Deposing Dictators

Tim Peters tim.one at home.com
Mon Aug 13 05:14:49 CEST 2001


[Tim, not surprised that this is the only part quoted <wink>]
> Guido isn't obsessed with Carnegie Mellon.  We have years of reports
> from many Python users about the real-life consequences of division
> semantics, and VPython is notable primarily because they cared so much
> they forked Python to change it.  Should Guido ignore that?  Of course
> not.

[Arthur Siegel]
> Alice forked Python on the case sensitivity issue.

Yes.

> Where is the evidence on the VPython site that they have forked
> Python on the div issue?  The docs specifically caution on the
> div semantics. Obviously unnecessary if they were using
> a patched Python.

I don't even know where the VPython site is.  Looking back at what I did
see, the closest I can find is Bruce Sherwood's prospective discussion about
why:

    http://mail.python.org/pipermail/idle-dev/2000-April/000138.html

and David Scherer's explanation of the patch he implemented to change
division:

    http://mail.python.org/pipermail/idle-dev/2000-April/000140.html

If they didn't fork after all, cool.

> So yes I was disturbed that Guido's original announcement
> of the PEP0238 acceptance was a few sentences long, *at least *
> one of which was wrong - the VPython fork bubbermeister.
>
> And I am disturbed to hear Tim Peters repeat it as fact.

My apologies.

> Worst case scenario we are talking about a student, warned
> of Python's behavior, not being careful to apply what he knew.
>
> Took him three hours to debug. Poor guy.
>
> I've fallen into every trap there is to fall into in learning
> Python.  Never thought to blame it on the language.
> Kind of thought the language was doing exactly what
> the documentation says it would do, and *I* did something
> wrong.  The finding it, and fixing it - was invaluable.
>
> But you all know all this.
>
> I can't help feeling half the time that I am up against an
> entire community playing dumb.
>
> Tim responds to everything except my central point.  That
> there is an excellent basis to assert that if the universe
> of Python was VPython, the change makes little sense. ...

I'll spell it out:  I don't care about VPython.  Never did.  That's
something you go on about, pretty much all by yourself.  If the rest of the
community doesn't join you, the obvious reason isn't that they're playing
dumb, but that they don't care about VPython either.

While Alice and VPython played some brief roles in focusing attention on the
existence of "a problem" here, once the problem was appreciated it gained a
life of its own, quite independent of who first made a big stink about it.
If Carnegie Mellon fell into a black hole tomorrow, taking Alice and VPython
with it forever, it wouldn't affect the motivation for PEP 238 at all.

It's not about VPython, although the VPython folks have their own reasons
for wanting this.  It doesn't matter to me either whether their analysis is
right or wrong for VPython, only whether Guido's analysis is right for
Python.  I agree with his reasoning wholeheartedly (read the PEP).  I'm not
sure that, were I him, I would actually make the change, though:  despite
that "it's right", it's a painful change all around.  But it's not about
VPython regardless.  Neither is it visibly about CP4E in any larger sense.
It's about the language's numeric rules making sense (read the PEP).





More information about the Python-list mailing list