Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • Dec 24 2015 04:29

    freeman-lab on master

    Update README.md (compare)

  • Dec 09 2015 19:18

    freeman-lab on master

    Adjust color (compare)

  • Dec 09 2015 19:16

    freeman-lab on master

    Add logo (compare)

  • Dec 01 2015 06:59

    freeman-lab on master

    Update README.md (compare)

  • Dec 01 2015 06:59

    freeman-lab on master

    Update README.md (compare)

  • Dec 01 2015 06:58

    freeman-lab on master

    Update README.md (compare)

  • Dec 01 2015 06:52

    freeman-lab on master

    Add Gitter badge Merge pull request #1 from gitt… (compare)

  • Dec 01 2015 06:52
    freeman-lab closed #1
  • Dec 01 2015 06:42
    gitter-badger opened #1
Nicholas Sofroniew
@sofroniewn
ah nice suggestion - have to run now, but will look into it - thanks!!
Jason Wittenbach
@jwittenbach
@sethvincent have you used choo much? I see that you’ve been active on a few issues in the repo...
Kyle Kelley
@rgbkrk
@freeman-lab and @sofroniewn - could I get your opinion on something for a notebook
I'm extracting the core of ipywidgets out of jupyter to provide a flexible API for allowing people to update a model front <---> back
ooh, and @mathisonian if you happen to have time as well
:)
Matthew Conlen
@mathisonian
:+1: happy to take a look
Kyle Kelley
@rgbkrk
yay, thanks!
context on the custom mimetypes there - we now support direct geojson as well as Plotly's JSON spec directly
Matthew Conlen
@mathisonian
wow nice
Matthew Conlen
@mathisonian
what would it look like on the python side in this case to update the model?
e.g. if you are doing patching, are you imagining automated diffing to happen internally or is that something that woudl require structured updates on the kernal side as well
Kyle Kelley
@rgbkrk
I'm not real sure
ipywidgets has mechanics for it
They do it over the comms channel
Which is a wide-open pandora's box escape hatch in jupyter
I'd like to prescribe that the diffing is done on the kernel side
What would you want it to be on the python side (or other backend)?
We can always make helpers per kernel to send the right messages and make it easier
IPython.models.update(model_id, ...)
IPython.models.set(model_id, ...)
I'm not a total fan of the naming there
Matthew Conlen
@mathisonian
yeah, something in that direction makes sense to me
i’d be okay with being a little verbose if it provides a way to minimize data transfer
this transform thing -
i’m out of the loop i think, is it something that already exists within nteract?
Yeah
Matthew Conlen
@mathisonian
ah cool
Kyle Kelley
@rgbkrk
We map mimetypes to a React component
and shove it the data that comes from the backend
Matthew Conlen
@mathisonian
so in this case theres always a 1-1 mapping between data model and react component instance
that seems pretty reasonable
Kyle Kelley
@rgbkrk
Yeah, that's the thinking
Matthew Conlen
@mathisonian
i think
Kyle Kelley
@rgbkrk
well
I think I wrote this so you could set a model id assigned to different mimetypes
This is what the jupyter protocol looks like for the display under the hood: http://slides.com/kylekelley/sidecar-5#/6
Matthew Conlen
@mathisonian
okay, i see
Kyle Kelley
@rgbkrk
You may have multiple views of the same model on the page
Matthew Conlen
@mathisonian
yes, was just going to ask that
Kyle Kelley
@rgbkrk
It's up to the contract of how you write the frontend part
Matthew Conlen
@mathisonian
v cool
Kyle Kelley
@rgbkrk
as for registering these transforms, that's a bit of an open question -- we can easily make it part of the state tree
right now I just add the plotly and geojson ones into our default transforms: https://github.com/nteract/nteract/blob/master/src/notebook/components/transforms/index.js
Those don't handle any updates of course, up until we figure out this spec
Once it's happy, I'll start prototyping
Would you want to see lightning embedded in here directly to hack with?