Re: [Matplotlib-users] Event blocking

[John H. <jdh...@ac...>]

>>>>> "Malte" == Malte Marquarding <Mal...@cs...> writes:

    Malte> John, When do you expect this to be implemented?

    Malte> I'd like to roll out a package depending on matplotlib and
    Malte> have a request for this specific feature.

I took a look at implementing this last week and talked with Fernando
Perez who did the matplotlib pylab support for ipython, we came to the
conclusion, that this needs to be done at the interactive shell
level.  The basic problem is that although you are blocking, you need
to still share cpu time with the gui thread, so that you can continue
to interact with your plot and generate events.  What shell and
backend are you using?

In the meantime, you might want to suggest this idiom to your user who
is asking for a blocking event

class Catcher:
    event = None
    def __call__(self, event):
        self.event = event

plot([1,2,3])
c = Catcher()

connect('button_press_event', c)

# no go click on the axes
print c.event.xdata, c.event.ydata

# and click somewhere else
print c.event.xdata, c.event.ydata

So even though the call is not blocking you have ready access to the
event data in as way that doesn't feel like event driven programming
with callbacks.

I like this approach, actually.  I wonder if it would be useful to
assign some new pylab module variables

bp = Catcher()
connect('button_press_event', bp)

br = Catcher()
connect('button_release_event', br)

kp = Catcher()
connect(key_press_event', kp)

and so on.  Then in the pylab interface, w/o making any connections
yourself, you would always have access to the last keypress event as

kp.event

and so on.

Do people think this would be useful for interactive work?

JDH