Issue1510227

I'm seeing weird differences between repr() and
__repr__() that don't seem to appear in regular Python.

For example:

>>> foo='foo\b\n'
>>> repr(foo)
"'foo\\b\\n'"
>>> foo.__repr__()
'foo\b\n'

This is especially bad from Java, as the __repr__() is
all you have access to.  To reproduce from Java, I put
code in _codecs.java, in a new escape_encode function;
I'll submit this as part of my first pickle patch.

Matt, I thought you'd applied your patch, but this still seems to be broken in trunk and on the 2.3 branch.  Did this not end up in the patch?

Im going to take a wild guess here that the problem is because
repr(foo)

will go to __builtin__ repr that calls:


public PyString __repr__() {

    return new PyString(encode_UnicodeEscape(string, true));

}

while
foo.__repr__

calls
    final String str_toString() {

        return string;

    }

Incidentally, python returns:
"'foo\\x08\\n'"
for both repr and __repr__

leouser

this is fixed on trunk as of r4346