Python bug in Windows 8--report now, or later?

Terry Reedy tjreedy at udel.edu
Sat Sep 17 17:19:26 EDT 2011


On 9/17/2011 2:01 PM, Kevin Walzer wrote:
> I have been testing my Python application on the just-released developer
> preview of Windows 8 and have noted an error: the application does not
> create an app folder in the user's "application data" directory. This
> causes the app to crash on startup. Manually creating the directory
> solves the problem. Since the app uses an os.mkdir() call to create the
> directory, and since the app runs fine on Windows 7, my guess is that
> the bug lies somewhere in the interaction between Python (I'm using
> ActivePython 2.7) and Windows.

We need more that guesses to act. I think is premature to call this a 
'Python bug'.

> Here's the relevant code:
>
> #make preferences directory if it does not exist
> def makePrefsDir(self):
> self.appdir = os.path.join(os.path.join(os.environ['APPDATA'], 'MyApp'))
> if not os.path.exists(self.appdir):
> os.mkdir(self.appdir)
>
> I realize that this developer preview of Windows is still at somewhere
> between alpha- and beta-level, and it's possible things will get better.
> Should I wait to report this as a bug until Windows 8 is released, or do
> the Python developers test Python on pre-release versions of Windows?

3 days ago (Sept 14) someone asked about 'Windows 8 support' on pydev 
list. The answers were 1) 2.7 and 3.2 appear to run fine on the Dev 
release (but there was no report of test suite results); 2) Python 
directly uses so little of the Win interface that problems are not 
anticipated; 3) applications might have to make Win 8 specific 
adjustments and should test before the release.

Of course, if MS accidentally changes thinly wrapped system calls such 
as os.environ, .exists, and .makedir, there will be a problem but that 
is their bug. My impression is that they are not intentionally breaking 
such things.

I anticipate 3.3 and some future 2.7.z will be officially supported (and 
tested) on Win 8.

-- 
Terry Jan Reedy




More information about the Python-list mailing list