Thread: [matplotlib-devel] pyplot commands appear to ignore interactive status

using a recent svn (r8900), I've noticed that after starting from a 
regular python shell:

 >>> import matplotlib.pyplot as plt
 >>> plt.isinteractive()
False
 >>> plt.plot([1,2,3],[1,3,2])
[<matplotlib.lines.Line2D object at 0x114e7a090>]
 >>> plt.plot([1,2,3],[1,2,3])
[<matplotlib.lines.Line2D object at 0x114e7a590>]
# plt.draw() is not required, the figure pops up
# and both plots are shown
 >>> plt.xlim(1,2)
(1, 2)
# again this works immediately no draw() required
 >>> plt.xlabel('aaa')
<matplotlib.text.Text object at 0x1033a2290>
# ditto, no draw() required

but if the axes methods are used, then interactive status is honored:

 >>> plt.gca().set_xlabel('bbb')
<matplotlib.text.Text object at 0x1033a2290>
 >>> plt.gca().plot([1,2,3],[3,2,1])
[<matplotlib.lines.Line2D object at 0x114e7cf90>]
 >>> plt.gca().set_xlim(1,3)
(1, 3)
# all these require a plt.draw() to show up...

I think that this is a misfeature, but maybe this is desired behavior?

M

On Mon, Feb 7, 2011 at 7:20 PM, Mike Kaufman <mc...@gm...> wrote:

>
> using a recent svn (r8900), I've noticed that after starting from a
> regular python shell:
>
>  >>> import matplotlib.pyplot as plt
>  >>> plt.isinteractive()
> False
>  >>> plt.plot([1,2,3],[1,3,2])
> [<matplotlib.lines.Line2D object at 0x114e7a090>]
>  >>> plt.plot([1,2,3],[1,2,3])
> [<matplotlib.lines.Line2D object at 0x114e7a590>]
> # plt.draw() is not required, the figure pops up
> # and both plots are shown
>  >>> plt.xlim(1,2)
> (1, 2)
> # again this works immediately no draw() required
>  >>> plt.xlabel('aaa')
> <matplotlib.text.Text object at 0x1033a2290>
> # ditto, no draw() required
>
> but if the axes methods are used, then interactive status is honored:
>
>  >>> plt.gca().set_xlabel('bbb')
> <matplotlib.text.Text object at 0x1033a2290>
>  >>> plt.gca().plot([1,2,3],[3,2,1])
> [<matplotlib.lines.Line2D object at 0x114e7cf90>]
>  >>> plt.gca().set_xlim(1,3)
> (1, 3)
> # all these require a plt.draw() to show up...
>
> I think that this is a misfeature, but maybe this is desired behavior?
>
> M
>
>
Which backend are you using and using which OS?

Ben Root

On 2/7/11 9:02 PM, Benjamin Root wrote:
> On Mon, Feb 7, 2011 at 7:20 PM, Mike Kaufman <mc...@gm...
> <mailto:mc...@gm...>> wrote:
>
>
>     using a recent svn (r8900), I've noticed that after starting from a
>     regular python shell:
>
>      >>> import matplotlib.pyplot as plt
>      >>> plt.isinteractive()
>     False
>      >>> plt.plot([1,2,3],[1,3,2])
>     [<matplotlib.lines.Line2D object at 0x114e7a090>]
>      >>> plt.plot([1,2,3],[1,2,3])
>     [<matplotlib.lines.Line2D object at 0x114e7a590>]
>     # plt.draw() is not required, the figure pops up
>     # and both plots are shown
>      >>> plt.xlim(1,2)
>     (1, 2)
>     # again this works immediately no draw() required
>      >>> plt.xlabel('aaa')
>     <matplotlib.text.Text object at 0x1033a2290>
>     # ditto, no draw() required
>
>     but if the axes methods are used, then interactive status is honored:
>
>      >>> plt.gca().set_xlabel('bbb')
>     <matplotlib.text.Text object at 0x1033a2290>
>      >>> plt.gca().plot([1,2,3],[3,2,1])
>     [<matplotlib.lines.Line2D object at 0x114e7cf90>]
>      >>> plt.gca().set_xlim(1,3)
>     (1, 3)
>     # all these require a plt.draw() to show up...
>
>     I think that this is a misfeature, but maybe this is desired behavior?
>
>     M
>
>
> Which backend are you using and using which OS?

Good question. Snow Leopard and the MacOSX backend. If I use the Gtk 
backend this bug does _not_ occur (though I have to use the plt.show() 
command to bring up the window --- which hangs the shell...)

M

On 02/07/2011 04:13 PM, Mike Kaufman wrote:
> On 2/7/11 9:02 PM, Benjamin Root wrote:
>> On Mon, Feb 7, 2011 at 7:20 PM, Mike Kaufman<mc...@gm...
>> <mailto:mc...@gm...>>  wrote:
>>
>>
>>      using a recent svn (r8900), I've noticed that after starting from a
>>      regular python shell:
>>
>>       >>>  import matplotlib.pyplot as plt
>>       >>>  plt.isinteractive()
>>      False
>>       >>>  plt.plot([1,2,3],[1,3,2])
>>      [<matplotlib.lines.Line2D object at 0x114e7a090>]
>>       >>>  plt.plot([1,2,3],[1,2,3])
>>      [<matplotlib.lines.Line2D object at 0x114e7a590>]
>>      # plt.draw() is not required, the figure pops up
>>      # and both plots are shown
>>       >>>  plt.xlim(1,2)
>>      (1, 2)
>>      # again this works immediately no draw() required
>>       >>>  plt.xlabel('aaa')
>>      <matplotlib.text.Text object at 0x1033a2290>
>>      # ditto, no draw() required
>>
>>      but if the axes methods are used, then interactive status is honored:
>>
>>       >>>  plt.gca().set_xlabel('bbb')
>>      <matplotlib.text.Text object at 0x1033a2290>
>>       >>>  plt.gca().plot([1,2,3],[3,2,1])
>>      [<matplotlib.lines.Line2D object at 0x114e7cf90>]
>>       >>>  plt.gca().set_xlim(1,3)
>>      (1, 3)
>>      # all these require a plt.draw() to show up...
>>
>>      I think that this is a misfeature, but maybe this is desired behavior?
>>
>>      M
>>
>>
>> Which backend are you using and using which OS?
>
> Good question. Snow Leopard and the MacOSX backend. If I use the Gtk

This is a known bug in the MacOSX backend.

> backend this bug does _not_ occur (though I have to use the plt.show()
> command to bring up the window --- which hangs the shell...)

It should simply block until the window is closed; is this what you mean?

Eric

>
> M
>
> ------------------------------------------------------------------------------
> The ultimate all-in-one performance toolkit: Intel(R) Parallel Studio XE:
> Pinpoint memory and threading errors before they happen.
> Find and fix more than 250 security defects in the development cycle.
> Locate bottlenecks in serial and parallel code that limit performance.
> http://p.sf.net/sfu/intel-dev2devfeb
> _______________________________________________
> Matplotlib-devel mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel

On 2/7/11 9:28 PM, Eric Firing wrote:

>>> Which backend are you using and using which OS?
>>
>> Good question. Snow Leopard and the MacOSX backend. If I use the Gtk
>
> This is a known bug in the MacOSX backend.
>
>> backend this bug does _not_ occur (though I have to use the plt.show()
>> command to bring up the window --- which hangs the shell...)
>
> It should simply block until the window is closed; is this what you mean?

Yes I do, although I'm not sure why it should block: I may want to add 
additional plots to the window --- though I do notice that if 
isinteractive=True, then the window doesn't block. It makes sense I 
guess, but not really intuitive, especially not coming from the 
behaviour of the OSX backend.

M

On Mon, Feb 7, 2011 at 9:17 PM, Mike Kaufman <mc...@gm...> wrote:

> On 2/7/11 9:28 PM, Eric Firing wrote:
>
> >>> Which backend are you using and using which OS?
> >>
> >> Good question. Snow Leopard and the MacOSX backend. If I use the Gtk
> >
> > This is a known bug in the MacOSX backend.
> >
> >> backend this bug does _not_ occur (though I have to use the plt.show()
> >> command to bring up the window --- which hangs the shell...)
> >
> > It should simply block until the window is closed; is this what you mean?
>
> Yes I do, although I'm not sure why it should block: I may want to add
> additional plots to the window --- though I do notice that if
> isinteractive=True, then the window doesn't block. It makes sense I
> guess, but not really intuitive, especially not coming from the
> behaviour of the OSX backend.
>
> M
>
>
The behavior in gtk and other backends is the designed/intended behavior.
macosx backend is actually the odd-man out because it was coded to only work
in one of those modes.

Maybe we should emit a warning when macosx backend is used when
interactive=False in order to dispel misunderstanding?

Ben Root

On 2/7/11 10:30 PM, Benjamin Root wrote:
>
> The behavior in gtk and other backends is the designed/intended
> behavior.  macosx backend is actually the odd-man out because it was
> coded to only work in one of those modes.
>
> Maybe we should emit a warning when macosx backend is used when
> interactive=False in order to dispel misunderstanding?

possibly, but maybe adding a blurb about OSX, and maybe the blocking 
behavior to the documentation here: 
http://matplotlib.sourceforge.net/users/shell.html and maybe to the 
dostrings of ion(), ioff() would be good.

M