re bad leadership

Paul Boddie paul at boddie.net
Thu Jul 26 10:53:28 CEST 2001


Chris Barker <chrishbarker at home.net> wrote in message news:<3B5EEED7.C9C7117B at home.net>...
> 
> I'm sure that was the intention, but we need to keep in mind that the
> PEP proposed two types of division, NEITHER of which is the current one,
> so in order to keep total backward computability, we would need to add
> two operators, which would really get messy.

Well, having read the PEP more thoroughly I accept that the two
division operators will work differently from the existing one, given
that they will each provide consistent operations across all numeric
types, and this would be a good thing to have when designing a *new*
language. (Although many people accustomed to working with floating
point numbers would probably explicitly signal ceiling and floor
operations rather than have them implicitly performed by some strange
new operator.)

However, it's arguably more relevant to discuss the fact that this
changes well-defined semantics in the language. It's extremely
important for pragmatic reasons to discuss how much damage this change
will cause. It's not at all acceptable or interesting to respond to
such concerns with how "nice" this will make Python's numeric system.

> And, this has been repeated MANY times, but despite the working in the
> PEP "Rationale" section, there are some very good and compelling reasons
> to have the proposed division that are relevant to experienced
> programmers, not just people learning. Guido recently said he will
> re-write the Rationale section to include this. I have been convinced by
> those reasons, though I think we need a better way to handle the
> backward computability issue.

I'm sure we'd all love to see a nice and consistent numeric system,
but Python is a mature language in widespread use. It's tempting for
some to think that with a supposedly inevitable huge influx of new
users that the volume of existing code will be dwarfed by new code
within a short period of time, but regardless of whether this will
happen or not, I think such "ideological" changes to the language with
such limited justification send the message to numerous
highly-respected community members that their work is considered to be
of little value and can be sacrificed to the "language gods" on a
whim.

Paul



More information about the Python-list mailing list