[IPython-dev] Regression bug in 0.8.x input filtering

[Dan Milstein]

Okay, I think I've fixed this (just committed it to svn).

Now, as per the requests below, any magics, aliases or autocalls must  
be separated from the 'rest' of the line by a whitespace for ipython  
to treat them special.

Thus, e.g.

  $ r'a_string'

Will no longer trigger the %r magic even if automagic is on.  Added  
tests to verify this.

I'm hopeful that this isn't going to screw anything else up, but  
everything prefiltering related is complex, so lemme know if it blows  
things up interestingly.

-D


On May 24, 2007, at 10:50 AM, Hans Meine wrote:

> On Donnerstag 24 Mai 2007, Dan Milstein wrote:
>> Okay, I'm sold on the %r as a reasonable magic, and whitespace
>> splitting as a reasonable req.
> I too think that whitespace splitting is reasonable, but I would  
> never use %r.
> OTOH, I see that every ipython user has his/her own habits.
>
>> Specifically, we jump through some hoops right now so that:
>>> !thing?
>>
>> and
>>
>>> !!thing?
>>
>> Trigger the shell handler, and *not* the help handler.  This was
>> particularly important/useful because the qmark is a part of some
>> valid shell commands.
> Part of globs in particular.  globs are not evaluated by the shell  
> in the
> first (command) word, so I would suggest..
>> maybe it's okay if !thing? triggers help, as long as:
>>> !thing (.*)?
>>
>> and
>>
>>> !!thing (.*)?
> exactly this.
>
> -- 
> Ciao, /  /                                                    .o.
>      /--/                                                     ..o
>     /  / ANS                                                  ooo
> _______________________________________________
> IPython-dev mailing list
> IPython-dev at scipy.org
> http://lists.ipython.scipy.org/mailman/listinfo/ipython-dev