@rough-minister-58256 That display looks awesome! 😄 Reading up this discussion a bit… I agree with @aloof-angle-91616 that we need some manual curation here, and I suspect we should probably borrow something from @fancy-queen-20734’s cut points thinking. It feels like an

@rule

should be annotated with “Is this a significant level for the UI”, and we should collect subgraphs under those groupings. As a user, I think I’d like lines to look something like:

path/to/target:name: compile: running javac [10 seconds so far]
path/to/other/target test: listing files in directory X [1 second so far]

Where the first column is pretty stable, the second column scrolls slowly with a high-level goal, and the third column scrolls quickly with the actual graph node being evaluated. Probably 3 isn’t a fixed number, and we show as many columns as there are “cut-point”-annotated

@rules

. Maybe even we fix the first column as the target roots, but until we start doing actual work for a target root, we either hide it entirely, or have its second column being the dependency it’s working on to work for that target root, and severely down-rank it because it’s not very relevant.

i hadn't been assuming that the "cut points" in that idea were going to be manually specified... rather, calculated.

cut points is a good word for it