macosx broken on py3.4 (homebrew, anaconda)

Unless there is something odd about my machine and attempted setups, we have a very bad situation. I have OS X 10.9 (Mavericks). I am seeing the same problem with homebrew python and pip-installed numpy, pip-installed mpl from git 1.4rc3, and from Anaconda 2.0.1, all with python 3.4.

Do the normal thing: fire up ipython, use %matplotlib magic, import numpy and pyplot, make a plot. The macosx backend comes up as the default. A plot appears, but it is not interactive.

Maybe this is related to the framework vs non-framework build question. In a way, I don't care--the important point is that using *common* methods of trying to get started with ipython and mpl on up-to-date python, one immediately runs into a basic failure.

If this can't be fixed *fast*, I suggest we make the macosx backend available only by special request, not by default, as of a 1.4 release.

Yes, this can go to a github issue if appropriate, but first I want to get thoughts from a wider audience.

Comments?

(I'm wondering why this hasn't driven someone else nuts some time ago--hasn't anyone else run into this?)

Eric

Clue: my first try with homebrew was in a virtualenv. Now I installed everything directly, without the virtualenv, and the first plot trial worked. Then I tried restarting ipthon and using "%matplotlib qt". Plotting was OK, but after closing the plot window and going away for a bit, upon coming back, the ipython terminal was extremely sluggish. App-nap problems? I was having similar response problems in some earlier trials as well, with ipython in anaconda. I thought all this was supposed to have been cleared up by now...

I went back to anaconda, and verified I am still getting the unresponsive window--it can't even be resized.

Eric

···

On 2014/08/13, 1:19 PM, Eric Firing wrote:

Unless there is something odd about my machine and attempted setups, we
have a very bad situation. I have OS X 10.9 (Mavericks). I am seeing
the same problem with homebrew python and pip-installed numpy,
pip-installed mpl from git 1.4rc3, and from Anaconda 2.0.1, all with
python 3.4.

Do the normal thing: fire up ipython, use %matplotlib magic, import
numpy and pyplot, make a plot. The macosx backend comes up as the
default. A plot appears, but it is not interactive.

Maybe this is related to the framework vs non-framework build question.
  In a way, I don't care--the important point is that using *common*
methods of trying to get started with ipython and mpl on up-to-date
python, one immediately runs into a basic failure.

If this can't be fixed *fast*, I suggest we make the macosx backend
available only by special request, not by default, as of a 1.4 release.

Yes, this can go to a github issue if appropriate, but first I want to
get thoughts from a wider audience.

Comments?

(I'm wondering why this hasn't driven someone else nuts some time
ago--hasn't anyone else run into this?)

Eric

Can you make a blocker issue for this?

Is it worth looping the ipython devs in on this email/issue as well?

···

On Wed, Aug 13, 2014 at 8:13 PM, Eric Firing <efiring@...229...> wrote:

On 2014/08/13, 1:19 PM, Eric Firing wrote:

Unless there is something odd about my machine and attempted setups, we
have a very bad situation. I have OS X 10.9 (Mavericks). I am seeing
the same problem with homebrew python and pip-installed numpy,
pip-installed mpl from git 1.4rc3, and from Anaconda 2.0.1, all with
python 3.4.

Do the normal thing: fire up ipython, use %matplotlib magic, import
numpy and pyplot, make a plot. The macosx backend comes up as the
default. A plot appears, but it is not interactive.

Maybe this is related to the framework vs non-framework build question.
  In a way, I don't care--the important point is that using *common*
methods of trying to get started with ipython and mpl on up-to-date
python, one immediately runs into a basic failure.

If this can't be fixed *fast*, I suggest we make the macosx backend
available only by special request, not by default, as of a 1.4 release.

Yes, this can go to a github issue if appropriate, but first I want to
get thoughts from a wider audience.

Comments?

(I'm wondering why this hasn't driven someone else nuts some time
ago--hasn't anyone else run into this?)

Eric

Clue: my first try with homebrew was in a virtualenv. Now I installed
everything directly, without the virtualenv, and the first plot trial
worked. Then I tried restarting ipthon and using "%matplotlib qt".
Plotting was OK, but after closing the plot window and going away for a
bit, upon coming back, the ipython terminal was extremely sluggish.
App-nap problems? I was having similar response problems in some
earlier trials as well, with ipython in anaconda. I thought all this
was supposed to have been cleared up by now...

I went back to anaconda, and verified I am still getting the
unresponsive window--it can't even be resized.

Eric

------------------------------------------------------------------------------
_______________________________________________
Matplotlib-devel mailing list
Matplotlib-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/matplotlib-devel

--
Thomas Caswell
tcaswell@...149...