[ Closed ] Can the tool import multiple point-series curves? (Page 1 of 2)

There is a little in the help file, but you are probably right that I need to improve the documentation of import of point series.

Yes, but only if they use the same x-coordinates.

When the data points do not share the same x-coordinates, you have to put each point series in a separate file. The format could look like this:

1;2
3;4
5;6
7;8

Graph supports separation by semicolon, comma, tabulator or space.

The beta version of Graph 4.4 has support for scripting, but it has even less documentation than the import feature.

You got yourself a deal.
But I think it is best to keep responding in the forum as it might have interest to others.

You need to install Python 3.1 to use the plugin scripting feature. You can access it by pressing F11 inside Graph, which will show a Python interpreter console, or you can place a .py file in the Plugins directory, which will be automatically run when Graph is started.

There are already a few example plugins installed with Graph you can look at.

It is probably mostly because no one has asked about it before. I think csv files are mostly used to import data points from Excel or a similar program, where you have the x-coordinates in the first column and the y-coordinates for every point series in the other columns.

I will consider using a wiki, but I don't think it will work very well with Graph. The users of Graph are very different from the users of CodeLite.

Python is a scripting language. When you know other programming languages it should be easy to learn. There is a great tutorial in the documentation.

If it is just a matter of adding some kind of separator, I guess it would be easy. I will take a look at it. Graph itself is written in C++, so there isn't any scripting code in it yet though.

Maybe I am.
At least I don't think most users of Graph are programmers and are not used to writing documentation.

OK, I've installed "the full Monty" ... Python that is :-) and skimmed the docs, and I still don't see how it connects to Graph. If a Python script is able to do Graph commands, then I don't see why I can't have a Python script do multiple point-curves and points with different X values, and in fact, just about anything that Graph can do without help from Graph.

Perhaps it'd help if you knew my application...

I've got some performance curves in graphical form created from experimentation. I want to write software that analyzes/uses these graphs, so I've created a set of points that I manually read from the graph, which I'm hoping is a piece-wise linear approximation to the curves.

And I'd like to make sure that the point data describing the curves that I've put into C initialize statements looks about right, so I'd like to process the C initializer data into a set of points that I can display on your Graph program and look for inaccuracies. For instance, maybe I read a graph point wrong and a point is misplaced. I can't tell that from just looking at numbers. I need to see the graph of it.

When I find an error or an unrealistic bend, I'd fix it in the C initializers, then run it thru again and replot it in Graph to make sure the result is smooth and accurate. So as you can see, I'm expecting multiple passes.

And I'll probably have to write something to munge the C initializer format into whatever Graph needs. One graph I've got has about 18 different curves, so doing 18 individual files would be inconvenient to do once, nevermind doing the set multiple times.

Does that make sense?

Seems like anybody doing software analysis of experimental data might be interested in this process. I can post my software format munger later if I do one in case anybody is interested.

OK, now we're getting somewhere. I understand better about Python, and I think you're right, it really doesn't add any value to what I'm doing. But the "#" thing does.

I think you've implemented the bare minimum, but while I've got your ear, let me make a suggestion that would make life (mine anyway :-) much easier.

Suppose you were to ignore any characters that you didn't understand in an import file and look only for numbers - specifically pairs of numbers on a line, and the # symbol. Then I could feed my .H file in directly if I were careful about my comments. The file looks like this:
/*****************************************************************
*Here's a bunch of comments, some with a single
*number in the line like copyright 2010.
*/

#include "ga.h"

point_t
tbl_lr35_to8_wt[MX_CURVES][MX_POINTS] =
{
  {//Curve 0
    {NULLNR, NULLNR},
    {10000, 1980},
    {12400, 2150},
  },
//# (for Graph to know it's another curve)
  {//Curve 1
    {NULLNR, NULLNR},
    {10000,  2130},
    {12400,  2310},
    {12600,  2330},
  }
}

So if you take the browser philosophy of being as error tolerant as possible, I could import this file directly into Graph and it'd just ignore the lines it can't understand, and plot only the point lines with two numbers. Or even better, it'd be neat if I could embed commands in the C-comments for Graph like:
//Color: Blue
//Line: Interpolation Linear
etc.

What do you think?

It is actually not as easy as it might sound. Graph handles expressions and not only numbers, so you can actually have a coordinate like (pi+2,e^-1)

Now you are starting to invent a new scripting language, and I already have Python for that.

Great!

Under the type you can select the size. You can even set it to 0 if you don't want the markers shown.

It looks like when an import has a blank line (perhaps in conjunction with the new "#" command you put in), the program crashes.

Also, if my first value is "0,1540" nothing displays and it got imported as X=0,1540 and Y is blank.
I had to put a space in: "0, 1540"

I'm unable to reproduce the crash now (I tried a bunch of stuff on what was pretty close to the file as it was when the import crashed I think. Oh well...)

But if you import the following file:
0,1540
10800,123
15000,150

It thinks both numbers on each line are X, and there's nothing for Y.
But it works if you insert a space after each comma like:
0, 1540
10800, 123
15000, 150

BTW, it'd be neat if you had a 'clear' or 'close' command so I wouldn't have to exit and start the program again for each attempt to reproduce the crash.

Oh, yes, looks like New will do what I want (create a new system), thanks!

I'm using V4.4 Beta, build 483.

OK, here's the file. When I import, then edit the curve, I see:
     X             |      Y
0,1540          |
10800,123    |
15000,150    |

and nothing plots.

Interesting - I got it to work by putting a space after each comma even with the extra space at the end of the first line. And it's strange that a space after the first line messes up the interpretation of the subsequent lines as well. I had to put a space after all the commas.

OK, I reproduced the crash on the import of the multiple curves, probably the one I saw before but couldn't reproduce. It turns out if I have a line composed *only* of spaces, I get an exception at EEFFACE

Note that line 3 has a couple of spaces and nothing else.
A blank line doesn't crash.

It must be allergic to spaces. :-)

Just tried it - looks like you successfully inoculated it against spaces. :-)