"Fraga, Eric" <e.fr...@ucl.ac.uk> writes:

> > I now have one remaining problem: my time values in the second column
> > contain the name of the time zone - the purpose is to flag the time as
> > summer or winter time.  Currently I need to remove these indicators from
> > the table, else plotting fail, gnuplot doesn't understand the data.
>
> You can specify the format of dates/times in gnuplot and then manipulate
> them.  See "time/date specifiers" in the gnuplot info manual and also
> read about the strptime() function.  I don't have any examples of these
> at hand unfortunately.

Yes thanks - but I really think Org gets it wrong.  Look:
With

| plot data using 1:(timecolumn(2,"%H:%M") ...

gnuplot happily accepts a data file looking like

| 2024-09-18      11:40 CEST
| 2024-09-19      11:27 CEST
|    ...             ...

It just discards the timezone which is good enough.  No problem with the
kind of data for gnuplot (AFAIR I read in the gnuplot manual that time
zones are not supported ad-hoc - anyway, I don't need this for
plotting.)

But the real problem is that the org table export mechanism thinks that
the time field doesn't look like a numerical field, so it wraps every
such field in quotes, and the exported data file will look like

| 2024-09-18      "11:40 CEST"
| 2024-09-19      "11:27 CEST"
|    ...             ...

instead.  Then gnuplot doesn't find data matching "%H:%M in the second
column of the data file and interprets it as zero or 00:00am or
whatever.

Can I make gnuplot disregard the quotation marks?  Else I must tell org
that these fields should not be quoted.  But ironically the only way I
know to do this is again a hack: I must adjust
`org-table-number-regexp' which... sounds like a good solution but since
org disregards a file local binding I must change the global binding of
the variable!  Ugh.  Better solutions welcome.


Thanks,

Michael.

Reply via email to