It all started when a cross-platform Python application, Blogofile, tried to use the standard regular expression module, re, to munge paths.
I was investigating a small Blogofile bug on Windows which involved the regular expression library operating on Windows paths. Blogofile's test server was using re.sub to map requests to a subdirectory that would otherwise be directed to the base directory of the server. This worked fine on Linux paths, but on Window's paths the match wasn't happening.
This time I quote the text that I want to substitute. The match is made but the re module appears to helpfully interpret the "\t" as a tab. I'm glad I happened to use a directory that starts with a "t," or I wouldn't have noticed this problem.