The documentation http://docs.python.org/library/configparser.html states that optionxform() is used only beginning ConfigParser.ConfigParser, whereas it is ALSO in effect for ConfigParser.RawConfigParser. (As I checked in the source)

The documentation may be poorly worded but `optionxform()` has always been used also by RawConfigParser. It has been that way since the introduction of this class in Python 2.3 (previously there was only one ConfigParser class which used `optionxform()` since at least Python 1.6).

Both the documentation and the code have seen significant updates in Python 3.2 and there the situation is more explicit.

I'll rephrase the docs for 2.7.

New changeset c7ce67c9237a by Łukasz Langa in branch '2.6':
Closes #11786: ConfigParser.[Raw]ConfigParser optionxform().
http://hg.python.org/cpython/rev/c7ce67c9237a

New changeset a6b772599594 by Łukasz Langa in branch '2.7':
Merged solution for #11786 from 2.6
http://hg.python.org/cpython/rev/a6b772599594

2.5 and 2.6 are in security mode.  Other bug fixes, build changes, documentation improvements, etc. should not go in these branches.  Your commit does not break anything, but for process clarity, please back it out.

Sorry about that. Since I'm not technically touching the source code, I thought the security fixes restriction does not necessarily apply. Especially that my patch only updates what ends up here:

http://docs.python.org/release/2.6.6/library/configparser.html

The point is, all gratuitous changes to such branches are a nuisance to the release manager who decides what needs to go into a security release.

Also, doc changes in 2.6 are almost completely gratuitous, since the docs are never again built and put online somewhere.  (Source-only releases don't have build doc releases.)

Yup, the fact that the docs won't be rebuilt ever again makes my 2.6 commit effectively pointless and harmful for the RM. I will clean it up when I get home.

Thanks for the thorough explanation.