[Kst] Vector skip patch ?

[Brisset, Nicolas]

> > 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.
> 
> Ah, that is why the f0 requested is not always what the user 
> gets.  I'll modify the testcases.
Which "consistency" gets to win here ? Code consistency or honouring
user expectation ? I'd advocate the second, that is:
- we do what everybody would expect in count from start mode (f0 does
not get changed) as there is no reason to do anything different here
- we align on skip boundaries in cound from end mode because of the
performance benefit, and document it clearly (tooltip and user manual)
so that users are warned, even though this is probably used with
streaming data and I doubt so many users would even notice that there is
an offset between the last point displayed and the actual last point
available in the vector

Does that sound right ? I'm not sure whether this is exactly the same as
you are saying, so I thought I'd reformulate my idea once more. I hope
it does not add to the confusion...

Nicolas