• Improve navigation so that axes appear right before the data
• Improve navigation so that titles for facets appear before the axis
• Remove the faceted variable from tooltips. For example in https://vega.github.io/vega-lite/examples/trellis_bar.html, we don't need to repeat "female" since it's the same value for the whole subview.

Interested in the possibilities of narrative schemas with litvis but found insuficient documentation to get started. Can you link to additional resources?

Currently 'tests' consist of working examples that are verified by visual inspection.

A useful addition would be tests that confirm that the json spec produced by a spec-generating function is what is expected. This would also allow documentation examples to be tested through elm-verify-examples.

We might also consider validating against Ve

@domoritz please let me know if I should go forward with this :)

Look at on_altair_error() to see we can print out the original error message; then temporarily set dist: trusty on Travis to make sure we get a descriptive error.

Currently the preview is not updated, when the data file referenced via url is changed and saved.

The file seems already to be known, because it can be opened in the preview. Would it be possible to listen to its save events and refresh the preview accordingly?

Many thanks for this extension!

documentation says that there is an autoResize option (link below)... however when using it, it doesn't seem to resize automatically. Adding autosize options along with it does not seem to work either. Is there something I am doing wrong?

<vega-lite :data="data" :mark="mark" :encoding="encoding" :autosize="{type:'pad', contains:'padding', resize:true}" autoResize/>

https://ne

• sequelize-typescript based
• lightweight schema - tensors saved as jsonb
• migration scripts
• store training data in table 1
• store predict input and output in table 2
• load training data on start to train the model
• load old predict to frontend graph

The current readme.md has explanations about how to write python code that does the things we want. It needs a new section, directed at people who only want to use the code and never want to look inside the code. You might have more than one .md file if you want, and link them together, that's fine, but here's the kinds of things we want it to cover:

1. Overall flow: simulation -> JSON stats -