[Python-3000] Removing repr

[Jim Jewett]

On 4/6/06, John J Lee <jjl at pobox.com> wrote:

[suggests changing the docs for repr for Py3k]

The current help(repr) may be even more misleading.

I would suggest making the change as early as 2.4 (2.5 for the docstring.

Changes:

(1)  It doesn't mention backquotes.  They are already deprecated, so
it makes sense to mention the equivalency in the backquote docs,
rather than the repr docs.

(2)  Saying "for debugging" instead of "canonical" makes it less of an
attractive nuisance.  (canonical can mean "the one right way")

(3)  Giving the < > example along with -- or even before -- the eval
makes it clear that you shouldn't rely on the eval loop in the general
case.

These are good changes, even from a 2.x perspective.

As to the exact wording -- what are the current guarantees on the type
of repr's return.  Can it be unicode?  If we don't want to explicitly
ban unicode, I would modify your suggestion (even today) to

"""repr(object) -> string

Return a printable representation of an object, useful for debugging.

For simple objects, repr may the inverse of eval, so that
eval(repr(obj)) == obj.

More complex objects typically use '<...useful info ...>' such as
'<__main__.C instance at 0x00B818C8>'.
"""


-jJ