Nice find.
That one for concatenating strings is good and works for every language, not just Python.
In Data Aggregation I noticed something that wasn't mentioned in the text. I increased the performance in Atres dramatically a few weeks back by removing the temporary variables in most functions and using member variables instead. Basically the allocation of temporary variables is quite expensive. I've had a speed-up over 100% (from 30 FPS to 47 FPS).
Lol, xrange instead of range is mentioned there.
Though, I do not agree with one thing on the map/dict initialization. Exceptions are VERY slow and should be avoided whenever possible, especially in cases where they are used a lot like in a loop with many iterations. We had an huge overhead because of this in AprilUI at one point. After I removed the exception handling and instead changed to a different method (it was reading an XML document and checking the availability of an attribute through an exception) and implemented an alternative, suddenly reading the XML definition of our dataset became so fast that you wouldn't notice it anymore. Before that there was a significant an actually time required to read it.