These are chat archives for ipython/ipython

11th
Dec 2015
NEETFR3AK
@NEETFR3AK
Dec 11 2015 00:20
Anyone here willing to look over a students poor code? It'll help me get a job : )
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 05:48
Hi everyone.
Has anyone experimented with OpenGL/WebGL in the notebook environment?
As mentioned in gibiansky/IHaskell#619, @conal has a haskell to GLSL compiler, which he suggests can be hooked into IHaskell.
Sylvain Corlay
@SylvainCorlay
Dec 11 2015 13:36
@sumitsahrawat There is PyThreejs and vispy
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 16:19
Thanks @SylvainCorlay and @AaronWatters. I'll take a look.
Sylvain Corlay
@SylvainCorlay
Dec 11 2015 17:27
@sumitsahrawat if you have a gamepad, you can play the Jupyter flight simulator
which uses pythreejs
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 17:28
Great. I'd like to try that out.
The is a binder for it
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 17:29
:+1:
Nicholas Bollweg
@bollwyvl
Dec 11 2015 17:32
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 17:35
Awesome :+1:
The voxels example didn't work for me though.
It worked on a reload :smile:
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 17:43
@SylvainCorlay The flight simulator is taking a long time :smile:
Sylvain Corlay
@SylvainCorlay
Dec 11 2015 17:44
Taking a long time to?
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 17:45
What are the dependencies, I'd like to try it offline.
Sylvain Corlay
@SylvainCorlay
Dec 11 2015 17:45
load the terrain?
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 17:45
Nope, execute some cells.
The "Terrain modeling..." part.
It might be a problem with my internet connection.
Sylvain Corlay
@SylvainCorlay
Dec 11 2015 17:46
I don't know how powerful the VM is, but I take a large chunk of grand-canyon terrain elevation data, which is a bit violent.
@sumitsahrawat did it work eventually?
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 18:01
I downloaded the notebook, and installed pythreejs.
It errors out saying couldn't find Controller in ipywidgets. Are you using a modified version?
Sylvain Corlay
@SylvainCorlay
Dec 11 2015 18:02
ipywidgets is in recent releases of ipywidgets
which version are you using?
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 18:03
The one which came bundled with ipython. pip show ipywidgets shows 4.0.2.
Sylvain Corlay
@SylvainCorlay
Dec 11 2015 18:03
(I think it was in 4.0.3, and current is 4.1.1)
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 18:04
That gives me one more thing to implement for IHaskell :smile:
Sylvain Corlay
@SylvainCorlay
Dec 11 2015 18:04
:)
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 18:04
Have there been any more changes?
Sylvain Corlay
@SylvainCorlay
Dec 11 2015 18:04
in ipywidgets?
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 18:04
New widgets, or changes to the spec.
Yup.
Sylvain Corlay
@SylvainCorlay
Dec 11 2015 18:05
Yes: changes re: persistence.
But I don't think that it should impact the design of models.
The best way to make this work in the future (notebook 4.1)
is to implement the comm_info_[request/reply] in the IHaskell kernel.
Which will make it work starting with notebook 4.1.
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 18:08
Oh, I remember. That is the suicidal comm workaround, right?
Sylvain Corlay
@SylvainCorlay
Dec 11 2015 18:08
IPywidgets currently implements a workaround to the absence of this message, which we will remove in ipython/ipywidgets#175
yes
But there were changes in the implementation of persistence to take advantage of this new message (or the workaround)
But the best thing to do would be that the IHaskell kernel implements the message.
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 18:10
Yup. I'll look into it.
Sylvain Corlay
@SylvainCorlay
Dec 11 2015 18:10
(And switch to notebook 4.1 as soon as Matthias releases the beta ;) )
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 18:11
Yup. I'm not a python person primarily, but I'll try to keep up with notebook versions.
matplotlib is huge. I'll let it install and then try the flight simulator.
Sylvain Corlay
@SylvainCorlay
Dec 11 2015 18:15
@sumitsahrawat ipywidgets 5.0 will have model changes, that will require the haskell binding to adjust.
But then, we will have a proper npm package, it should be easier moving forward.
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 18:16
What are the proposed changes?
Sylvain Corlay
@SylvainCorlay
Dec 11 2015 18:17
DOMWidget has a top level attribute "Style" which itself is a widget
and carries all the styling attributes.
instead of having them top-level.
That is the main one.
Sumit Sahrawat
@sumitsahrawat
Dec 11 2015 18:18
Great. To create an experience similar to ipywidgets, the ihaskell-widgets package has a LOT of repetitive code.
This will simplify it, I think.