[Kst] trunk open

[Brisset, Nicolas]

> 1.3 includes only what is listed in the Bug list as HI 
> priority or higher.
I have browsed through the list messages of the past week, and seen that
you've been "cleaning up" a lot of things. Looking at the next
features/improvements planned and comparing with my expectations, I have
a few remarks:
1) the very professional improvement discussed in bug #110734 is only
marked as NORmal priority. It does not appear anywhere in the
RELEASE_PLAN either, so that I'm wondering when/whether it is planned to
tackle it ? Hopefully it will be done at some point.
2) thinking about datasources and working a lot with multi-sampled files
(i.e. files where data come in blocks with different sampling times and
periods), I've come to the conclusion that we could improve things a lot
by just adding the concept of a "buddyVar" (the name could be changed!).
This would work roughly as follows: a virtual method like bool
buddyVar(Qstring& field, Qstring& buddy) gets added to kstdatasource. It
returns the preferred associated variable of a given field (like the
time corresponding time field), if there is any. The datasource should
know the structure of the datafile and be able to handle that easily,
while it could and should remain transparent to the user. The datawizard
gets a new "auto" value for the X field to use when creating new curves.
In that case (i.e. no explicit X field chosen), the wizard asks for each
Y field which is the preferred var to use as X via the previous method.
If there isn't any, it reverts to some sensible default (like "time"),
or INDEX, or asks the user to specify it.

I can elaborate a bit more on the second point and create a bugzilla
entry if there is interest...

Nicolas