[ZODB-Dev] Error running test.py in zodb3.1b

[Tim Peters]

[Magnus Lycka]
> Where can I learn more about this bug? (Not tonight
> though. I hope I'll get some sleep despite these
> news...)

It was a wild load (a read from a nonsense address).  The code was clearly
wrong (once found <wink>).  The only person who saw it blow up, though, was
David Leblanc, on a Windows box with an AMD processor.  I tried like hell to
get it to blow up, and never could, *except* when I ran the binaries that
David had compiled on his machine and shipped to me.  They didn't blow up
when I compiled them myself, and not on any of the four Windows boxes I
tried that on.  Neither was anyone able to provoke a blowup under Linux.
When it failed for David (and for me when running his binaries), you never
got off the ground:  the mere attempt to initialize the persistence
machinery died with a memory fault (due to the wild load).

> I read the "[ZODB-Dev] ZODB3-3.1b1" mails from last
> week, but I didn't get a lot wiser. I don't quite
> understand when the buggy function is called.

During initialization of ExtensionClass objects.

> I'm naively hoping that maybe my app isn't affected
> after all? :)

You're definitely getting the wild load if you're running ZODB.  Whether it
harms you or not is much harder to say -- a wild load can pick up anything.
It only blows up at once if you're lucky <wink>.

> ...
> Do I understand correctly that this function would
> be called if I for instance did "type(6)(3)"?

No, there's no ExtensionClass involved in that.