Early binding as an option
tjreedy at udel.edu
Tue Aug 2 22:23:21 CEST 2011
On 8/2/2011 12:55 PM, Chris Angelico wrote:
> As I understand it, Python exclusively late-binds names; when you
> define a function, nothing is ever pre-bound.
By 'pre-bound' you presumably mean bound at definition time rather than
call time. Default arg objects *are* pre-computed and pre-bound to
internal slots at definition time.
> Argument in favour: Efficiency is also a Good Thing, and with staples
> like 'len', it's unlikely anyone will want to change them - yet the
> interpreter still has to do a namespace lookup every time.
Three approaches to machine efficiency.
1. Better algorithm: Python's people efficiency makes this easier than
in most other languages.
2. Hand-optimize the code that actually chew up time (as revealed by
profiler). This often means removing repeated expressions *and* global
names from inner loops.
_len = len
for line in somefile:
n = _len(line)
*might* give a worthwhile speedup in a function if not too much else
happends in the loop. But the CPython global name lookup code (in C) has
been thoroughly examined and optimized as best as several developers
could think of.
3. Convert critical code to native language (or C).
The idea of 'early binding' comes up periodically but I do not remember
many concrete proposals.
Terry Jan Reedy
More information about the Python-list