Alexander Alexandrov
Alexander Alexandrov
The `protobuf_roundtrip` is the only way we currently validate that the protobuf serialization works. It has been a good tool to catch bugs in the past, so I would be...
Updating the issue with findings so far based on the following investigation steps: 1. Re-created the issue (in `debug` mode, but it's still indicative) and save the output in [`gdb.txt`](https://github.com/MaterializeInc/materialize/files/9715249/gdb.txt)....
Another thing that we should do is to simplify [the `proptest_oneof!` macro in our `proptest` fork](https://github.com/MaterializeInc/proptest/blob/master/proptest/src/sugar.rs#L326-L476) by eliminating all but the last case which emits the `$crate::strategy::Union::new_weighted` call.
I tried replacing `prop_oneof![$(args),*]` macros with `Union::new([$(args.boxed()),*])` calls, but the big stack frames won't go away. I think there are some `#[derive(Arbitrary)]` instances on enum types that need to be...
I like the general idea, but I am not sure whether we should allow the execution of arbitrary bash commands through our codebase. A bit more robust is to have...
> Once thing I thought about is to add `algorithm` to `sys.path`. I like that idea.
> By this, we can have our cake and eat it too [...] Not sure if this is entirely true, though. We still have to leave the original predicate in...
From #19378: > As part of evaluating an MFP's filters to determine whether it will filter out an entire set of rows, we use AND to combine the results of...
Another related issue: #22529.