ipython configuration and isympy

[krastano...@gmail.com]

Sometime ago I mentioned that for the correct working of the new
plotting module (actually only for the matplotlib backend) one must do
some magic with the event loops. It seems that this magic was so hard
that the matplotlib documentation simply states that if you want
interactive non-blocking plotting you _must_ use ipython as the
ipython team has taken care of all this. So, now the implementation of
the new plotting module requires a small one line change to the sympy
profile in ipython.

I will leave the discussion surrounding all the event loop magic and
plotting for another time and focus on the problem of updating the
sympy profile in ipython.

The fact is that one needs to add different new configuration options
to this profile as sympy evolves and it is suboptimal to require an
upstream change in ipython for every such change.

To remedy this problem Aaron proposed some time ago to dump all such
configurations in the interactive module in sympy and just make the
ipython's sympy profile call them.

Could we proceed with this idea?

[Aaron Meurer]

Yes, let's do it. Will you be able to do it? The relevant SymPy
issues are 3031, 2632, and 2945.

Aaron Meurer

[krastano...@gmail.com]

>
> Yes, let's do it.  Will you be able to do it?  The relevant SymPy
> issues are 3031, 2632, and 2945.
>

Well my mail was more of a plead "please somebody do it" ;)

However as it is a blocker on one of my pull request I will actually
try to do it now that I have your approval.
(in a week as I have many exams at the moment)

[Aaron Meurer]

On Fri, Apr 27, 2012 at 4:11 PM, krastano...@gmail.com
<krastano...@gmail.com> wrote:
>>
>> Yes, let's do it.  Will you be able to do it?  The relevant SymPy
>> issues are 3031, 2632, and 2945.
>>
>  Well my mail was more of a plead "please somebody do it" ;)

Well I won't be free to do it for at least two weeks, when I graduate.
But maybe someone else will step up.

>
> However as it is a blocker on one of my pull request I will actually
> try to do it now that I have your approval.
> (in a week as I have many exams at the moment)

If my approval is all it takes, then I hereby approve everybody to
implement any idea or fix any bug in SymPy. :)

Aaron Meurer

[Joachim Durchholz]

Am 28.04.2012 00:33, schrieb Aaron Meurer:
>> However as it is a blocker on one of my pull request I will actually
>> try to do it now that I have your approval.
>> (in a week as I have many exams at the moment)
>
> If my approval is all it takes, then I hereby approve everybody to
> implement any idea or fix any bug in SymPy. :)

I think I'll at least participate once I get that lambdify-raises issue
off my desk (and my mind). I was heading towards doing more review work,
but I guess that can help with the plotting and interactive stuff, too
(besides, I do have some solid GUI architecture background, which may or
may not help here).

Which means that I'm blocked for now. I understand that graduation has
some relevance, too ;-)
I'll just wait until that happens.
And good luck to all who're about to do their exams!

[Bharath M R]

I have been trying to get this working.

Should we also bring the sympyprinting extension in IPython to sympy or leave it there?
For additional configuration, we can have an ipython_config in interactive and call it when
we run ipython with sympy.

[krastano...@gmail.com]

I would prefer that everything is kept inside sympy and not inside
ipython. However you may want to ask Brian Granger what makes most
sense (he is among the lead developers of ipython, I have cc-ed him).
The only thing that I know is that Aaron already expressed the desire
to have all in sympy in order to be able to make changes without
pushing upstream to ipython.

[Brian Granger]

I am fine if sympyprinting moves to sympy proper. IPython's extension
loading mechanism works fine with anything in sys.path so we could do:

%load_ext sympy.sympyprinting

Or something like this. It should also remain in IPython though for
the transition period.

Cheers,

Brian

--
Brian E. Granger
Cal Poly State University, San Luis Obispo
bgra...@calpoly.edu and elli...@gmail.com

[Aaron Meurer]

For what it's worth, Min (two other core IPython devs) also agrees.
See https://github.com/ipython/ipython/pull/1773#issuecomment-5996649.
And given the fact that the only IPython developer that knows the
SymPy internals pretty well is Brian, I think the extension will see
much more love inside of SymPy.

Aaron Meurer

> --
> You received this message because you are subscribed to the Google Groups "sympy" group.
> To post to this group, send email to sy...@googlegroups.com.
> To unsubscribe from this group, send email to sympy+un...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/sympy?hl=en.
>

[Bharath M R]

ok. I will move the extension to the interactive module.