[reportlab-users] Spider plot checked in

Robin Becker reportlab-users@reportlab.com
Tue, 3 Dec 2002 18:23:58 +0000

In article <PGECLPOBGNBNKHNAGIJHAEICEBAA.andy@reportlab.com>, Andy
Robinson <andy@reportlab.com> writes
>> Nice. The real challenge with this is, of course, multiple data
>> series of different lengths plus circular axes, which would also
>> be useful for polar plots.
>This one had exactly what was needed for a customer demo so
>I will happily leave that challenge to others, or to 'next
>time'.  In the meantime I submit that
>(a) spider plots are a deranged sociologists' adaptation of a bar
>    or line plot anyway, carry less information and have many
>    gotchas
>(b) anyone calling it with negative numbers should be shot
>(c) anyone calling it with different length series should
>   be hung, drawn, quartered and then finally shot :-)
transparency is available with PDF 1.4, but sadly it has more than just
alpha ie there's shape which somehow controls the blending at the edges.

If we had access to the internal rendering engine we could do the same
in other models. All the smart people in the Gnome/Mozilla world seem
unable to make libart work well at bitmap intersections. I know that PDF
uses a fixed point representation, but that just puts off the evil day
when you have to decide what things cross and what don't; when that day
comes you need to go onto the grid and the set of all grid crossings is
of zero measure in the original space so then you need to ........ zzzzz
Robin Becker