Since the huge update, I'm having to fix lots of API changes with my python interfaces to caffe. How do I wade through the commit stream to find out what all needs to change?

This is my most recent problem. I try this:

classifier = caffe.Classifier(deploy_file, snapshot_file,
             image_dims      = (height, width),
             mean            = image_mean,
             raw_scale       = 255,
             channel_swap    = channel_swap,
             )
classifier.predict([image], oversample=False)

And I get this error:

ValueError: operands could not be broadcast together with shapes (3,227,227) (3,256,256) (3,227,227)

Apparently, the mean_file has to be cropped before passing it to the Classifier constructor now. How was I supposed to know that? I don't see it in any of these commits. And how am I supposed to do it? Should I be cropping it myself manually, or should I try to use Classifier.transformer.preprocess()?