These are chat archives for ipython/ipython

10th
Apr 2015
Min RK
@minrk
Apr 10 2015 04:13
@/all whether the example notebooks (Notebook and/or Widgets) and/or notebook-focused rst docs should follow the notebook repo is really the last question in the way of finishing the big split.
Thomas Kluyver
@takluyver
Apr 10 2015 04:16
of the examples, I think the notebook examples (http://nbviewer.ipython.org/github/ipython/ipython/blob/3.x/examples/Notebook/Index.ipynb ) should follow it, and perhaps the widgets, as we're splitting them later
Min RK
@minrk
Apr 10 2015 04:16
and leave the rst docs for the jupyter-doc repo?
Thomas Kluyver
@takluyver
Apr 10 2015 04:17
I was just typing that I'd be inclined to move the notebook docs along with it
though perhaps not the nbformat doc
and possibly not the nbconvert doc
However. It's past midnight here, and the conference proper starts tomorrow morning, so I'm turning in
night all!
Min RK
@minrk
Apr 10 2015 04:21
That's roughly what I was thinking, thanks.
There's no great cost to losing history or carrying extra .rst files we decide to ditch, since they are small and haven't much history.
Min RK
@minrk
Apr 10 2015 05:26
The example notebooks are large, though. They are a major fraction of the entire IPython repo, so I don't want to copy them to the notebook repo if we are just going to remove them shortly after.
Bas Nijholt
@basnijholt
Apr 10 2015 13:52

I'm trying to start an IPython notebook server on a hpc trough a tunnel. I can already work in the notebook from my computer, but the engines are not launching. I followed the docs: https://ipython.org/ipython-doc/stable/parallel/parallel_process.html

/home/user/python/env_stable/lib/python2.7/site-packages/IPython/parallel/client/client.py:446: RuntimeWarning: Controller appears to be listening on localhost, but not on this machine. If this is true, you should specify Client(...,sshserver='you@192.168.3.206') or instruct your controller to listen on an external IP. RuntimeWarning)

Should I run parallel.Client() with extra arguments or something?

Min RK
@minrk
Apr 10 2015 18:32
@basnijholt how are you starting the engines and controller? Are you running the client on your laptop, or on the hpc system?
Min RK
@minrk
Apr 10 2015 18:51
@/all I'm doing the notebook split now, which should conclude this stage of The Big Split™. I imagine much of my day will be spent working on getting the js tests to run without iptest.
Matthias Bussonnier
@Carreau
Apr 10 2015 18:55
I'm in BIDS, ping we when you want to have lunch.
Brian E. Granger
@ellisonbg
Apr 10 2015 20:58
Jonathan Frederic
@jdfreder
Apr 10 2015 21:16
I like the markdown editor
S. Chris Colbert
@sccolbert
Apr 10 2015 21:36
@ellisonbg @KesterTong Do you guys have any feedback on the component api? I'm getting ready to merge all of my widget cleanup, and will have some time to address any concerns you have for the components.
^^ Phosphor stuff
Jonathan Frederic
@jdfreder
Apr 10 2015 22:06
@Carreau may have feedback, he's been using it too.
S. Chris Colbert
@sccolbert
Apr 10 2015 22:19
cheers
Matthias Bussonnier
@Carreau
Apr 10 2015 22:33
@sccolbert One queston I'll open an issues.
why addStrech as a default of 0
S. Chris Colbert
@sccolbert
Apr 10 2015 22:36
@Carreau because that's the same default used for the optional 'stretch' arg of addWidget. The effect is that, if you use the defaults for everything, it will work as expected. This is because the spacer item generated for addStretch has an expanding size policy, which will take precedence over zero stretch items in the layout algorithm
Matthias Bussonnier
@Carreau
Apr 10 2015 22:38
Sorry opened an issues on the meantime.
wouldn't 1 everywhere works ?
^^^ that's the chain of logic to follow
S. Chris Colbert
@sccolbert
Apr 10 2015 22:43
I added some comments to the issue
r.e. 1 everywhere
that wouldn't be as effective (imo), because as soon as you want something else to stretch, but not others, you have to set everything else to zero
imo, zero is the better stretch default, and it's the same default as used by Qt
Matthias Bussonnier
@Carreau
Apr 10 2015 22:46
Ok, I'll trust your experience.
I was just expecting addStretch(), addStrech(2) to respectively take 1 and 2/3 of screen space.
I'm fine having 0 I'll add some notes.
S. Chris Colbert
@sccolbert
Apr 10 2015 22:47
I could make the stretch value required, that would have symmetry with addSpacing
what are your thoughts on that?
Matthias Bussonnier
@Carreau
Apr 10 2015 22:49
Could be better.
at least people using 0 will know what they are doing.
I mean both make sens, I just found the behavior surprising, so reported it.
S. Chris Colbert
@sccolbert
Apr 10 2015 22:56
fixed
Matthias Bussonnier
@Carreau
Apr 10 2015 22:58
OK, thanks !
Jeff Hussmann
@jeffhussmann
Apr 10 2015 23:07

i just tried to clone master then run 'python -m IPython'. This fails with the message

IPython/config.py:10: UserWarning: The `IPython.config` package has been deprecated. You should import from traitlets.config instead.
  warn("The `IPython.config` package has been deprecated. "
/usr/bin/python: No module named loader; 'IPython' is a package and cannot be directly executed

Am I doing something wrong?

Matthias Bussonnier
@Carreau
Apr 10 2015 23:07
No we are refactoring so these days are not the good one to use master.
we did announce that on the ML and twitter.
you will have to install Jupyter_ repost wit being core, clients, and a few other
as well a ipython/traitles...
Jeff Hussmann
@jeffhussmann
Apr 10 2015 23:09
OK, thanks!
Matthias Bussonnier
@Carreau
Apr 10 2015 23:09
or revert to 3.x
Jeff Hussmann
@jeffhussmann
Apr 10 2015 23:13
if you don't mind bearing with someone who doesn't really understand git - I have a fix for a small bug in tab completion of dictionary keys that I was going to submit a pull request for - how should I go about this with respect to 3.x vs master?
Min RK
@minrk
Apr 10 2015 23:16
Ideally, submit a PR to master and we'll backport it to 3.x. It might be a little while before it's convenient to work from master, so if you want to submit a PR to 3.x, we can foward-port it to master.
Matthias Bussonnier
@Carreau
Apr 10 2015 23:19
In terminal to checkout 3.x you want to do something like git checkout -b 3.x origin/3.x
It will put the repo in the state where you work with the 3.x branch.
You can do your fixes on that.
you can also stay on master and manually install the few dependencies. If you need specific help with git, you can ask.
Jeff Hussmann
@jeffhussmann
Apr 10 2015 23:24
thanks! I'll submit to 3.x