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

[Chris Gonnerman]

----- Original Message -----
From: "Neil Schemenauer" <nas at arctrix.com>
Subject: Re: Gratuitous Change (Was: Re: "in" operator for strings)


> Can you give some examples of changes you have to "put up with"?
> The only controversial changes in 2.1a2 that I know of are nested
> scopes, function attributes and weak references.  The nested
> scope changes are the only ones that affect existing code.  You
> have to forgive me if I think your spreading FUD.

Well, that hurts, but I guess I have it coming.  FUD is my enemy, let me
spread no FUD.

My primary complaint is against changes to the language semantics.  I am
against change in a language in general when the language is already IMHO
the finest I've ever worked with.

You are probably right, in that these changes in 2.1 are not so bad.  On the
other hand, I probably won't use them.  Heck, I still import string.  I'm
not
against the changes in 2.0, although I don't use most of them; I still have
servers
running 1.5.2 that would be a pain to upgrade.

One thing I see a lot in 2.0 code that is legal but UGLY is:

    nstr = ''.join((s,t,u,v))

(for instance) but I guess you can write ugly code in any language.

My only real complaint is the need to rebuild ALL MY EXTENSIONS
every time the version is upgraded.  The build process is painless under
Linux but a pain under Windows and I must live in both worlds.  Even
under Linux, I wind up rebuilding a LOT of software because I use a LOT
of Python.

Yeah, I'm rambling.  The point is (yes, I have one) that I think all changes
are potentially dangerous.  I do respect and trust GvR though, so perhaps
I'll try 2.1 when it's not alpha anymore.

BTW I've lost track... is 2.1 stackless or was that just a rumor?