Re: Re: Precision graphics
- To: mathgroup at smc.vnet.net
- Subject: [mg16785] Re: [mg16631] Re: Precision graphics
- From: Mark Evans <NoSpam at NoWhere.Net>
- Date: Wed, 24 Mar 1999 17:07:11 -0500
- Organization: gte.net
- References: <7d73kj$chv@smc.vnet.net>
- Sender: owner-wri-mathgroup at wolfram.com
Bob Hanlon,
Thank you for that but this is not the most common problem. I hardly ever
use more than two lines in a plot label.
In my remark (a), the problem is with the tick numbers running along the
horizontal axis. Currently, the only way to resolve that problem easily is
to enlarge the plot such that it may no longer print on the page. A much
cleaner solution would be horizontal tick numbers that read vertically
upwards. These could not possibly overlap.
In my remark (b), the real problem involves the tick marks running up the
vertical axis. This is the plot range. By default, Mathematica decides
what these tick marks should be. That is generally a good thing. Yet from
one plot to the next, Mathematica may change its decision, resulting in
misalignment of the two plots. From a report writing standpoint this
situation is painful. What I mean is that you may have on one plot vertical
ticks {0,0.5,1.0 } and on the next {0.01234, 0.34987, 0.76584}. Things get
really bad when Mathematica decides that it?s time for scientific notation
on your vertical ticks. Then you have a full expression including
ten-to-the-power-of multipliers.
One ends up having to play guessing games with the PlotRange option to
achieve compatible tick marks on the vertical axis of two plots. When you
change the PlotRange by hand, Mathematica must, by definition, recompute the
vertical ticks.
The other way around this vertical-tick alignment problem is a very tedious
expenditure of time using PaddedForm to label the ticks by hand. When you
are writing a report with dozens of plots, this exercise gets painful very
fast.
These are the main reasons why Wolfram should implement a graphics alignment
option of the kind I indicated.
One way that Wolfram could achieve the result is to include ?meta?
information in each plot. The meta information would communicate which
pieces of the post script code describe the features which must be changed
to realign plots.
Another way to achieve the result would be to break apart the current
monolithic PostScript object into constituent pieces. One block of
PostScript for the rectangle, another for each of the tick areas, and
another for each of the label areas.
In some cases it is better if the ticks are all but dissociated from the
plot. When you have scientific notation going on, it would be better to
have fewer ticks with special indicator lines running from the tick label to
the corresponding grid line on the plot. That way the ticks could be nicely
spaced with no sacrifice in clarity.
So there are many things to be improved.
Mark