[daip] Another technical detail

[Eric Greisen]

Elias Brinks writes:

 > I think I stumbled on another small problem. UVGLU doesn't do exactly
 > what it advertises to do. When gluing two IFs together, it purports  
 > to check
 > if the first channel of the second IF differs by exactly one channel  
 > from that
 > of the last channel of the first IF. Well, this isn't what it does!
 > 
 > If the first channel of the second IF coincides with the last channel  
 > of the first IF,
 > it indeed refuses to glue the two together. If the difference is  
 > exactly 1 channel,
 > it works fine BUT IT STILL WORKS when the difference is 2 channels,
 > 3 channels, or n channels (I stopped running tests after a while).
 > 
 > So, it seems to be keen to prevent an overlap, but it doesn't seem to  
 > care
 > if there is a gap. COuld you have a quick look to see if this makes  
 > sense?
 > I can send you the files I have been playing with, of course, for you  
 > to run
 > SPLIT and UVGLUE on.

UVGLU is an antique that no one much uses.  The test of frequencies
computes the freq od data set 1 channel N+1 and data set 2 channel 1.
It then does
           (F2-F1)/DeltaF > 0.1
but of course this had better be absolute value of the ratio.  I have
fixed the 31DEC06 and 31DEC07 versions.

Eric Greisen