[IPython-user] 0.4.0 ready for Monday

[Prabhu Ramachandran]

>>>>> "FP" == Fernando Perez <fperez at colorado.edu> writes:

    FP> Prabhu Ramachandran wrote:
    >>>>>>> "GB" == Gary Bishop <gb at cs.unc.edu> writes:
    >>
    GB> Python 2.2.2 (#37, Oct 14 2002, 17:02:34) [MSC 32 bit (Intel)]
    GB> IPython 0.2.15pre5 -- An enhanced Interactive Python.
    >>
    GB> This is from CVS
    >>  When I get the time I'll try and check if the CVS version
    >> fixes this.

    FP> I doubt it does, as the problem is that the cvar objects
    FP> generate a funny NameError exception which trhows the standard
    FP> lib's inspect.py for a loop.  Here's the relevant part of the
    FP> traceback from ipython's crash report:

[snip]

    FP> As I said, implementing a workaround in ipython is trivial,
    FP> and I can do it.  But I do think that the bug lies really in
    FP> the way the SWIG code works.  I can see why they chose to
    FP> raise a NameError, since it seems natural for missing variable
    FP> names.  But since they are implementing name access via
    FP> attributes, the proper exception is an AttributeError.
    FP> Otherwise they end up breaking even code in the python
    FP> standard library, such as in the inspect.py case above.

Indeed but what stumps me is that this same thing works with
Python2.2.2 for Gary.  I looked at the version of inspect.py in CVS
and the getdoc still only looks for an attribute error.  Perhaps
something deeper is happening?  This is not a serious issue so I'm not
really worried too much about it.  Thanks.

cheers,
prabhu