On Saturday 26 May 2007, Joseph Armbruster wrote:
I noticed that one of the parts of ConfigParser was not using "for line in fp" style of readline-ing :-) So, this will reduce the SLOC by 3 lines and improve readability. However, I did a quick grep and this type of practice appears in several other places.
Before the current iteration support was part of Python, there was no way to iterate over a the way there is now; the code you've dug up is simply from before the current iteration support. (As I'm sure you know.)
Is there motivation for these changes other than a stylistic preference for the newer idioms? Keeping the SLOC count down seems pretty minimal, and unimportant. Making the code more understandable is valuable, but it's not clear how much this really achieves that.
In general, we try to avoid making style changes to the code since that can increase the maintenance burden (patches can be harder to produce that can be cleanly applied to multiple versions).
Are there motivations we're missing?