[Geotools-devel] Classification Functions

[Jody G. <jga...@re...>]

Andrea Aime wrote:
> Which test cases are you talking about? The build is working?
> Oh, do you mean the postgis feaure collection ones depending on
> a database you only have? Those I killed yesterday replacing them
> with tests that do work over the standard testing data (a correctness
> test does not need tables with thousands of data) and yes, they
> are working now (after my yesterday's patches, that is).
Thanks - that is great news. It is one of the things holding udig back 
(see the technical dept page). No the test cases I am having problems 
with are for all of the java implementations of the Classification 
functions; ie process the featurecollection and produce something like a 
histogram.

Let me find a list of them;
- EqualIntervalFunction
- QuantileFunction
- StandardDeviationFunction
- UniqueIntervalFunction

I see that a few specific test cases are "X"ed out and are not being 
run. I remember having problems trying to get some splits going based on 
Strings; the interaction between comparable was killing me.
> SLD was not in the picture during OWS5? We participated in WCS 1.1.1
> cite, WFS 1.1 + xlink cite, KML experiments. We used a bit of SLD 1.0
> in the last one, both us and Cubewerx (SE 1.1 was taken out of the
> question from day one since none of the testers supported it).
Wonder who did the work then, I will be very said if the define 
functions we cannot represent in CQL.
>> Andrea note that the "Concatenate" function takes any number of 
>> strings which is a new trick for our Funtion API.
> I see, interesting... Does it mean we have to change the API (did not 
> look).
That is something that Eclesia is slowly beating me over the head with - 
ask him :-) I think we can handle all the functions they define; except 
possibly their version of a classification function (think something 
like a function that takes a series or "threashold" / "value" pairs ).

Jody