From: Dave H. <da...@ho...> - 2021-02-02 22:34:31
Attachments:
health.gp
|
Gnuplot 5.4 patchlevel 1 on MacBook Pro Sierra 10.12.6 (updated each week). On 1st February (I don't know whether the date is significant) my daily health plot suddenly lost its R/H axis. I didn't do nuffink... Well, there was a MacPorts update that morning, but I didn't see Gnuplot in the list; I guess it's possible that something got borked, but just the R/H axis? Gnuplot was last updated on 25th Jan, and worked up until now. As it was on 31st Jan: http://www.horsfall.org/health.pdf.old (no HTTPS yet). Now: http://www.horsfall.org/health.pdf Well, all I did was to add an entry for 1st Feb: # Date Time Wgt Gth Sys Dia HR 30/1/21 1800 89.9 109 115 72 77 31/1/21 1730 88.6 109 118 72 95 ==> 1/2/21 1830 88.6 109 115 67 82 2/2/21 1730 90.3 109 118 60 69 The error: gnuplot> load 'health.gp' "health.gp" line 99: warning: y2 axis range undefined or overflow Note that line 99 is actually the last continued line of the "plot" command, which I certainly had not changed: mackie:health dave$ ls -l *.gp -rw-r--r-- 1 dave staff 5511 Sep 27 2019 health.gp (Note that the one I attached has potentially sensitive comments removed; it's my personal information, after all, but I'm willing to share the lot with an individual, not the entire list.) It's worth noting that the plot has probably saved my life a few times, as I suffer from hypertension (with medication)... The script itself is attached (I trust that it's within the guidelines). Thanks. -- Dave |
From: Ethan M. <eam...@gm...> - 2021-02-02 22:41:40
|
On Tuesday, 2 February 2021 14:17:59 PST Dave Horsfall wrote: > Gnuplot 5.4 patchlevel 1 on MacBook Pro Sierra 10.12.6 (updated each > week). > > On 1st February (I don't know whether the date is significant) my daily > health plot suddenly lost its R/H axis. I didn't do nuffink... > > Well, there was a MacPorts update that morning, but I didn't see Gnuplot > in the list; I guess it's possible that something got borked, but just the > R/H axis? Gnuplot was last updated on 25th Jan, and worked up until now. > > As it was on 31st Jan: http://www.horsfall.org/health.pdf.old (no HTTPS > yet). > > Now: http://www.horsfall.org/health.pdf > > Well, all I did was to add an entry for 1st Feb: > > # Date Time Wgt Gth Sys Dia HR > > 30/1/21 1800 89.9 109 115 72 77 > 31/1/21 1730 88.6 109 118 72 95 > ==> 1/2/21 1830 88.6 109 115 67 82 > 2/2/21 1730 90.3 109 118 60 69 > > The error: > > gnuplot> load 'health.gp' > "health.gp" line 99: warning: y2 axis range undefined or overflow Since your script never sets a range for y2 at all that error message is evidently correct. I do not understand how the script could ever have worked correctly without either "set link y2" or "set y2range [min:max]". Could the script have gotten borked? Ethan > > Note that line 99 is actually the last continued line of the "plot" > command, which I certainly had not changed: > > mackie:health dave$ ls -l *.gp > -rw-r--r-- 1 dave staff 5511 Sep 27 2019 health.gp > > (Note that the one I attached has potentially sensitive comments removed; > it's my personal information, after all, but I'm willing to share the lot > with an individual, not the entire list.) > > It's worth noting that the plot has probably saved my life a few times, as > I suffer from hypertension (with medication)... > > The script itself is attached (I trust that it's within the guidelines). > > Thanks. > > -- Dave |
From: Dave H. <da...@ho...> - 2021-02-02 22:52:52
|
On Tue, 2 Feb 2021, Ethan Merritt wrote: >> gnuplot> load 'health.gp' >> "health.gp" line 99: warning: y2 axis range undefined or overflow > > Since your script never sets a range for y2 at all that error message is > evidently correct. > > I do not understand how the script could ever have worked correctly > without either "set link y2" or "set y2range [min:max]". That's the bit that had me confused; it had been working for ages... > Could the script have gotten borked? Not according to the modification time unless it was some weird corruption. Anyway I'll explicitly set the y2range as you suggested. Thanks. -- Dave |
From: Dave H. <da...@ho...> - 2021-02-03 02:05:23
|
On Tue, 2 Feb 2021, Ethan Merritt wrote: > I do not understand how the script could ever have worked correctly > without either "set link y2" or "set y2range [min:max]". Well, stuffed if I know how it managed to work since 2013 when I first wrote it (with minor tweaks); I think we can safely file this under "H" for "Heisenbug" :-) Explicitly setting "y2range [0:160]" (the same as "yrange" because it happened to be the same scale) fixed it; there's probably a better way but I like to spell things out (I might want to change the scale(s) one day), and it may have something to do with different captions. Many thanks! I dunno how I missed that, or how it even worked at all... -- Dave |
From: Hans-Bernhard B. <HBB...@t-...> - 2021-02-02 23:08:41
|
Am 02.02.2021 um 23:17 schrieb Dave Horsfall: > The error: > > gnuplot> load 'health.gp' > "health.gp" line 99: warning: y2 axis range undefined or overflow That warning message appears justified, because there's a bug in the plot script (at least the one you showed). It doesn't really make much sense to request ticks on the y2 axis, but nothing else: no data on the y2 axis itself, nor any stated relation between y1 and y2 axis, FWIW, I suspect adding set link y2 should reinstate the behaviour of older versions of gnuplot for you. OTOH, if you were actually going to use the second y axis, there should be 'axis x1y2' options on the applicable datasets (Sys, Dia). |