altair-viz / jupyter_vega Goto Github PK

Model after jupyterlab's new IMimeModel API:

(error, result) => {
  const imageData = result.view.toImageURL().split(',')[1];
  this._data.set('image/png', imageData)
}

https://github.com/altair-viz/jupyterlab_vega/blob/master/labextension/src/output.js#L52-L53

These have not been released officially but will be released around April. Just as a heads up.

It looks like jupyterlab_vega is no longer available for pip to install.

$ pip install jupyterlab_vega                      (jlab)
Collecting jupyterlab_vega
  Could not find a version that satisfies the requirement jupyterlab_vega (from versions: )
No matching distribution found for jupyterlab_vega

When I try to use altair in jupyter lab 0.31.5. I get the message

JavaScript output is disabled in JupyterLab

Can I use altair in the current version of jupyterlab?

is it normal ?

Follow up from jupyterlab/mimerender-cookiecutter#39
We can try:

git filter-branch -f --tree-filter 'rm -rf ./jupyterlab_vega/static' HEAD
git filter-branch -f --tree-filter 'rm -rf ./nbextension/embed' HEAD

This can cause weird conflicts between branches, so I think it's best to do when there are no outstanding PRs...

In #3 we discussed adding versioned MIME types:

application/vdn.vega.v2+json

application/vnd.vegalite.v1+json

I think this is some copy-paste error here ?

I thought I'd get the maintainer's blessing before removing it in a PR...@gnestor ?

Hi,

Is there support for layered vega-lite plots (especially the kind that uses multiple SingleViewSpec instances, each one having its own data source)?

If not, is it planned?

Thanks!

Otherwise npm install will fail

See this:

https://github.com/jupyterlab/jupyterlab_geojson/blob/master/jupyterlab_geojson/__init__.py#L23

I will work on this now...

jupyerlab-extensions

Hi guys,

We are building a Java kernel for ImageJ and we would be interested in adding plotting capabilities in it. See this issue for details scijava/scijava-jupyter-kernel#56

So I am looking for a standardized and well-supported way of plotting things from a Jupyter kernel.

I feel like vega could do the job. If I get it correctly vega is a specification. So I would need to generate some vega-compatible JSON on the kernel side and send it to the Jupyter client with the appropriate mimetype so it can render it using a vega compatible JS library. Am I correct?

Some questions:

• Does it work now and if it does how can I install it? I got this error Could not find a version that satisfies the requirement jupyterlab_vega (from versions: ) trying to install with jupyterlab 0.21

• Is this extension working on both Notebook and Lab?

• Are you aware of a Java library that can output vega JSON?

Thank you !

ping @ellisonbg @gnestor @jakevdp

Thinking about adding a conda-forge recipe now that the release is stable (and now that we're transitioning altair master to this package).

Is there any reason to keep the old ipyvega, or can we essentially replace it with this? IMO it might be nice to use the vega name on PyPI and conda-forge for this package.

Is there any good reason not to do this?

In ipyvega, when a vega/vegalite viz is rendered we also save a static png version in the notebook output. This is so that these things render on GtiHub and other contexts where full HTML is not supported. We need to implement that here as well. Here is the code (commented out) to do this in JupyterLab:

https://github.com/altair-viz/jupyterlab_vega/blob/master/src/rendererwidget.ts#L66

@gnestor

Jupyterlab now comes with Vega by default. I think this repo should be archived.

Similar to jupyter/jupyterlab_geojson#24

If we create a conda package we can automate the installation of all the extensions. We should do this before releasing a new version of Altair. The ipyvega feedstock has the needed scripts:

https://github.com/conda-forge/vega-feedstock

@jakevdp