[Kst] healpix status

Ted Kisner tskisner.public at gmail.com
Sat Oct 29 22:51:35 CEST 2005


On Saturday 29 October 2005 13:40, George Staikos wrote:
>    I'm not too worried about the different configurations even.  Just the
> basics of the loading code would be great to have tested.  We can also use
> these for performance regression tests as they become more full-featured.

ok, I'll see what I can do.  I will say that when using large projection 
matrices, things are slow.  There is no real way around this, since we have 
to go through every pixel in the matrix and assign it a value.  However, I 
have a future idea that will make it better:

1.  leave the resolution at <= the screen resolution.
2.  have the datasource query the current kst2dplot window for the current 
zoom range.  (need to work out details here)
3.  have a button in the datasource that when clicked will "project to current 
window range".  
4.  the user can now zoom in and out until they get to an interesting feature.  
They click the button and the map gets re-rendered at higher detail.

>   Beyond the automated unit tests, it's also useful to have test sessions
> we can just open in Kst to make sure everything looks right.  

ok, so basically an example data file and a corresponding *.kst file that we 
can just open and play around with?

> Once we start to get a useful collection, I'll have an automated
> visual-regression suite implemented.  We have one in KHTML and it 
> works very well. 

so you write some code that actually goes and manipulates the widgets and 
dialogs?  sounds like it would be similar to watching a chess computer play 
itself :-)

-Ted





More information about the Kst mailing list