protecting from div by zero when bbox has no extent

[ahocevar]

In GEM's Modeler's toolkit, there is a chance that model output results in a shapefile with just one point feature. The extent of such a shapefile will have no width and height. This pull request fixes the issue by protecting from div by zero exceptions that happen during updatelayers.

[ingenieroariel]

This is an interesting fix, can you share with us one of those shapefiles so we can add it to the default set of test data and add it to the upload tests in the integration test suite?

Even without that file I would be happy for this to go in because the logic is sound. I have another suggestion, though:

try:
    dataAspect  = dx / dy
except ZeroDivisionError:
    dataAspect = 1

This seems a little bit more explicit and easier to read, but at the end of the day is a matter of style.

[ahocevar]

See http://dl.dropbox.com/u/5489321/zeroextent.zip for an example shapefile that causes this issue.

[ahocevar]

Regarding the try/catch or avoid zero value approach: I don't know about Python coding style or good practices, I only did what I would have done in JavaScript, where try/catch is usually only used when there is no other way to deal with the problem.

[ingenieroariel]

This is the first time I used the github interface to automatically merge the pull request. Now I have to find out how to propagate this change to synth :)