Can confirm, am still unsure if DDlog [0] can be switched to Worst-case optimal joins (WCOJ) [1] with the recent (unreleased, but almost 1y old) calculus operat

Can confirm, am still unsure if DDlog [0] can be switched to Worst-case optimal joins (WCOJ) [1] with the recent (unreleased, but almost 1y old) calculus operators of DDflow [2][3], because at least the original dogs^3 approach supposedly doesn't work in iterative contexts (which are necessary for recursive operations, like graph computations). The calculus blog post ends on a promising note, however. I'm trying to help a couple (friends?) with getting the analysis of rslint [4] running well on DDlog or at least DDflow, with the end-goal being a perceptually zero-latency linter that typically responds faster than a human types. We're currently seeing initial delays in the single-digit second range, and that's not even on large projects (the incremental performance is far better, but we would like to out-compete the official TS typechecker even in CI settings that don't keep the linter's state across runs). The good news: we're making nice progress on profiling tools and I might get to trying some WCOJ code later today. [0]: https://github.com/vmware/differential-datalog https://github.com/vmware/differential-datalog [1]: https://github.com/TimelyDataflow/differential-dataflow/tree/master/dogsdogsdogs#worst-case-optimal-joins https://github.com/TimelyDataflow/differential-dataflow/tree... [2]: https://github.com/frankmcsherry/blog/blob/master/posts/2020-02-15.md https://github.com/frankmcsherry/blog/blob/master/posts/2020... [3]: https://mtrlz.dev/api/rust/dogsdogsdogs/calculus/index.html https://mtrlz.dev/api/rust/dogsdogsdogs/calculus/index.html (3rd-party hosting of docs for the calculus/dogs^3 crate) [4]: https://github.com/rslint/rslint https://github.com/rslint/rslint