Python Rocks!

Tim Peters tim_one at email.msn.com
Mon Jan 24 19:30:40 EST 2000


[Paul Prescod]
> Wouldn't it be easier if we just added an optional, (but
> compiler-recognized if present) block delimiter to Python 2?

The option was offered to Guido years ago, via a patch by Donald Beaudry to
do just that.  He declined then, and I doubt he'd change his mind now.

> Then people would have to stop complaining that Python is
> broken

No, they wouldn't.  That's the whole problem here:  people with a bug up
their butt about indentation are no different than people with a bug up
their butt about integer division (like me!), or people who hate reference
counting, or those who hate the lack of an address-of operator, or lack of
lexical closures, or lack of an ISO std, or ... I once drew up a list of
about 30 of these "crucial show-stoppers" culled from semi-hysterical c.l.py
threads.

When people say that Python's indentation drives away thousands of
programmers, they're almost certainly correct!  And so do each of dozens of
other things.  The same is also true of every other language on earth.
Guido is on the receiving end of *all* these gripes, so has a sobering
global perspective the proponents of individual features X, Y and Z lack in
their single-minded zeal.  Over the years, and only after a great deal of
thought and care, he occasionally "yields" on the few he judges most likely
to add more power than cause harm.

I don't know whether he keeps a formal priorty list of these things, but, if
he does, optional block delimiters must be scraping the bottom of it (adds
no power, at best leads to the usual flamewars about where the delimiters
"should be" placed, at worst ruins the easy visual uniformity of Python code
that so many old-timers (incl. esp. Guido!) actively love).

> and instead admit that Python programmers choose to
> program without delimiters.

The existence and utter non-use of Guido's Tools/Scripts/pindent.py is
sufficient proof of that.

that-was-indeed-his-last-word-on-the-subject-ly y'rs  - tim






More information about the Python-list mailing list