#
# TODO list for lyngby.
#
# $Id: TODO,v 1.37 2004/06/01 12:37:28 fnielsen Exp $
# $Revision: 1.37 $


- Jan Larsen delay (50% energy) in FIR filter
- TTP in FIR discarded
- 2 Lange-Zeger model: clean up.
- Lange-Zeger noise model implementation
- Lange-Zeger generalized with respect to the paradigm
- lyngby_ui_error and lyngby_message extended to handle long strings
- lyngby_ui_option renamed!?
- Dialog for putting up ui_options
- Tophat filtering
- Trend filtering in GUI
- Gaussian filtering
- Despeckle filtering
- IDL Kirt specification 
- ROI mouse specification
- Setup/data save export
- Masking of pixels depending of the variance: Both in the low AND high end
- SVD filtering  
- Show ROI in lyngby_ui_volume
- Rounding errors in view (at the border)? When the mouse clicks. check!
- Same size of buttons. I choose:
	lyngby_ui_fir_init: 
		height 0.12 x 200 = 24
		width  0.20 x 350 = 70 (Should this be the same?)
		edit control: x at 0.5
		text control: x at 0.48 (right alignment
- Text in right and left view should be UI_COLOR_FRONT
- Common format for the lyngby_log counting. 
	sprintf('Computing ?-test: %d / %d')
	when comparing it to the grid search, ... or is it the grid search LZ
- Every method returns: 
	RESULT_<method> with the parameters that exists
	STRENGTH_<method> aktivation
	DELAY_<method> delay if possible
	PROB_<method> probabil. map if possible. 1-P !!
- The ui_*_init windows should show some indication on whether the 
	fields have been edited or they are valid (i.e. already written
	to the global variables).                   
- lyngby_image: extra button in lyngby_ui_view? log, -log, sqrt, ...
- Error with "Show singular values" in FIR and NNR when there is no paradigm	
	and no data.
- Meta K-means should be able to choose its input.
- Error message if the data_paradigm.mat does not contain a PARADIGM variable
- Vectorizing of Ordinary t-test (to make it faster)
- Vectorizing of Kolmogorov-Smirnov (to make it faster)
- Placement of windows so the fit on a 640x800 screen (protable computer.
	Maybe a function should be used.
- Vectorization of lyngby_xcorr.
- DISCRIM_TMASK and the "edgedrop" design issue should be sorted out.
- Proper tick marks on masked volume with background image.
- Sliders should slider in step of one and it should be possible to reach
	all numbers.
- Change the 'See also' functionnames to capital so that helpwin can parse them.
- Move the CVS information so it is possible for the user to see the version number
	with helpwin and help.
- Save the time course shown in the 'time' window, - not just XN.













