/group/chronoscope/browse_thread/thread/d7907e08577f08bf

Original comment by tom...@gmail.com on 5 Oct 2010 at 7:55

[GoogleCodeExporter]

My guess is that you can implement caching yourself inside your handler.

Original comment by johan.ry...@gmail.com on 5 Oct 2010 at 9:31

[GoogleCodeExporter]

Not really. A new mip-map will have to be constructed anew for every call. 
Please efer to AbstractDataset.setIncrementalData().

In any case, the integration is extremely limited. The initial dataset is the 
baseline.
As I mentioned in the user group post, you will always see it when you pan and 
zoom before seeing your incremental data.

Original comment by tom...@gmail.com on 5 Oct 2010 at 9:48