Re: [Gnuplot-info] tics, mtics, and logscale

[Tait <gnu...@t4...>]

> > Consider the following:
> > reset
> > unset key
> > set xrange [2**4:2**8]
> > set logscale x 2
> > set grid xtics mxtics ytics
> > set ytics 1
> > set multiplot layout 5,2
> >   set title 'default'; set mxtics 8; plot log(x)/log(2)
> >   set title 'xtics, 8'; set xtics 4; set mxtics 8; plot log(x)/log(2)
> >   set title 'xtics, 2'; set xtics 4; set mxtics 2; plot log(x)/log(2)
> >   set title 'xtics, 3'; set xtics 4; set mxtics 3; plot log(x)/log(2)
> >   set title 'xtics, 4'; set xtics 4; set mxtics 4; plot log(x)/log(2)
> >   set title 'xtics, 5'; set xtics 4; set mxtics 5; plot log(x)/log(2)
> >   set title 'xtics, 6'; set xtics 4; set mxtics 6; plot log(x)/log(2)
> >   set title 'xtics, 7'; set xtics 4; set mxtics 7; plot log(x)/log(2)
> >   set title 'xtics, 8'; set xtics 4; set mxtics 8; plot log(x)/log(2)
> >   set title 'xtics, 9'; set xtics 4; set mxtics 9; plot log(x)/log(2)
> > unset multiplot
> > 
> > It produces output like: http://i.imgur.com/AvUUZ.png
> > 
> > In the default plot, you can see the minor tics are spaced logarithmically,
> > as one would expect. As soon as I set the xtics spacing explicitly,
> > the mxtics lose their spacing.
> > 
> > But the oddity does not end there. The "xtics, 2" graph is probably
> > expected. But then the behavior for powers-of-two mxtics is nothing like
> > the expected logarithmic spacing (that works normally when xtics are not
> > set explicitly). I can't explain the non-powers-of-two mxtics at all. It
> > appears the spacing is uniform (not logarithmic), but different after
> > the first xtic for some reason.
> > 
> > In any case, the help doesn't adequately explain what's happening here,
> > and I'm sort of wondering if maybe there isn't a bug or two. Can anyone
> > explain?
>
> hello Tait,
> thx for your report. It would help when you tell what version of gnuplot
> you are using so developers can check that.
> 
> re,
>  wh

I was using 4.4. I just tried it out on 4.6rc1 and the behavior is the same,
there.