I've been looking at a javascript library for drawing graphs, Dygraph, source on github.
It has several enormous classes, such as here and here.
Personally, I don't think justifications for large classes are valid, since it goes against principles of OOP, such as the single responsibility rule. In this day and age, even phones have plenty of power, so except for the most fine-grained objects, it seems like bad practice. But even then, javascript often uses boxing anyway.
Now, I don't really know what javascript pre-compiling does besides shortening symbol names and removing whitespace, but perhaps it makes large classes even less significant in helping performance?
In short, my question is, if the above code was refactored to abide with the single responsibility rule, what sort of performance degradation might be expected? I know there's a lot of code there that could be analysed, but perhaps there are some rule-of-thumbs that code statistic gurus could apply? (I know I've seen lint engines that do these kinds of estimations)
Edit: my opinion is that it is bad practice and should be avoided unless critical to performance. I am just giving the authors the benefit of the doubt