[IPython-dev] Argument parsing (was: Qt api selection re. ipython and matplotlib)

[MinRK]

On Thu, Jul 7, 2011 at 20:42, Ilan Schnell <ischnell at enthought.com> wrote:
>> only because 0.11.rc1 got into EPD
>
> I hope this is not the only reason.  We will probably do an EPD
> bug fix release in a few weeks, in which I'm planing to update
> to 0.11 final.

That helps a lot, thanks for reminding us, Ilan.  Now that I read
that, I'm pretty sure you told us this earlier, but my brain is pretty
fried right now.  This takes some pressure off, and we can make sure
we get it right, without worrying about 6 months of EPD users having
an IPython that doesn't match those before or after.

We will favor natural argparse-style arguments, even if we don't
finish migrating to argparse itself, and the vociferously opposed
'foo=value' without leading hyphens will likely be removed. master is
currently in a temporary state that supports both.

-MinRK

>
> - Ilan
>
>
> On Thu, Jul 7, 2011 at 6:41 PM, MinRK <benjaminrk at gmail.com> wrote:
>> We have relaxed the syntax in 0.11, so you can do the more familiar
>> '--gui=qt' pattern, and 'ipython -i foo.py' should work as expected.
>> The 'gui=qt' way will still work in 0.11, only because 0.11.rc1 got
>> into EPD, but I think we will remove it in 0.12.  I also added support
>> for '--' to halt parsing, as Robert described.
>>
>> Since we aren't doing unusual things for the simple cases, I opened an
>> issue to replace the existing machinery for everything short of fully
>> specified Class.trait=value with argparse, and only using our extra
>> code for that single case, since it just makes more sense. I don't
>> believe this will happen in 0.11.
>>
>> @Julian '-i' should work fine now (its alias was --i, since we used to
>> require two leading '-' for flags, as described in the help output).
>> We don't (and won't) support 'ipython -c "stuff"' until we go back to
>> using argparse. Its alias is 'c' (used as 'c=stuff' or, with current
>> master, '--c=stuff'), as described in the help output, until then.  I
>> also believe I fixed the inherited-argv bug for scripts that you
>> described.
>>
>> -MinRK
>>
>> On Thu, Jul 7, 2011 at 09:55, Robert Kern <robert.kern at gmail.com> wrote:
>>> On 7/7/11 11:32 AM, MinRK wrote:
>>>> On Thu, Jul 7, 2011 at 07:00, Darren Dale<dsdale24 at gmail.com>  wrote:
>>>
>>>>> Or, suppose I have a file named something pathological like foo=bar?
>>>>> Python will run it without any problems. Even if I rename it
>>>>> foo=bar.py, ipython appears to interpret it as an option, not a
>>>>> filename.
>>>>
>>>> You can always give more path information, so `ipython ./profile` or
>>>> `ipython ./foo=bar` will work*.  Python has the exact same
>>>> vulnerability to pathological filenames matching their argument
>>>> patterns, like files called '-foo' or '-zebulon'. Obviously, given our
>>>> more flexible syntax, our cases cast a wider net.
>>>
>>> Well, with argparse, you can use the standardized "--" marker to indicate that
>>> everything following is an argument, not an option. E.g.
>>>
>>>   $ ipython -pylab -- -foo.py
>>>
>>> --
>>> Robert Kern
>>>
>>> "I have come to believe that the whole world is an enigma, a harmless enigma
>>>  that is made terrible by our own mad attempt to interpret it as though it had
>>>  an underlying truth."
>>>   -- Umberto Eco
>>>
>>> _______________________________________________
>>> IPython-dev mailing list
>>> IPython-dev at scipy.org
>>> http://mail.scipy.org/mailman/listinfo/ipython-dev
>>>
>> _______________________________________________
>> IPython-dev mailing list
>> IPython-dev at scipy.org
>> http://mail.scipy.org/mailman/listinfo/ipython-dev
>>
>