Allow variable symbol sizes in ScatterPlot

[kjordahl]

Move functionality from VariableSizeScatterPlot to parent ScatterPlot class. Also allows ColormappedScatterPlot to use variable marker sizes, closing #97.

[kjordahl]

I have two remaining concerns before I think this is ready to be merged:

1. Performance could be better, especially for the case of constant symbol size.
2. There is still a bug in indexing in ColormappedScatterPlot. I think it is fixed in ScatterPlot now.

[kjordahl]

I've addressed my previous concerns, and IMO this is now ready to merge. Any other opinions?

[pberkes]

If you can give me a couple of days, I'd like to update the docs.

[pberkes]

I made a few updates to examples and documentation, the rest looks great. Unless somebody objects, I'll merge it tomorrow.

@kjordahl do you see anything stopping us from lifting the color up to scatterplot? performance?

[kjordahl]

@pberkes Thanks.

I still think everything should ultimately go into the base scatterplot. I am concerned about performance, but don't have good tests for it yet. More than that, it would be really nice if every attribute that could apply to each marker could be specified as an array (particularly marker). Doing that right would take a little more work, but I don't think it's unreasonable.

Meanwhile, I think we should merge this as a step forward (that also fixes a previous bug), and I'll make a new branch to start working on a unified ScatterPlot class.

[cfarrow]

On Sun, Mar 10, 2013 at 8:50 PM, Kelsey Jordahl notifications@github.comwrote:

@pberkes https://github.com/pberkes Thanks.

I still think everything should ultimately go into the base scatterplot. I
am concerned about performance, but don't have good tests for it yet. More
than that, it would be really nice if /every/ attribute that could apply to
each marker could be specified as an array (particularly marker). Doing
that right would take a little more work, but I don't think it's
unreasonable.

You might find the flyweight pattern userful:
http://en.wikipedia.org/wiki/Flyweight_pattern. It handles performance and
memory issues associated with the "most things are the same but some are
different" pattern quite well. (The write up in Design Patterns is much
better than the wikipedia page, IMO.)

Meanwhile, I think we should merge this as a step forward (that also fixes
a previous bug), and I'll make a new branch to start working on a unified
ScatterPlot class.

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/101#issuecomment-14694332
.

The information contained in this message is Enthought confidential & not
to be disseminated to outside parties without explicit prior approval from
sender. This message is intended solely for the addressee(s), If you are
not the intended recipient, please contact the sender by return e-mail and
destroy all copies of the original message.

[pberkes]

@kjordahl I agree about merging this PR, but was interested in your opinion since you worked on this class.

@cfarrow Thanks for the pointer!