Bug or inaccurate documentation about .csv files with data visualization

[quark67]

Version: 3.1.9a

In pgf/doc/generic/pgf/pgfmanual-en-dv-main.tex (https://github.com/pgf-tikz/pgf/blob/28b9756fef06fc59ebee7241261ebd0e51d83c21/doc/generic/pgf/pgfmanual-en-dv-main.tex), lines 370-372, we read:

\datavisualization ...
  data [read from file=file1.csv]
  data [read from file=file2.csv];

This is false (or bugged). This don't work if the .csv file has an header (see https://tex.stackexchange.com/questions/627516/how-to-plot-function-from-a-csv-file-in-tikz-the-csv-file-created-from-the-ex).

If the bug is not corrected, the documentation must be updated with this precision:

The .csv file must not contain an header (e.g. : the first line must not be x,y).

And the lines 371-372 must be replaced by:

  data [headline={x, y},read from file=file1.csv]
  data [headline={x, y},read from file=file2.csv];

Edit: alternatives options:
Removing commas in the csv file, replacing it with spaces, while keeping the header, and replace lines 371-372 with:

  data [separator=\space,read from file=file1.csv]
  data [separator=\space,read from file=file2.csv];

[hmenke]

This is a bug in datavisualization. There should be no difference between handling of inline data and data read from a file.

[muzimuzhi]

When reading from a file line by line, each line is appended a space derived from the line end. So the read in headline would be x,<space>y<space> and the second recognized attribute would be y<space>, which is different from the expected y. This difference causes the reported problem.

Using option headline={x, y} avoids the trailing space hence avoids the problem. To show it's not the use of option headline, but the value parsed to it that works around the problem, one can play with data[headline={x, y } , ...], note the space after y and the one between } and ,.

I'll provide a PR.