Gratuitous Change (Was: Re: "in" operator for strings)

[bowman]

Neil Schemenauer <nas at arctrix.com> wrote in message
>
> scope changes are the only ones that affect existing code.  You
> have to forgive me if I think your spreading FUD.

language changes that affect existing code have their own way of spreading
FUD. I'm not supporting a major software effort in Python at this time, just
some relatively simple tools for my own use. However, I've had a few
previously working tools fail or report errors since upgrading. This doesn't
do a lot to encourage me to undertake a project with a projected lifetime of
10 years or so in Python.

OTOH, the actual deliverable apps I am working on at present have already
passed the 10 year mark since their original release, have been constantly
enhanced and expanded, and ported to several platforms. They are written in
C, and not a line of code has had to be changed because of underlying
changes in the C language. I believe it is only in the last release of the
gnu toolchain that the decision was taken to drop support of the long
obsolete K&R declarations. It is this sort of stability that has made C,
despite its warts and blemishes, a widely used language.

Personally speaking, I am not a conservative, but there is something to be
said for conservative principals when dealing with a language. A parallel
would be when Java's event model changed -- altogether it was a good idea,
but forced one to make the choice of using the new model, or staying with
the old, knowing full well that many users were not going to upgrade their
jvm's until absolutely necessary.