vmware / differential-datalog

DDlog is a programming language for incremental computation. It is well suited for writing programs that continuously update their output in response to input changes. A DDlog programmer does not write incremental algorithms; instead they specify the desired input-output mapping in a declarative manner.
MIT License
1.34k stars 117 forks source link

Dev dependencies in prod release #1157

Open yhack opened 2 years ago

yhack commented 2 years ago

The generated Cargo.toml files contain dependencies that don't look production ready.

#differential-dataflow = "0.11.0"
differential-dataflow = { git = "https://github.com/ddlog-dev/differential-dataflow", branch = "ddlog-4" }
#timely = "0.11"
timely = { git = "https://github.com/ddlog-dev/timely-dataflow", branch = "ddlog-4", default-features = false }

I understand this might be done during development to fix a bug or add a feature, but now the forks have fallen behind. What would it take to re-point these dependencies to official releases on crates.io?

ryzhyk commented 2 years ago

@yhack, those aren't really dev dependencies, but slightly customized versions of timely and differential we use in DDlog. I agree we should try to keep them up to date. We aren't that far behind though, in particular I am not aware of bug fixes/significant optimizations that we're missing.

yhack commented 2 years ago

@ryzhyk, gotcha, I guess I took "dev" in the URL too literally. Taking a closer look at your slight customization, I see your comment

Ideally, we would like an adaptive channel size that scales up an down with demand.

If I'm not mistaken, something like that was implemented in master to replace the customized function.