Hey Knit noggins! Are you reaadddyyyyyy to ruuuuuuuuum-
Web stuff
-ahem. Are you ready to data pull request?
It's not much right now, just diffs and comments. But it's great to have a platform to explore richer ideas and present higher fidelity demos. I've been curious about Svelte, so that's what this is using for now. Lots of improvements are needed to make this useful, but a couple interest me most right now.
Unlike code, most data is not manipulated as text. So it's probably important to figure out a flexible way to present graphical data and data summaries like aggregates or anomalies. Also unlike code, data often changes on its own.
Code can be cleanly prepared into "commits" so reviewers can just compare two points in time; but for data it's more valuable to be able to look at it over many points in time, such as to see trends. This is a tricky, somewhat meta problem, but I think it's pretty interesting.
Diff stuff
To build a diff, first we traverse the graph structure of a flow invocation and build an index (essentially a run manifest). We do this for a second flow invocation, and diff the two indexes, just like two directories of files.
-uuuumbbblllleeeeeee!!!