[curveplot] Improve *.dat export

[vallsv]

Feedback from a user. That's not a high priority at all.

Tested with silx 2.1.1.

The dat export with multi curves is like exporting multiple scans instead of multiple curves.

• Run examples/plotCurveLegendWidget.py
• Export as dat
• Import the dat file with pymca

• Notice that there is 3 scans, and 2 counters per scans.

They would expect a single scan, with the 4 counters (3 counters + 1 axes).

BTW during the export we can notice that:

(bliss2.2) valls@lvalls:~/workspace/silx.git(fix-get-display-mode)$ python examples/plotCurveLegendWidget.py 
Warning: Ignoring XDG_SESSION_TYPE=wayland on Gnome. Use QT_QPA_PLATFORM=wayland to run on Wayland anyway.
Qt: Session management error: None of the authentication protocols specified are supported
X ['Y'] %.7g 2 [-3.14159265 -3.14120843 -3.14082421 ...  3.14082421  3.14120843
  3.14159265] [[-0.56797773  0.67579871 -0.03870308 ... -0.52785258 -0.67234882
  -0.98853755]]
X ['Y'] %.7g 2 [-3.14159265 -3.14120843 -3.14082421 ...  3.14082421  3.14120843
  3.14159265] [[-1.22464680e-16 -3.84222170e-04 -7.68444284e-04 ...  7.68444284e-04
   3.84222170e-04  1.22464680e-16]]
X ['Y'] %.7g 2 [-3.14159265 -3.14120843 -3.14082421 ...  3.14082421  3.14120843
  3.14159265] [[-1.         -0.99999993 -0.9999997  ... -0.9999997  -0.99999993
  -1.        ]]

There is an extra print somewhere.

[t20100]

This part of the code hasn't changed much in the last 6/7 years, so it's a bit risky to change the behavior now...
The most recent relevant change looks to be "Save error-bars when saving a plot" #3199 (from 2020...)

I guess it was saved with different scans rather than counters because it allows for each curves to have a different number of points.

What could be done is:

• check that all curves have the same number of points (+ probably no error bars) and then save as a single scan... but that's kind of obscure behavior IMO
• provide yet another save options that save all curves as one scan eventually clipping the curves that are longer or filling the ones that are too shorts

[t20100]

BTW, there is indeed a print:

+++ b/src/silx/io/utils.py
@@ -423,7 +423,6 @@ def savespec(
     ncol = data.shape[0]
     assert len(labels) == ncol
 
-    print(xlabel, ylabel, fmt, ncol, x_array, y_array)
     if isinstance(fmt, str) and fmt.count("%") == 1:
         full_fmt_string = "  ".join([fmt] * ncol)
     elif isinstance(fmt, (list, tuple)) and len(fmt) == ncol:

[vallsv]

In this case maybe you can close this issue, until somebody else request that. As i said, the user will use the HDF5 reader from PyMCA anyway, and will not export anymore to .dat.

Just remove the print.

[t20100]

PR #4192 removes the print.
I'm closing this issue then