[IPython-dev] use of InteractiveShell and R magic

[MinRK]

On Thu, Jun 14, 2012 at 4:30 PM, Brian Granger <ellisonbg at gmail.com> wrote:

> But don't we have similar assumptions about InteractiveShell being a
> singleton throughout the code base.  The patterns used by the magic
> system are by no means unique.  I am not sure I would call this a bug.
>

The relevant assumption is that the Magics object itself is a singleton.
 The magics table is *per class*, but on initialization of the first
instance, the class mapping clobbered with the methods of the first
instance.  Simply copying the magics dict before altering it so that it is
an instance attribute rather than a class attribute edited by the instance
fixes the issue. I think this is inappropriate for a class that doesn't
have any singleton-like instance() api like the InteractiveShell does.

For instance:

from IPython.core.magic import magics_class, Magics, line_magic
@magics_class
class MyMagic(Magics):

    @line_magic
    def foo(self, line):
        print 'hi'

magic = MyMagic(None)
magic2 = MyMagic(None)

assert magic2.magics['line']['foo'].im_self is not magic # fails


>
> Brian
>
> On Thu, Jun 14, 2012 at 4:26 PM, MinRK <benjaminrk at gmail.com> wrote:
> >
> >
> > On Thu, Jun 14, 2012 at 3:57 PM, Jonathan Taylor
> > <jonathan.taylor at stanford.edu> wrote:
> >>
> >> Thanks, that seems to work in a separate python process rather than the
> >> ipython process:
> >>
> >> ~/workbook (metadata*) $ python -c "import
> >> IPython.frontend.terminal.interactiveshell as IS; shell =
> >> IS.TerminalInteractiveShell(); shell.run_cell('%load_ext rmagic');
> >> shell.run_cell('%R X=rnorm(20); print(X)')"
> >>  [1]  0.2125483  0.1030451 -1.4036373 -0.5617847 -0.7571675 -0.3985085
> >>  [7] -0.4386932  0.7164915 -0.4794236 -0.1373804  0.2831013 -0.7248862
> >> [13]  0.5260675 -0.3789804  0.1535592  1.5569203  1.0225972  0.8149846
> >> [19] -0.9629060 -0.4161373
> >>
> >> ~/workbook (metadata*) $ python -c "import
> >> IPython.frontend.terminal.interactiveshell as IS; shell =
> >> IS.InteractiveShell(); shell.run_cell('%load_ext rmagic');
> >> shell.run_cell('%R X=rnorm(20); print(X)')"
> >>  [1]  0.44037484  1.31372601 -0.37253955  2.21955270 -0.07886852
> >> -0.09022193
> >>  [7]  0.38464850 -0.41004855  1.62648041 -0.13687832  0.91537699
> >> -2.70125217
> >> [13]  1.49125226  0.11580012 -0.64993087  2.49367657  0.01447792
> >>  0.57408249
> >> [19] -0.80559871 -0.76346964
> >>
> >> So, for interactively running this in ipython I should expect to see
> this
> >> load_ext failure?
> >>
> >
> > You are running into some singleton assumptions in the new magics system
> -
> > each Magics class methods are only registered once, and thus bound to the
> > first Shell that load the magic.
> >
> > So what you are seeing is that shell.magic("load_ext rmagic") is actually
> > triggering the load_ext magic on the *original* IPython instance.  I
> think
> > this should be considered a bug.
> >
> >>
> >> Thanks,
> >>
> >> Jonathan
> >>
> >> On Thu, Jun 14, 2012 at 3:48 PM, Thomas Kluyver <takowl at gmail.com>
> wrote:
> >>>
> >>> On 14 June 2012 23:44, Jonathan Taylor <jonathan.taylor at stanford.edu>
> >>> wrote:
> >>> > Actually, I want to create an InteractiveShell inside another python
> >>> > process. Maybe I should use embed?
> >>>
> >>> Embed's not terribly relevant here - that creates an InteractiveShell,
> >>> but attempts to use the namespace from which it's called.
> >>>
> >>> Try it in a separate Python process. Don't forget to %load_ext rmagic
> >>> first.
> >>>
> >>> Thomas
> >>> _______________________________________________
> >>> IPython-dev mailing list
> >>> IPython-dev at scipy.org
> >>> http://mail.scipy.org/mailman/listinfo/ipython-dev
> >>
> >>
> >>
> >>
> >> --
> >> Jonathan Taylor
> >> Dept. of Statistics
> >> Sequoia Hall, 137
> >> 390 Serra Mall
> >> Stanford, CA 94305
> >> Tel:   650.723.9230
> >> Fax:   650.725.8977
> >> Web: http://www-stat.stanford.edu/~jtaylo
> >>
> >> _______________________________________________
> >> 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
> >
>
>
>
> --
> Brian E. Granger
> Cal Poly State University, San Luis Obispo
> bgranger at calpoly.edu and ellisonbg at gmail.com
> _______________________________________________
> IPython-dev mailing list
> IPython-dev at scipy.org
> http://mail.scipy.org/mailman/listinfo/ipython-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/ipython-dev/attachments/20120614/2596b181/attachment.html>