Am 03.05.2025 um 22:05 schrieb Cottrell, Allin:
On Sat, May 3, 2025 at 10:00 AM Sven Schreiber
<sven.schreiber(a)fu-berlin.de> wrote:
> follow-up, the behavior is not restricted no matrix input; example:
>
> <hansl>
> open AWM
> smpl 1980:1 1980:1
> gnuplot GCR --time-series # same error (missing values)
> </hansl>
>
> Does anybody have an idea for a workaround? It would be nice to have
> this corner case covered for general algorithms where this may occur.
I don't have any idea for a workaround, other than detecting the case
where the sample contains a single observation and therefore
abandoning the notion of producing a plot.
Right, that's what I did in the
meantime.
However, in current git there's now code to ensure that we
don't pass
an empty xrange (such as [1980:1980]) to gnuplot, even when there's
just a single observation available. I suppose this could be OK for
personal usage but I wouldn't really want to see a function package
that was willing to "plot" a single observation.
Thanks, I think it's good that the error is avoided, even if the value
of a single-obs plot is certainly limited.
cheers
sven