gratuitous new features in 2.0

[Grant Griffin]

Tim Peters wrote:
> 
> [Grant Griffin]
> > I know that it's very important not to break old code, and I know Guido
> > and other language designers are very committed to that (which is
> > generally A Very Good Thing.)  But it's a tradeoff: I would posit that
> > people who have a problem with old code vis-a-vis "to" could either
> > change it, or just stick with 1.5.2. <<toheck with 'em>>
> 
> Grant, just yesterday you were posting on behalf of the commerical world in
> praise of "stability". 

But a wise man once said something to the effect that the key to any
successful philosophy was having enough contradictory points that you
can justify doing what you wanted to do anyway. ;-)

> Nobody is amused when their old code fails to
> compile under a new release, and Guido hasn't risked a new keyword since ...
> 26 January 1994, with release 1.0.0.  That's when he released contributed
> code for the then-new "lambda", and it doesn't help matters any that he's
> regretted that ever since <0.8 wink>.  The only reason he risked it then is
> that he figured he could risk a *little* incompatibility in the first
> release of the 1.0 line.
>
> > Also, so many points of Tim's "Python Philosophy" point in favor of "to"
> > and against ">>"* that I guess my personal instinct is that The Greater
> > Pythonic Good would be served by it.
> 
> "No new keywords" has been one of the few absolute rules here.  Although
> Guido often shows signs of being on the brink of breaking it <wink>.

In his role as a language designer, I respect and admire Guido's
steadfastness here.  But being just a member of the Peanut Gallery
myself, this issue seems much less important.  I look at it a tradeoff:
you lose something in terms of compatability and you gain something in
terms of readability and Pythonic (in my interpretation).  If one small
to-letter word can be introduced every six years or so, the commercial
world will regard that as stability.  <<we usually think in the scale of
six minutes, not six years>>

In C, for example, I don't think anybody is served by having six
different ways to use the "const" keyword, based on position: if someone
had introduced "const_ptr" or something like that, it would make it
easier to write and read correct code C.  (And I bet it would be easier
for compiler authors, too. ;-)

> 
> > *BTW, what's wrong with ">" for redirect?  Heck, anything that both DOS
> > and Unix do can't be _all_ bad! <wink>
> 
> To both Unix and DOS users, ">" implies overwrite and ">>" implies append.
> The print semantics are much closer to the latter.  Ditto to the meaning of
> ">>" in C++ in an output context.  Using ">" instead would *really* be
> gratuitous novelty.

OK...nevermind. :-)

emily-littella-ly y'rs,

=g2
-- 
_____________________________________________________________________

Grant R. Griffin                                       g2 at dspguru.com
Publisher of dspGuru                           http://www.dspguru.com
Iowegian International Corporation	      http://www.iowegian.com