PEP-xxx: Unification of for statement and list-comp syntax

[Carl Banks]

Edward Elliott wrote:
> Special cases aren't special enough to break the rules. (proposal eliminates
> the current special case for comprehensions/generators)

It really isn't a special case, though.  It might seem like it is, but
it's not at all when you remember the rules of equivalence between
listcomps and regular statements.

This listcomp:

[ x for x in y ]

is, by definition, semantically equivalent to this (except it's an
expression instead of a statement):

for x in y:
    _.append(x)


If you have two fors in the listcomp, the second for is equivalent to a
nested for.  This:

[ x for y in z for x in y ]

is, by defintion, equivalent to this:

for y in z:
    for x in y:
        _.append(x)

Likewise, an if gets its own nested block in the equivalent statements.
 This:

[ x for x in y if x is not None ]

is, by definition, equivalent to this:

for x in y:
    if x is not None:
        _.append(x)

In other words, each for and if in a listcomp is equivalent to a nested
block in a regular statement.  There's a regular way to nest, and
listcomp way, and they're separate.  This PEP suggests we should mix
them up--now *that's* a special case.


Carl Banks