[Kst] Vector skip patch ?
netterfield at astro.utoronto.ca
Wed Oct 18 15:09:12 CEST 2006
On Tuesday 17 October 2006 10:59, Adam Treat wrote:
> What I am supposing (and the patch is doing) is that when in count-from-end
> mode that we count back from the end the requested length and the frame we
> arrive at is then f0. We then divide up the length according to skip.
Imagine what then happens if one more frame gets added to the file... f0
increments by 1, and the entire vector changes. So as data comes in, you get
very unnerving flicker, and you take a big performance hit.
The solution we came to was to insist that new f0 = n*skip where n is an
integer ("skip boundries"). This is what the old code did. This is required
for count from end, and so, for consistency, we did it everywhere.
The only change in behavior from the old code we need to make is to ensure
that the last skip-boundry point is read.
More information about the Kst