matplotlib-users

[Matplotlib-users] plotting from xemacs always 1 step behind

[Neal B. <ndb...@gm...>]

This is a bit wierd.

If running from xemacs, it seems interactive plotting is always 1 step
behind.

For example, I do:

M-x py-shell
This brings up a *Python* buffer in Comint mode running Ipython.

In [1]: from pylab import *

In [2]: plot ([1,2,3])
Out[2]: [<matplotlib.lines.Line2D instance at 0xf563b0>]

In [3]: xlabel('time')
Out[3]: <matplotlib.text.Text instance at 0xf52758>

In [4]: ylabel ('volts')
Out[4]: <matplotlib.text.Text instance at 0xf533b0>

When I execute step 3, the time is not show, but the line is.  When I do
step 4, now time is added.

Any clues?

Re: [Matplotlib-users] plotting from xemacs always 1 step behind

[John H. <jd...@gm...>]

On Feb 12, 2008 6:49 AM, Neal Becker <ndb...@gm...> wrote:
> This is a bit wierd.
>
> If running from xemacs, it seems interactive plotting is always 1 step
> behind.

I assume you have set 'interactive : True' in your rc file?  What
backend are you using (you should be using tkagg from xemacs) since it
does not require GUI threading support.

  http://matplotlib.sf.net/interactive.html

JDH

Re: [Matplotlib-users] plotting from xemacs always 1 step behind

[Neal B. <ndb...@gm...>]

On Tuesday 12 February 2008, John Hunter wrote:
> On Feb 12, 2008 6:49 AM, Neal Becker <ndb...@gm...> wrote:
> > This is a bit wierd.
> >
> > If running from xemacs, it seems interactive plotting is always 1 step
> > behind.
>
> I assume you have set 'interactive : True' in your rc file?  What
> backend are you using (you should be using tkagg from xemacs) since it
> does not require GUI threading support.
>
>   http://matplotlib.sf.net/interactive.html
>
> JDH

I did set interactive.

I was using QtAgg.

I hate tk.

Re: [Matplotlib-users] plotting from xemacs always 1 step behind

[John H. <jd...@gm...>]

On Feb 12, 2008 8:50 AM, Neal Becker <ndb...@gm...> wrote:

> I did set interactive.
>
> I was using QtAgg.
>
> I hate tk.

[reposting in the correct thread]

Well, the problem is that GTK, WX and Qt require threading support to
use them properly interactively.  ipython has special modes for these
to run the GUI mainloop in the separate thread.  tk is special in this
regard, in that it runs from a standard python shell w/o threading
support.  You may be able to configure ipython within xemacs to take
advantage of both xemacs and ipython's support for interactive qt
plotting in pylab.

JDH

Re: [Matplotlib-users] plotting from xemacs always 1 step behind

[Phil A. <pha...@gm...>]

John Hunter wrote:
 >
 > Well, the problem is that GTK, WX and Qt require threading support to
 > use them properly interactively.  ipython has special modes for these
 > to run the GUI mainloop in the separate thread.  tk is special in this
 > regard, in that it runs from a standard python shell w/o threading
 > support.  You may be able to configure ipython within xemacs to take
 > advantage of both xemacs and ipython's support for interactive qt
 > plotting in pylab.

FWIW, everything's working correctly for me using
xemacs + ipython:

Xemacs 21.5
current svn snapshot of python-mode.el from
   http://sourceforge.net/projects/python
current ipython svn snapshot
Centos 5.1 and python 2.5.1


 -- Phil

Re: [Matplotlib-users] plotting from xemacs always 1 step behind

[Neal B. <ndb...@gm...>]

Phil Austin wrote:

> John Hunter wrote:
>  >
>  > Well, the problem is that GTK, WX and Qt require threading support to
>  > use them properly interactively.  ipython has special modes for these
>  > to run the GUI mainloop in the separate thread.  tk is special in this
>  > regard, in that it runs from a standard python shell w/o threading
>  > support.  You may be able to configure ipython within xemacs to take
>  > advantage of both xemacs and ipython's support for interactive qt
>  > plotting in pylab.
> 
> FWIW, everything's working correctly for me using
> xemacs + ipython:
> 
> Xemacs 21.5
> current svn snapshot of python-mode.el from
>    http://sourceforge.net/projects/python
> current ipython svn snapshot
> Centos 5.1 and python 2.5.1
> 
Works for me if invoked from xemacs shell as ipython -pylab.

If invoked from M-x py-shell doesn't work.

Now if I could just figure out where to modify the way py-shell invokes
ipython to add the -pylab.  Doc say C-u M-x py-shell would prompt for args,
but the doc lies.

Re: [Matplotlib-users] plotting from xemacs always 1 step behind

[John H. <jd...@gm...>]

On Feb 12, 2008 12:17 PM, Neal Becker <ndb...@gm...> wrote:

> If invoked from M-x py-shell doesn't work.
>
> Now if I could just figure out where to modify the way py-shell invokes
> ipython to add the -pylab.  Doc say C-u M-x py-shell would prompt for args,
> but the doc lies.

You can probably get help with this on the ipython mailing list.

JDH

Re: [Matplotlib-users] plotting from xemacs always 1 step behind

[Phil A. <pha...@gm...>]

John Hunter wrote:
 > On Feb 12, 2008 12:17 PM, Neal Becker <ndb...@gm...> wrote:
 >
 >> If invoked from M-x py-shell doesn't work.
 >>
 >> Now if I could just figure out where to modify the way py-shell invokes
 >> ipython to add the -pylab.  Doc say C-u M-x py-shell would prompt 
for args,
 >> but the doc lies.

I use:

(require `python-mode)
; for python-mode.el from http://sourceforge.net/projects/python
; If you are using the default emacs python-mode, comment out the following
; two lines
(setq py-shell-switch-buffers-on-execute nil)
(setq py-python-command-args '("-pylab"  "-colors" "LightBG" "-p" "phil"))
(require `ipython)