Logged In: YES user_id=31435 Curious! The Uninstall Start Menu entry for 2.4a1 worked fine for me, WinXP Pro. If it's a "deep" problem, I expect we could just toss the Uninstall menu entry -- all versions of Windows I care about have the Add/Remove gimmick, which most Windows users expect to use for uninstallation anyway.
Logged In: YES user_id=11105 Hm, it works for Martin's MSI (the one downloaded from python.org), but not for those created myself from the CVS tree. Using the menu entry is much faster for people testing the installers (Add/Remove take a long time to open on my machine). Normal users shouldn't uninstall Python anyway ;-) [5 minutes later] I guess he has set 'alpha = 0' in the msi.py script he used to build the installer. The 'alpha' version seems to use two conflicting product codes for the MSI: one hardcoded, at the top of the script, the other created by msilib.gen_uuid() at line 81. I'll try it out again, and report back.
Logged In: YES user_id=21627 There was a bug in the script that would use different product codes for an alpha (i.e. the "final" product code) for uninstall. I have fixed this in my CVS sandbox before the release of 2.4a0, but not yet committed - will do so tomorrow.
Logged In: YES user_id=31435 Thomas, do yourself a favor and Google on "Raymond Chen add remove" -- you'll find an hilarious/depressing weblog about *why*add/remove is so slow to come up in modern Windows. Believe it or not, there are impossible-to-fix reasons for it!
Logged In: YES user_id=11105 Tim, I already know and read from time to time Raymond's weblog - but it is a useful pointer for the casual reader of Python's bug reports.
