Stefan Matthias Aust wrote:

> Well, I was interested in how many different categories the most often used
> methods are put.  #initialize (which is implemented 311 times) can be found
> in 26 different categories.
> 
> 175->#initialization 126->#'class initialization' 67->#'as yet
> unclassified' 53->#initialize 33->#'initialize-release' 7->#private 6->#all
> 4->#'private-initialization' 2->#'initialize-reset-release' 1->#events
> 1->#misc 1->#'class variables' 1->#'simple keyboard' 1->#'parser support'
> 1->#'initialization-release' 1->#simulation 1->#'ST initialize' 1->#'like
> fullCopy' 1->#public 1->#'common requests' 1->#initializing 1->#accessing
> 1->#'instance creation' 1->#application 1->#project 1->#Initializing
> 
> This makes the core system really untidy IMHO.  If I look for a certain
> method, I should only need to look into one or only a few obvious
> categories.  It seems however that people life out their creativity (or
> laziness in the case of 'as yet unclassified') in picking strange category
> names.

... or that the "new category" menu selects its available choices poorly. It
might, for instance, present all existing method categories in all the
classes in the same system category (of classes). And there might be a
standardized list of categories somewhere that is always available. And it
may look in subclasses as well as superclasses.

And to classify an "as yet unclassified" method, you might be presented also
a choice of the categories in which this selector has been placed elsewhere.

Nice optimization essay btw!

Henrik