Question about "stats" patch

[Ethan M. <merritt@u.washington.edu>]

I have a question about the intended behaviour of the stats command.
What I observe is that the current xrange affects the result of a
command like
	stats 'file' using 3

I find this unexpected.  Is it intended, or is it a bug?

Here's what I was trying to do...

# preliminary exploration and plot of data column 2
 stats 'silver.dat' using 2 noout var "Ag2_"
 print Ag2_records, Ag2_min, Ag2_max
 set xrange [0:1]
 set yrange [Ag2_min:Ag2_max]
 plot 'silver.dat' using (1):2 with points pt 6

# So far so good, Ag_records = 58
# and the plot shows the scatter of column 2 values

# preliminary exploration and plot of data column 3
 stats 'silver.dat' using 3 nout var "Ag3_"
 plot 'silver.dat' using (1):3 with points pt 6
 ...

Oops.  The stats command + plot fails, claiming there is only 1 data point.
(Actually in one run it claimed 0 data points, but leave that for another day).
However...
 set xrange [*:*]
 stats 'silver.dat' using 3 nout var "Ag3_"
 ...

This works, indicating to me that the xrange is the cause of the problem.


Here's my thinking:
When you issue a plot command with a single column, e.g. "plot 'foo' using 3",
the values in column 3 are treated a y values, not x values.  I wouldn't expect
a filter on x to do anything here.  The stats documentation patch says:
 
  The `stats` command will only consider data points which fall into the
  plot range as defined inline or using `set xrange` and `set yrange`. If a
  value in either column falls outside of its corresponding range, the entire
  record is skipped and does not contribute to any of the summary statistics.

I read this to mean that if you say, e.g.
  stats 'foo' using N:M
then column N is checked against the x range and column M is checked against
the y range.  But I expected
  stats 'foo' using Q
to check Q only against the y range, with no check on x.

Bug?  Intentional?  
Am I looking at this the wrong way?

	cheers,

		Ethan