[IPython-user] Fwd: Preparing for the 0.7.4 release

[Fernando Perez]

On 3/23/07, Stefan van der Walt <stefan@sun.ac.za> wrote:

> I don't suggest you apply the attached patch, but it serves to
> demonstrate the workaround.

Thanks, Stefan.  I did something slightly different, but r2195 works
for me with Jorgen's original problematic example.  I think the point
is to ensure that the various points in the code that produce
variables susceptible of being treated as a Python identifier do so
via

.encode('ascii')

since pyhton itself mandates pure ascii for identifiers.  I'm trying
to enforce that rule strictly now, though I may still have missed a
few locations.

I didn't modify _ofind because I expect that it should always be
called with valid ascii, so it shouldn't be protecting itself.  We
should rather ensure that we don't feed it unicode, there's only a few
points where data flows into it and those can be well controlled.
Since the bulk of those start with input from split_user_input(),
that's what I fixed.

Feel free to continue sumbitting problem reports...

Cheers,

f