Home Original page

Case insensitivity

Tim Rowe digitig at cix.co.uk
Fri Aug 3 16:58:00 EDT 2001 
In article <mailman.995637214.19543.python-list at python.org>, 
guido at digicool.com (Guido van Rossum) wrote:

> > You seem to want to change the core of the language every other day.
> 
> Eh?  Evolution of the language is inevitable.  We *need* to fix old
> design mistakes.  When it's possible to do so without breaking old
> code, everybody cheers.  Sometimes it's not possible without breaking
> old code, and then I try my darndest to find a way that will let old
> code work.

But you seem to have the idea that there is a "perfect language". Below 
you refer to an 80% solution. An 80% solution to what? It is roughly a 0% 
solution to what I do professionally, but was a 100% solution to what I 
was doing for my postgrad project. If you fixed the things that stop me 
using it professionally [1] then you'd break the things that made it work 
on my postgrad. Inevitably. When you want to fix "mistakes", why do you 
think you get howls of protest? It's because for /some/ of us those are 
not mistakes at all, they're Just The Way It Should Be. The changes are 
not improving the language, they're just shuffling around the marketplace, 
and it's in serious danger of never staying in one market sector for long 
enough to take off as well as it should (because it really is a delight to 
work with).

You have lots of good ideas and suggestions from others. /Please/ don't 
change Python to accommodate them all; it would spoil it. Put them into a 
/new/ language. Maybe even build in call and data level compatibility so 
they can be well integrated. Then people can choose whichever one is best 
for the task. 

On the particular issue of case sensitivity, by the way, I'm in favour of 
an environment /option/ to warn if identifiers differ in case only. Not of 
a language change.

> > I've been there myself, on other projects... everything works 
> > correctly
> > and is almost perfect, but being a programmer you want to change
> > SOMETHING just to be fiddling with it.  "Almost perfect" is about the
> > best anyone in the Real World can hope for.
> 
> Sorry, that's not the affliction I am suffering from.  I know very
> well that Python will always be an 80% solution, and I'm happy to
> leave well enough alone.  There just are areas where I think future
> generations would thank me if I fixed it now...
> 
> > I've said it before, and I'll say it again... thanks with all my heart
> > for the best programming language I've ever worked in.
> 
> You're welcome.

Tim Rowe

[1] Strong typing, formal semantics, full control over memory management 
and deterministic timing for a start -- I work with safety-of-life 
systems!
More information about the Python-list mailing list