Re: weirdness and probable bug with set xdata/ydata time

[Ethan M. <eam...@gm...>]

On Saturday, 18 July 2020 15:51:04 PDT Juhász Péter wrote:
> Hi,
> 
> I've ran into an issue with time data that looks like a bug.
> 
> The script below works as expected:
> 
> ################
> $data << END
> 2020-07-01      1.1
> 2020-07-08      2.2
> 2020-07-15      3.3
> END
> 
> set xdata time
> set timefmt "%Y-%m-%d"
> #set ydata time
> plot $data u 1:2 w lp
> ###############
> 
> It produces a plot with correctly parsed and displayed dates on the x
> axis, and the x range is also sensible.
> 
> But uncomment the `set ydata time` line and it fails:
> 
> ################
>          "x.plt" line 10: warning: Skipping data file with no valid
> points
> 
> gnuplot> plot $data u 1:2 w lp
>                               ^
>          "x.plt" line 10: x range is invalid
> ##############
> 
> I can sort of understand that it fails because it can't parse the
> second column as date - but then why does it report the *x* range as
> invalid?

Since all of the points are invalid (because of y), it has no data left to
use for autoscaling x.  Therefore the range remains [VERYLARGE:-VERYLARGE],
which is an invalid range.

Perhaps easier to understand what's going on if you leave time data and
time formats out of it:

############
$NAN << EOD
NaN NaN
NaN NaN
EOD

plot $NAN using 1:2
         x range is invalid
show xrange
         set xrange [ * : * ] noreverse writeback  # (currently [8.98847e+307:-8.98847e+307] )
set xdata time
show xrange
         set xrange [ * : * ] noreverse writeback  # (currently ["         warning: time value out of range
32650/35/00,-468425088:00":"         warning: time value out of range
32650/35/00,-468425088:00"] )
############

> 
> Even more spooky is what it thinks about the x range after the failed
> plot:
> ##############
> gnuplot> sh xr
> 
> 	set xdata time
> 	set xrange [ * : * ] noreverse writeback  # (currently
> ["         warning: time value out of range
> 0000-35-32586":"         warning: time value out of range
> 0000-35-32586"] )
> #############
> 
> If I run gnuplot with valgrind, it barfs some "Conditional jump or move
> depends on uninitialised value(s)" errors, so some internal state is
> definitely corrupted.

Dunno. Valgrind doesn't complain here.

	cheers,

		Ethan


> 
> Best regards,
> Peter Juhasz
> 
> 
> 
> 
> 
> _______________________________________________
> gnuplot-beta mailing list
> gnu...@li...
> Membership management via: https://lists.sourceforge.net/lists/listinfo/gnuplot-beta
>