Am 06.12.2015 um 03:25 schrieb Allin Cottrell:
On Fri, 4 Dec 2015, Sven Schreiber wrote:
> I have a panel dataset which is monthly in the time dimension,
...
Reason: hours_in_day <= 24 (allowing for sleeping). It's a fairly
substantial coding effort to generalize all "privileges" associated with
time-series data to the time dimension of panels. (The --panel-time
option was originally intended just to produce a correct-looking time
dimension in panel plots.)
I understand; my point is that these limitations are not obvious to the
user. I believe that from reading the manual and other documentation the
impression is that the time dimension of panel data would be equivalent
to the time series case.
The "strange message" is trying to say that isn't implemented for panel
data ("This command won't work with the current periodicity").
I wasn't explicit about "strange", but I certainly didn't just get that
message which would be self-explanatory. Instead I got a sequence of
about ten very weird and seemingly unrelated messages. I'll see if I can
replicate them.
thanks,
sven