Bryan Harris wrote:
> Veusz needs to be able to handle datasets with headers. All my data is in
> ASCII format, but it has headers in the first two lines identifying the
> column
> type and unit. I can work around this by deleting the headers, but I
> shouldn't have to.
>
> Actually, I just told Veusz to skip the first two lines and that works
> fine:
>
> stream.newLine()
> stream.flushLine()
> stream.newLine()
> stream.flushLine()
>
> I think if people have to modify their data sets to make them work they
> just
> won't use the program. At the very least there ought to be a box to
> select
> how many lines to skip. (That's what Kaleidagraph does.) As it stands,
> data with text headers gets imported, but can't be plotted because it's
> not
> recognized as numbers. (Kaleidagraph has the same flaw if you don't skip
> the
> headers. It's a pain.) I think it would be slick to automatically
> recognize
> when the first few lines are text and auto-populate the descriptors box.
> We
> could also look for and just ignore text. (This is more or less what
> excel
> does. I think this program ought to be at least as clever as excel...)
I certainly think there is a problem with the data import. I'd like to make
it more intelligent and recognise headers (if required), or allow a click to
name a column. The problem is to make the behaviour predictable and sensible
Do you have some examples of where it could be useful? The CSV import
currently uses the column headings as descriptors.
A simple fix for your problem might be check-box option could be to ignore
text in the input file, though this is not ideal.
One simple idea (though not ideal for what you want here) would be to have
user-definable input plugins to support general file types, or perhaps a
text filter.
> I think it makes sense to prepend the filename to the imported datasets.
> It will make the dataset names long, but will make it much easier to
> distinguish
> between data from multiple files. We will need to scrub the name for
> illegal
> characters. (It doesn't like - or + because those are used for error
> limits. Comma (',') would probably cause problems too.)
We could do this, though I'm reluctant to do it automatically as the dataset
names would get very long (my filenames are often very long). How about a
prefix /suffix option which would add a specified prefix/suffix to each
dataset name (with filename as an option)? I don't want the dialog to get
too complex, so we need a simple way of allowing flexibility in import.
Perhaps an advanced settings menu/dialog would help for the dialog to not
confuse new users.
> Also, I think the formulas used to relate datasets ought to be editable.
> You
> can always delete datasets and re-enter them, but that gets messy. Is
> there already some way to do this that isn't obvious?
Yes - they really need to be editable. The only way currently is to edit the
saved document file.
> Also, in the plot window. A cool feature of some plotting software
> programs is that you can put markers on your trace to identify the line
> rather than
> corresponding to specific data points. My sets (stress strain curves)
> have
>>1000 data points typically and displaying all data points just looks bad.
>>It
> would be nice to be able to display just a small percentage of the data
> points as markers and connect them with a line.
This is a good idea as I plot very large datasets occasionally and it gets
very slow with antialising switched on. Is this just for display on the
screen or do you want all output thinned?
There could be the equivalent of a wireframe preview which optionally thins
large datasets on the screen, but perhaps you want fewer points in the
dataset. A workaround is to create a new dataset with an expression such as
olddataset[::10]
which only includes every 10th point.
> What do you think? To be clear, I'm offering to do all this stuff not
> asking
> for somebody else to do it. I just wanted to bounce ideas off somebody.
> This looks like an excellent program and I'm excited to contribute if I
> can! Bryan
You can certainly can have a go at implementing your ideas, but please
discuss the specifics of what you'd like to do so we can get the right and
most general solution!
Thanks again
Jeremy
_______________________________________________
Veusz-discuss mailing list
[email protected]
https://mail.gna.org/listinfo/veusz-discuss
