re bad leadership

Paul Boddie paul at boddie.net
Wed Jul 25 12:38:39 CEST 2001


Thomas Wouters <thomas at xs4all.net> wrote in message news:<mailman.995988041.22317.python-list at python.org>...
> On Mon, Jul 23, 2001 at 11:48:00PM +0100, Robin Becker wrote:
> 
> > it doesn't matter what the leadership says is right. The popular vote is
> > against the stupid float '/' operator.
> 
> Really ? I don't see a popular vote. I just see a very long and very heated
> discussion with a small group of repeat offenders (oops, did I say
> offenders ? I meant posters) that just keep on proclaiming their
> righteousness with little in the way of new arguments (either way, for that
> matter.)

On the contrary, I see some pretty strong arguments for not changing
the behaviour of the / operator. The argument that changing the
existing semantics breaks potentially huge amounts of existing code,
cannot be trivially tested for, and is therefore highly undesirable,
has not been adequately refuted. It is not sufficient to say that "new
style division" is nicer, looks better, or is "the way we wanted it
from the start"; such statements do not constitute a response to that
concern - they merely attempt to change the subject of discussion.

I see that the burden should be on those proposing these changes to
address the existing strong arguments against them, rather than being
on those against the changes to raise countless arguments only to see
them brushed aside in an ideological haze by people unwilling to
consider legitimate points of view.

The PEP document in question does not summarise the valid concerns of
the community, many of which have emerged in the recent discussions. I
can understand the author not having time to update it since these
discussions started - I have much less ambitious documents to maintain
myself, and it can be weeks before I get the chance to implement
changes in them - but I cannot believe that the original document did
not consider such obvious pitfalls in the proposed scheme, especially
before an implementation was produced.

Until such time as the PEP attempts to summarise the issues properly,
this discussion will continue with all sides feeling that their views
have not been heard, taken into account, or addressed.

> I strongly suspect the change will actually *fix* more code that was relying
> on float inputs than it will break code that depended on int/int being
> special-cased.

No-one ever fixes code, or should seek to have their code fixed, by
having the ground rules changed.

Paul



More information about the Python-list mailing list