I have a version of Canviz for PyJamas python->javascript compiler

[Joel]

Hi folks;
The project I've been working on needs the functionality of
something
like Canviz, but we need to use it in embedded windows, ideally in the
PyJamas development environment (pyjs.org). I've translated the
algorithms of Canviz to python in a form that the PyJamas compiler
can
use to generate the appropriate widgets. I've *temporarily* put a
demo on
line at http://staff.psc.edu/welling/canviz_demo/ . To use it, click
on the
bottommost tab on the left edge and then start selecting sample graphs
to try.
Since the algorithm follows that of the original Javascript Canviz
so
closely (although none of the original code remains), I thought I
should
mention it to this group. After some additional tweaking I hope to
hand
it off to the PyJamas people to include as an 'add-on' to their
software.
There is a remaining problem with text positioning that arises
because
different browsers handle the tabbed panels differently; I'm still
sorting that out. I've added culling of graph entities based on their
bounding box and the known bounding box of the page, and
simplification
of very small entities (less than 2x2 pixels). These things help with
the very large graphs we're trying to support.

-Joel
wel...@psc.edu

[Ryan Schmidt]

Thanks for letting us know! If it moves somewhere else, please let us know as well.

Can you share the code or the algorithm you used for small entities and for culling unseen entities? It sounds like it's probably simple to implement, I just haven't focused on optimizations in Canviz yet.