[Kst] Fits and Plugins

C. Barth Netterfield netterfield at physics.utoronto.ca
Mon Jul 26 20:37:52 CEST 2004


Andrew (and other interested parties)

I would like to discuss a few things about fits and filters.  Please respond! 
(even with 'I have nothing to say'),

**What is the advantage of having a separate fitcurve type?
**Should we fix issues with the current architecture, or modify the 
architecture?

I will explain the following questions below:

I see that fits are not plugins.  They are a new data object which 'has a' 
plugin. 

So fits show up in the data wizard, as do their plugins.  I find this odd.  
Currently, best I can tell, neither dialog can be used to edit the fit 
parameters (eg, change a linear fit to a quadratic fit).

I thought the idea we had discussed was that there is no such datatype as a 
fit.  Rather, the fit dialog was just a convenient way to create a plugin, 
make a curve from the ouputs, and put the parameters in a label.  
The curve properties of fits would be edited via the curve which was created, 
and the plugin properties via the plugin dialog.

Same for filters.

There are also some issues wrst the fit parameter labels: they are sort of 
legends, and sort of fit parameter lists.  I find some oddnesses with them.
	-The legend line shows up both in the fit parameter label, and the legend
		(should be only in legend?)
	-Multiple fits in a plot produce multiple fit parameter labels, which all lie 
on top of each other, and on top of the legend, unless one moves them
		(shouldn't lie on top of each other)
	-the label format can not be edited - it is what you got or nothing
	-you have to have them all, or none.
	-Plot dialog has two tabs, one for legends and one for fit labels, which are 
very similar, but not identical.

I am kind of stuck with fixing the fit dialog because I don't know where we 
should go here.

Barth



More information about the Kst mailing list