[gnuplot-beta]set missing malfunctions

[Hans-Bernhard B. <br...@ph...>]

Hello, guys,

someone just filed a bug that 'set datafile missing' is misbehaving, both
compared to both 'set missing' in 3.7 and to the documentation, if used
with simple using specs (the 'using 1:2' example in 'help missing').

Digging into this a little reveals that this must have been broken for
quite some time (at least 2 years).  The difference becomes obvious if
you run that example with the table terminal:

# using 3.7:
> env GNUTERM=table /usr/bin/gnuplot tmiss.gpl
#Curve 0, 4 points
#x y type
1 10 i
2 20 i
4 40 i
5 50 i

#compared to 4.0:
> env GNUTERM=table gnuplot tmiss.gpl
#Curve 0, 5 points
#x y type
 1  10  i
 2  20  i
 0  0  u
 4  40  i
 5  50  i


As you can see, the difference is that in 3.7 the '?' point is actually
treated as missing, i.e. as if it simply wasn't there, whereas 4.0 treats
it as an undefined datapoint. The culprit modification appears to be the
datafile.c:1161 ff., stamped by Ethan in October 2002.  It seems the 
rationale given in Ethan's comment is at odds with the documentation and 
with traditional behaviour.  So: which do we change: the code or the docs?

-- 
Hans-Bernhard Broeker (br...@ph...)
Even if all the snow were burnt, ashes would remain.