[Kst] A few (not so trivial) questions...

Brisset, Nicolas Nicolas.Brisset at eurocopter.com
Tue Jan 17 16:32:32 CET 2006

I am gradually introducing new users to kst, and in the process I get a
lot of feedback/requests. I know it is not the right time to discuss new
features :-), but while I'm thinking and discussing about them I thought
I might just as well document the ideas for future reference.

First thing: regarding the feedback, it is generally quite positive.
People need to be shown the basics to get started (otherwise they do
tend to be a bit overwhelmed), but once they've grown familiar with the
main concepts and the terminology, they adapt quickly and seem to enjoy
using kst. I will definitely try to continue producing Flash tutorials,
as I think it can really help there.

As for the requests, some are just existing features that users just
don't find at first (especially because nobody reads the manual, which
is probably not suprising when they use the svn version where it seems
it is not installed :-)). There are also quite a few interesting
requests for which I will submit reports in bugzilla, and some which
raise questions :
1) context information: many people have data files with some context
information in them (experiment date, system configuration, any other
relevant information) and it would probably make sense to be able to use
it somewhere in kst. I know the datasource API provides a place to put
metadata, but I'm not aware of what one can actually do with it ? It
would be nice to e.g. add a window or text frame with metadata
(especially in the printed output). One possible use would be to have
window top labels (like plot top labels, but above all plots) which
would display it. But that clearly needs to be refined as some
datasource might produce fairly large amounts of metadata... And maybe
we need to distinguish between per-field metadata (unit, comment, etc)
and per-file metadata (experiment date, weather conditions, etc...)
Basically, our need in the short term would be to be able to add 1 or 2
lines of context information read from the data file (up to 80 chars
each) somewhere at the top of the page. I don't know how to do that now
more or less automatically, but if some of you have ideas I'm interested
2) vector names: it happens quite often that we want to compare two data
files (corresponding to 2 different configurations). In that respect,
the "Change data file" tool with the option to duplicate and not only
replace is *absolutely gorgeous*. There is, however, one thing that is
not so nice: vectors are created with very similar names (e.g. VEC1 and
VEC1'), giving no hint as to which comes from what data file. As long as
you're inside your kst session, you can always look at vector properties
in the data manager, but if you have printed pages, there is no way you
can make that out. I understand this is a pretty difficult issue, but
here are some hints as to how this could be solved (comments welcome !):

	- change the way vector names are generated to include the data
file (ugly ?)
	- add an option to legend generation that optionally adds the
file name in the legend
	- implement per-vector metadata to store units and filename and
possibly other information, and allow to reuse it in the legend (a
generalization of the previous option)
	- ...

Now, that was for the more complicated requests. I also have much
simpler ones for which I'll file bugzilla reports soon, and it will be a
pleasure to implement them are they are much more straightforward :-)


More information about the Kst mailing list