[PEAK] Strange dispatch problem after cvs up

[Jay Parlar]

I give +1 to a Python2.4 requirement on the PEAK package, as soon as I
get my sysadmin to upgrade one of machines. It was just easier for now
to use the 2.3 decorators and not bug him, but I'll get that done on
Monday.

Thanks for the quick fixes, I'll try them out on Monday. I've got some
fairly elaborate doctests for the part of my program that uses the
dispatch module, so I'll let you know if they find any other issues.

Jay P.

On 4/24/05, Phillip J. Eby <pje at telecommunity.com> wrote:
> At 09:02 AM 4/21/05 -0400, Jay Parlar wrote:
> >Fantastic, importing dispatch.strategy first worked for me. Not a good
> >long term solution, but at least I can get back to it.
> 
> Okay, I've checked in a fix.  It turns out it wasn't an import circularity,
> but rather it's the fact that you can't nest "assignment advisors" -- my
> evil trace-hook hack to simulate decorators in Python <2.4.  What was
> happening is that if the first time dispatch.strategy was imported was from
> a decorator (like 'someGF.when()'), then an important 'dispatch.as()' call
> would silently fail, mucking up the "make_index" method's setup as a
> classmethod.
> 
> Anyway, it's fixed now, but we should probably make a move to Python 2.4 in
> the near future in order to get away from this problem.
> 
>