[Python-Dev] the "3*x works w/o __rmul__" bug

[Alex Martelli]

On Monday 27 October 2003 04:22, Greg Ewing wrote:
> Alex Martelli <aleaxit at yahoo.com>:
> > Exactly the same underlying reason as a bug I just opened on
> > SF: if x is an instance of a class X having __mul__ but not
> > __rmul__, 3*x works (just like x*3) but 3.0*x raises TypeError
>
> Seems to me the bug there is not giving X an __rmul__
> method and yet expecting y*x to work at all. The fact
> that it happens to work in some cases is an accident
> that should not be relied upon.

No, the bug is that it works in some cases where it should fail
(and, secondarily, that -- where it does fail -- it gives a weird error
message).  In other words, the bug (in Python) is that "accident".

Nobody's asking for 3.0*x to work where x is a user-coded type
without an __rmul__; rather, the point is that 3*x should fail too,
and ideally they'd have the same clear error message as 3+x
gives when the type has no __radd__.

Doesn't seem trivial to fix (though I hope I'm missing something
obvious) and doesn't affect perfect user-programs, but I do
think it should be fixed because it's sure extremely mysterious
and could send a developer on wild goose chases when met in
the course of development.


Alex