[Python-Dev] Re: os.path.commonprefix breakage

[M.-A. Lemburg]

Gordon McMillan wrote:
> 
> M.-A. Lemburg wrote:
> 
> > ... just look at what your browser does
> > when you request http://www.python.org/search ... the server
> > redirects you to search/ to make sure that the links embedded in
> > the page are relative to search/ and not www.python.org/.
> 
> While that seems to be what Apache does, I get 40x's from
> IIS and Netscape server. Greg Ewing's demonstrated a Unix
> where the trailing slash indicates nothing useful, Tim's
> demonstrated that Windows gets confused by a trailing slash
> unless we're talking about the root directory on a drive (and
> BTW, same results if you use backslash).
> 
> On WIndows, os.path.commonprefix doesn't use normcase
> and normpath, so it's completely useless anyway. (That is, it's
> really a "string" function and has nothing to do with paths).

I still don't get it: what's the point in carelessly dropping
valid and useful information for no obvious reason at all ?

Besides the previous behaviour was documented and most probably
used in some apps. Why break those ?

And last not least: what if the directory in question doesn't
even exist anywhere and is only encoded in the path by the fact
that there is a slash following it ?

Puzzled by needless discussions ;-),
-- 
Marc-Andre Lemburg
______________________________________________________________________
Business:                                      http://www.lemburg.com/
Python Pages:                           http://www.lemburg.com/python/