After a bit of goal-pruning, Thje #Vizier 2.0 #Notebook is finally at RC1 (https://github.com/VizierDB/vizier-scala/pull/116). This is a completely new UI with (i) snappier push-based updates, (ii) an overview UI, (iii) A cleaner, tighter design, (iv) native support for pyenv, (v) infinite-scroll datasets, (vi) better data vis.

Thanks to all who contributed to the redesign over the past 2 years, including mrb and lordpretzel for their dev efforts, and tslowe for testing and design feedback.

(1/N)

Version 2.0 by okennedy · Pull Request #116 · VizierDB/vizier-scala

This pull marks v2.0 as stable https://github.com/VizierDB/vizier-scala/milestone/6 Note: This release WILL very likely break the 1.x vizier.db file format. To migrate to the 2.x format: Export y...

GitHub

One new feature is native support for #PyEnv. You can create new environments for the project and then select which environment to use on a cell-by-cell basis. This is really handy for #Reproducibility (since the project records a snapshot of the environment settings). It also allows you to mix-and match incompatible python packages across python cells.

(2/N)

We've completely re-done the "new cell" interface as well, emphasizing the commonly-used cell types, giving them recognizable icons, and tying their positions to their role in the lifecycle of an artifact, rather than the "class" of operations to which they belong.

(3/N)

The workflow area provides a high-level overview of the notebook, with cells either auto-generated, or taken from the header (resp., comment) on the first line of a markdown (resp., script) cell. Markdown headers are used to signal emphasis.

Color indicates cell state (dark blue=done, light blue=stale or running, red=error). Clicking on a cell jumps you to it, and mousing over highlights it.

(4/N)

The brand new Artifacts area summarizes the state of your notebook, including any data visualizations generated, datasets loaded/created, etc... The interface provides convenient links to download the artifact, jump to the cell that created it, and/or open it in type-specific editors.

(5/N)

Loading datasets is now much easier with a re-designed data loading flow. Traverse your local filesystem or one of Vizier's supported data source integrations to find the dataset you want to import. As you do so, Vizier will automatically populate sensible defaults into the load configuration.

Although the UX flow is now much cleaner, the design is admittedly still a bit ugly... we're working on that for 2.1.

(6/N)

On that note, Vizier 'spreadsheet mode' now:
(i) supports infinite scroll (no more paging through the sheet)
(ii) is inlined directly into the notebook (see your edits right there)
(iii) is one cell, rather than the list of cells we used to have for each edit.

Thanks to caleblev for a TON of effort working through some of the new spreadsheet UI.

(7/N)

In short, there are tons of new features in 2.0.

There's also a lot of amazing features on the roadmap for 2.1:
(i) vdib is working on a slick Jupyter compatiblity feature
(ii) the UB EL Vizier team is working on a nifty data vis design tool based on #Vega
(iii) we have ideas for visualizing inter-cell dependencies.
(iv) An auto-generated docs area based on active packages and pypi/maven metadata.
(v) support for opening scripts in a local editor (based on work from a past UB 611 team)

(8/8)