Mail Archives: djgpp/2003/06/26/05:45:15
deckerben <bdeck AT lycos DOT co DOT uk> wrote:
> I really understand the DOS issue. But this would not be true under NT
> systems, where true filename cases are reported correctly.
I'm not all that sure about "are". They *can* be, just like they can
be under 9x. Whether they are or not is a configuration option.
E.g. the 9x versions of Explorer let you select whether to show
all-uppercase names in all-uppercase or all-lowercase. I've yet to see
anyone who actually toggled that option from its default lower case.
> Fooling with this causes problems. I was wondering why during some
> Python distutil builds, it was *supposed* to create an upper-case
> subdirectory, only to create a lower-case one, etc. I was banging my
> head wondering why the glob test was failing (I thought something
> was *really* wrong).
Some *is* really wrong, but I disagree that the behaviour of DJGPP is.
It's the expectations hard-coded into that glob test that simply are
false, for the case of any DOS-style platform up to and including both
OS/2 and all versions of Windows. It really should notice that it's
on a system where filename case is usually preserved, but *not* at all
respected.
> I am recommending all NT-based PythonD users to set FNCASE=y.
Then you should als recommend they be prepared for some rather strange
behaviour in all kinds of tools.
Trying to depend on the case of filenames in any way is usually a
futile exercise on Windows.
--
Hans-Bernhard Broeker (broeker AT physik DOT rwth-aachen DOT de)
Even if all the snow were burnt, ashes would remain.
- Raw text -