Jan Teske
Jan Teske
> Most of the views from the mz_catalog (SHOW VIEWS FROM mz_catalog) are cluster-dependent and identifiable by the cluster-id at the end of the name. E.g: mz_catalog.{type}-{cluster_id}. Not quite: Views...
> Produce example of problematic behavior after storage compaction lands I don't think we need to wait for compaction. Persist already returns an error when you try to read at...
> TODO: Discuss Alternative 2: Add a user-configurable parameter that determines a compaction lag for storage. Document it as an advanced feature. Looks like the STORAGE team is already thinking...
I took the liberty to rebase this onto the current main. Unfortunately it's not quite ready for merging yet, as I can still produce the `thread 'timely:work-0' panicked at 'cannot...
The issue is indeed that creating a `persist_source` with an `as_of` of `Timestamp::minimum()` for the introspection sinks fails if the target shard's `since` has already been advanced. We need to...
As far as performance is concerned, flamegraphs suggest that we are now spending signficantly more relative time in inside: * `mz_compute::arrangement::manager::TraceManager::maintenance` and * `mz_compute::compute_state::ActiveComputeState::report_compute_frontiers`. This seems to be because [the...
> Maybe in the HAL use-case it's more common to have larger blocks of code inside the closure, where this change does improve ergonomics? In the HALs you usually find...
I thought about the use of `StableDeref` a bit more. The guarantees it provides with regard to `DerefMut` are a bit hard to grasp. For example, `Vec` implements `StableDeref`, even...
@Sh3Rm4n No worries! And thanks for the effort you have put in so far! @dfrankland has given me contributor rights to this repo now too, so I will be able...
I'm interested in this too. The project I'm working on pretty much demands DMA (in tandem with ADC and SPI). That's a major reason I don't use this HAL for...