|
From: Daniel J S. <dan...@ie...> - 2016-06-08 16:07:59
|
I've been stepping through the demos, and my general observation from what I remember is that there is now much less white space around plots. The behavior is the same in two different terminals (x11 and qt). I've gone back to about -D2016-05-25 in the repository to see if this is a current inadvertent change. It's the same as current development version. Has anyone else noticed this? Perhaps it was intentional, and in some sense I don't mind it because it makes the plots use more of the available window screen. I mean, typically when I create a plot for a publication I will reduce the whitespace to minimal because word processors can always add it back in. That can be done changing the margins or post-gnuplot with some cropping tool specific to, say, eps or something. However, as far as the demos, the plots don't feel as though they have a good white space balance for several plots. Here are some notable ones: fillcrvs.dem (7) world.dat plotted with filledcurves multiplt.dem (1) [The above two illustrate how plot are is slightly skewed to the left] heatmaps.dem (1) (2) (3) (4) piecewise.dem (1) Piecewise function sampling matrix_index.dem (1) Data file contains labeled ascii matrices [annotation off screen] Dan |
|
From: Ethan A M. <sf...@us...> - 2016-06-08 18:40:13
|
On Wednesday, 08 June, 2016 11:07:28 Daniel J Sebald wrote: > I've been stepping through the demos, and my general observation from > what I remember is that there is now much less white space around plots. > The behavior is the same in two different terminals (x11 and qt). > I've gone back to about -D2016-05-25 in the repository to see if this is > a current inadvertent change. It's the same as current development version. The default margins for "set view map; splot ..." were changed to be more like those for an equivalent 2D plot. So that category of plot is expected to have less white space around it than in old gnuplot versions. > Has anyone else noticed this? Perhaps it was intentional, and in some > sense I don't mind it because it makes the plots use more of the > available window screen. I mean, typically when I create a plot for a > publication I will reduce the whitespace to minimal because word > processors can always add it back in. That can be done changing the > margins or post-gnuplot with some cropping tool specific to, say, eps or > something. > > However, as far as the demos, the plots don't feel as though they have a > good white space balance for several plots. Here are some notable ones: > fillcrvs.dem (7) world.dat plotted with filledcurves > multiplt.dem (1) I have no explanation for any changes seen in these > [The above two illustrate how plot are is slightly skewed to the left] > heatmaps.dem (1) (2) (3) (4) These are set view map + splot, so changes are expected > piecewise.dem (1) Piecewise function sampling > matrix_index.dem (1) Data file contains labeled ascii matrices > [annotation off screen] These demos did not exist prior to version 5, so I'm not sure what you are comparing to. The output looks normal when tested here. Ethan |
|
From: Daniel J S. <dan...@ie...> - 2016-06-08 20:05:54
|
On 06/08/2016 01:39 PM, Ethan A Merritt wrote: > On Wednesday, 08 June, 2016 11:07:28 Daniel J Sebald wrote: >> I've been stepping through the demos, and my general observation from >> what I remember is that there is now much less white space around plots. >> The behavior is the same in two different terminals (x11 and qt). >> I've gone back to about -D2016-05-25 in the repository to see if this is >> a current inadvertent change. It's the same as current development version. > > The default margins for "set view map; splot ..." were changed to be > more like those for an equivalent 2D plot. So that category of plot > is expected to have less white space around it than in old gnuplot > versions. OK, so it looks like primarily the issue is that a lot of demos need some fine tuning...or remove fine tuning that was geared toward older layout. >> Has anyone else noticed this? Perhaps it was intentional, and in some >> sense I don't mind it because it makes the plots use more of the >> available window screen. I mean, typically when I create a plot for a >> publication I will reduce the whitespace to minimal because word >> processors can always add it back in. That can be done changing the >> margins or post-gnuplot with some cropping tool specific to, say, eps or >> something. >> >> However, as far as the demos, the plots don't feel as though they have a >> good white space balance for several plots. Here are some notable ones: > > >> fillcrvs.dem (7) world.dat plotted with filledcurves >> multiplt.dem (1) > > I have no explanation for any changes seen in these Ah, there's a margin command in that example that was probably for tweaking the older layout behavior. This change looks much better: Index: gnuplot/demo/fillcrvs.dem =================================================================== RCS file: /cvsroot/gnuplot/gnuplot/demo/fillcrvs.dem,v retrieving revision 1.6 diff -u -r1.6 fillcrvs.dem --- gnuplot/demo/fillcrvs.dem 19 May 2007 20:35:53 -0000 1.6 +++ gnuplot/demo/fillcrvs.dem 8 Jun 2016 20:00:20 -0000 @@ -70,7 +70,6 @@ set object 1 rect from graph 0, 0 to graph 1, 1 behind fc rgb "#afffff" fillstyle solid 1.00 border -1 set xrange [ -180.000 : 180.000 ] set yrange [ -70.0000 : 80.0000 ] -set lmargin 1 plot 'world.dat' with filledcurve notitle fs solid 1.0 lc rgb 'dark-goldenrod' pause -1 'Press Return to continue' >> [The above two illustrate how plot are is slightly skewed to the left] >> heatmaps.dem (1) (2) (3) (4) > > These are set view map + splot, so changes are expected OK, map view, yeah there's quite a few of those. The historic map size was kind of small. This is the change that was noted in this bug report: https://sourceforge.net/p/gnuplot/bugs/1802/ I will follow up on that one on Sourceforge. >> piecewise.dem (1) Piecewise function sampling >> matrix_index.dem (1) Data file contains labeled ascii matrices >> [annotation off screen] > > These demos did not exist prior to version 5, so I'm not sure what > you are comparing to. The output looks normal when tested here. I'm seeing the y-axis text right up against the screen edge, and on the right side is three characters of white space. But, I see that margins are specified in this example. Note, however, that I'm seeing an error message for the "set multiplot" line: gnuplot> load 'matrix_index.dem' "matrix_index.dem", line 29: warning: must give margins and spacing, continue with spacing of 0.05 Hit return to continue If I just remove the "margins etc" option, the warning disappears and the spacing looks good, very balanced. Should we just remove the margins? Or keep them in as a way of testing the option? Here's what looks good for me: Index: gnuplot/demo/matrix_index.dem =================================================================== RCS file: /cvsroot/gnuplot/gnuplot/demo/matrix_index.dem,v retrieving revision 1.2 diff -u -r1.2 matrix_index.dem --- gnuplot/demo/matrix_index.dem 13 Jan 2015 18:17:29 -0000 1.2 +++ gnuplot/demo/matrix_index.dem 8 Jun 2016 20:00:20 -0000 @@ -26,7 +26,7 @@ set xrange [] noextend set yrange [] noextend -set multiplot layout 2,2 margins char 3,3,2,4 title "{/:Bold Data file contains labeled ascii matrices} +set multiplot layout 2,2 margins char 5,3,2,4 spacing char 6,3 title "{/:Bold Data file contains labeled ascii matrices} set title "Y range should be the same" plot '$MATRICES' nonuniform matrix i "set3" w image title "index 'set3'" Dan |
|
From: Ethan A M. <sf...@us...> - 2016-06-08 20:24:12
|
On Wednesday, 08 June, 2016 15:05:16 Daniel J Sebald wrote: > On 06/08/2016 01:39 PM, Ethan A Merritt wrote: > >> piecewise.dem (1) Piecewise function sampling > >> matrix_index.dem (1) Data file contains labeled ascii matrices > >> [annotation off screen] > > > > These demos did not exist prior to version 5, so I'm not sure what > > you are comparing to. The output looks normal when tested here. > > I'm seeing the y-axis text right up against the screen edge, and on the > right side is three characters of white space. But, I see that margins > are specified in this example. Note, however, that I'm seeing an error > message for the "set multiplot" line: > > gnuplot> load 'matrix_index.dem' > "matrix_index.dem", line 29: warning: must give margins and > spacing, continue with spacing of 0.05 > Hit return to continue That warning seems unnecessarily alarming. All it means is that in the absence of an explicit spacing keyword it will use the default value. > If I just remove the "margins etc" option, the warning disappears and > the spacing looks good, very balanced. Should we just remove the > margins? Or keep them in as a way of testing the option? Here's what > looks good for me: [shrug] Tastes differ. I think it looks better with identical left and right margins. To me the default looks lopsided. Ethan |
|
From: Daniel J S. <dan...@ie...> - 2016-06-08 21:26:04
|
On 06/08/2016 03:20 PM, Ethan A Merritt wrote: > On Wednesday, 08 June, 2016 15:05:16 Daniel J Sebald wrote: >> On 06/08/2016 01:39 PM, Ethan A Merritt wrote: > >>>> piecewise.dem (1) Piecewise function sampling >>>> matrix_index.dem (1) Data file contains labeled ascii matrices >>>> [annotation off screen] >>> >>> These demos did not exist prior to version 5, so I'm not sure what >>> you are comparing to. The output looks normal when tested here. >> >> I'm seeing the y-axis text right up against the screen edge, and on the >> right side is three characters of white space. But, I see that margins >> are specified in this example. Note, however, that I'm seeing an error >> message for the "set multiplot" line: >> >> gnuplot> load 'matrix_index.dem' >> "matrix_index.dem", line 29: warning: must give margins and >> spacing, continue with spacing of 0.05 >> Hit return to continue > > That warning seems unnecessarily alarming. > All it means is that in the absence of an explicit spacing keyword > it will use the default value. I agree. gnuplot does defaults all the time. I would say though the default is kind of small and I'd prefer it to be in characters because that is what "set tmargin", etc. uses. Screen is good for fine-tuning, but something near whole numbers, rather than .05, is a better place to start. >> If I just remove the "margins etc" option, the warning disappears and >> the spacing looks good, very balanced. Should we just remove the >> margins? Or keep them in as a way of testing the option? Here's what >> looks good for me: > > [shrug] Tastes differ. > I think it looks better with identical left and right margins. > To me the default looks lopsided. That's because it is, now that I look at. The top two plots have titles so they are smaller than the bottom two plots. Setting the margins must fix the axes locations rather than the overall space. Option margins char 5,3,2,4 spacing char 6,3 looks much better than the current setting though. Dan |