[IPython-dev] rant: "%matplotlib inline"

[Ondřej Čertík]

On Fri, Jul 3, 2015 at 11:53 PM, Fernando Perez <fperez.net at gmail.com> wrote:
> On Fri, Jul 3, 2015 at 9:15 PM, Matthias Bussonnier
> <bussonniermatthias at gmail.com> wrote:
>>
>>
>>
>> Well, that would be a step-back from what we have been doing for the last
>> 2 release to remove knowlege of IPython in Jupyter.
>> Or, actually break one abstraction that kernel do not know what have
>> started them.
>> If we do one I would prefer the more general one.
>
>
> No, it's not IPython-specific, it would simply be recording in any kernel
> how they were started. That information could be useful in a variety of
> contexts...
>
>>
>>
>>>
>>> The path forward is ti have fig object define a _repr_png_, the we don’t
>>> have to deal with %matpltlib inline
>>
>>
>> No, that wouldn't cover at all a case like the 'notebook' backend, that is
>> the preferred one for matplotlib, that allows panning/zooming (and hopefully
>> one day all MPL events in the notebook):
>>
>>
>> Well, also define _repr_html_ or _repr_js_ for notebook then.
>> The general message is to use our display_protocol the way it is design
>> for matplotlib too.
>>
>> Julia and R don’t need special activation and don’t have these issues, I
>> don’t see why we can’t figure
>> out how to do theses things.
>
>
> If we can find a cleaner solution with our display protocol, great!  Those
> ideas are certainly a lot more mature now than when we first hacked up the
> backend support in the early days of the qt console, and we also have a lot
> of real-world experience using it, teaching with it, etc.  It's a good time
> to take stock of that knowledge and update the toolchain...
>
> I see one or more PRs coming next week from Scipy :)
>
>> PS: Booooo Pylab :-)
>
>
> Actually, I use it a lot, and I think it's a mistake to berate it: %pylab is
> exactly the right thing to do when you need to quickly have something that
> puts at your fingertips a bunch of numerical tools.
>
> When I'm writing code that I *know* is purely numerical, with no chance
> whatsoever of a namespace conflict, what exactly do I gain by having to
> write
>
> import numpy as np
> x = np.sin(np.pi*np.e)+np.log(z)
>
> instead of the vastly more readable, debuggable, comprehensible
>
>  x = sin(pi*e) + log(z)
>
> ???  (and yes, I could do a lot of explicit single-function imports, but
> that gets old in a hurry when writing a lot of numerical code).
>
> For essentially one-off, matlab-style notebooks, I think that we need to be
> pragmatic and remember that ease of use, readability, ease of remembering
> what to write, are important.  These are cases where people are not building
> large software engineering projects, and where getting something done *now*
> is what matters, so let's not get too dogmatic in berating them for using
> tools that actually help them accomplish something.
>
> So yes, I still use, and will continue to use %pylab quite a bit...
>
> Remember, readability counts and practicality beats purity... Don't forget
> to import this every now and then, it's always a good antidote against
> dogmatism.

I just want to second William's request, I would love to just be able
to import matplotlib, and perhaps if I do show(), it would actually
show the plot, instead of silently fail, which is the worst possible
outcome.

As to pylab, I use it as well, as the only thing in fact. I start with
an empty notebook, do "%pylab inline" and quickly load some data and
plot it. Then I improve the plots and then when things work, I
actually keep it like that, as it seems the time needed to change the
imports and code completely to explicit imports is not justified.

Ondrej