[Kst] Re: branches/work/kst/portto4/kst/src

[Barth Netterfield]

On Sat, Mar 19, 2011 at 2:04 PM, Matthew D Truch <matt at truch.net> wrote:
>> > When only a filename is passed to kst without any further arguments
>> > and it is not a .kst file try to find a datasource which could load
>> > it, and show all vectors against INDEX. Each datasource could overwrite
>> > this behavior, like it was asked by Syam Krishnan C.R. on the mailinglist.
>> >
>> > Having this feature it is now possible to register Kst as default
>> > program for opening data files: double-click on your data file and see Kst
>> > previewing your data!
>> While the idea has some good sides to it, I think it is dangerous.
>> Imagine what happens if you pass kst a datafile with thousands of
>> variables?
>> I think we need to fine-tune the behavior a bit. One possibility would
>> be to load only the first x vectors (x being in the area of 30 to 40).
>> But other ideas are welcome.
>
> Well, kst already checks if there is enough memory available to open
> things.  Perhaps that check should be part of the process?  I certainly
> like the idea of kst auto starting up when someone double clicks on data
> and the like.  Perhaps to appease both sides, an setting which permits
> the choice between auto-loading and displaying all data (subject to
> memory limits) or just auto-starting the data wizard with that datafile
> pre-selected (thereby skipping the first screen of the wizard).
>
>> In fact, I'm still not completely sure what Syam's use case is. It
>> would be nice to clarify it a bit before spending time developing more
>> stuff. Using the -F switch to load the data vectors from a different
>> file, but using a standard layout is a pretty nice possibility if you
>> always have the same data. If it's not clear to you what this feature
>> is, I can give more explanations.
>
> Well, certainly when I have relatively "simple" data files I often just
> want to plot everything in the file.  And then I go from there (like
> zoom in) if something looks interesting to me.

This is really great.  Maybe for the first instance we could also limit to
1 million points or something like that, so it won't stall for ever if someone
tries to open a huge file.

-- 
C. Barth Netterfield
University of Toronto
416-845-0946