|
From: Tait <gnu...@t4...> - 2019-09-18 20:00:37
|
I think you misunderstand what "set datafile missing" does -- at least in the current gnuplot version. To quote from "help set datafile missing": Gnuplot makes a distinction between missing data and invalid data (e.g. "NaN", 1/0.). For example invalid data causes a gap in a line drawn through sequential data points; missing data does not. The help also notes missing/invalid handling changed between v4 and v5, so StackOverflow's advice may have been correct at one time. (I don't know why nobody is replying... I certainly don't have any special knowledge in this area.) Dave Horsfall <da...@ho...> said (on 2019/09/16): > Was my question too hard to answer, or merely mis-phrased? I simply want > to know whether I have misunderstood the meaning of "set datafile missing" > or should I be reporting it as a Gnuplot bug? > > I don't like reporting bugs if it turns out to be my fault... > > -- Dave > > ---------- Forwarded message ---------- > Odd; I just noticed that another of my graphs has a discontinuity at missing > data, yet I don't have the "set datafile missing 'x'" line, so what on earth > could be wrong? What have I not understood about missing data? > > Taking that directive out of spamlog.gp indeed shows a discontinuity (see > www.horsfall.org/spamlog-1.pdf) so something is definitely fishy here... > > -- Dave > > ---------- Forwarded message ---------- > Mac Sierra + latest patches (it won't go any further), Gnuplot 5.2 p6. > > Have a look at www.horsfall.org/spamlog.pdf and note the weirdness at the end; > there is missing data there thanks to my ISP cutting my fibre and taking ages > to fix it (great anti-spam measure, I suppose). > > In my spamlog.gp, I have: > > set datafile missing "x" > ... > plot datafile using 1:($2) lt rgb "red" title 'Reject', \ > ... > > Sample line from spamlog.dat: > > 2019/04/26 x x x x x x x x x > (Actually tabs, not spaces) > > Have I misunderstood what "missing" means? It's been a while since I studied > stats... My understanding is just that: missing data, hence leave a hole in > your graph (indistinguishable from zero in my case, but I can handle that). > > A message on StackOverflow also says that the above should work and not draw a > line across it (not that I trust StackOverflow anyway), but that's exactly what > Gnuplot is doing. > > Thanks. > > -- Dave |