Fwd: Overriding logical operators?

[Andrew Durdin]

On Fri, 20 Aug 2004 23:28:20 -0700, Robert Brewer <fu******@amor.org> wrote:

The first barrier is that 'or' and 'and' get compiled down to jump
codes, as opposed to operations:
[snip disassembly>]
....so you'd be hacking a completely different section of the compiler to
enable such overriding.

I wasn't yet planning to hack *any* part of the compiler :)

As far as I can see, the compiler changes would be limited to
compiling "a and b" to bytecode equivalent to

a.__logand__(lambda b: b)

And bool.__logand__ would work like this:

def bool.__logand__(self, otherlambda):
if not self:
return False
else:
return bool(otherlambda())

I believe this would handle the short-circuiting. How much
implementation this would require is something I don't know the answer
to yet... I've only just started looking at the python source. At the
least it'd mean compiler changes, but probably also changes to a
number of built-in objects.

[Greg Ewing]

Andrew Durdin wrote:

As far as I can see, the compiler changes would be limited to
compiling "a and b" to bytecode equivalent to

a.__logand__(lambda b: b)

That approach would work, but it would slow down the
evaluation of all logical expressions a lot, requiring
an extra Python function call to get the second operand,
even in the common case where no overriding is going on.
So I doubt it would be accepted.

--
Greg Ewing, Computer Science Dept,
University of Canterbury,
Christchurch, New Zealand
http://www.cosc.canterbury.ac.nz/~greg