Leonid Ryzhyk
Leonid Ryzhyk
The following worked fine in 0.12: ``` union stage {LS_IN_PRE_LB, LS_OUT_PRE_LB} table __Table_stage { v: stage; } table Flow { stage: stage; // NOTE: field name shadows union name, which...
The `ddshow` report consists of several parts, including the timeline, which can take a very long time to load in the browser. @Kixiron suggested disabling timeline generation using the `--disable-timeline`...
I am adding an option to ddlog CLI to start `ddshow` before initializing the dataflow graph. This should greatly simplify the profiling workflow as the user only needs to type...
`cargo install ddshow` doesn't work, but `cargo install --git https://github.com/Kixiron/ddshow` does.
With the new ovsdb2ddlog adapter generator, DDlog output tables have the exact same schema as OVSDB tables. The notions of uuid_names and uuid_or_str type are gone. Instead every output record...
@frankmcsherry, do you happen to know how come DD is so much faster on this benchmark?
Any chance we could ask benchmark developers to add reference output for each of the three benchmarks? If this is difficult, then at least knowing the size of output relations...
We currently use a private snapshot of the state of input relations to implement set semantics (ignore or fail on duplicates). We should use DD's [`upsert` operator](https://github.com/frankmcsherry/blog/blob/master/posts/2020-03-26.md) instead, which should...
`XFormArrangement` and `XFormCollection`, as well as `Relation` and `Arrangement` should contain a field pointing to the source code fragment(s). This information can be used to generate more informative profiles. cc...
This [branch](https://github.com/ddlog-dev2/differential-datalog/tree/optimize_intern) implements an optimization whereby we intern values by reference, avoiding cloning the value in the common case that the same value already exists in the interner. Unfortunately, this...