Keijo Kapp

Amended with the following patch. Otherwise that key would be outside the overall range, making this specific key unreadable. ```patch diff --git a/fdbclient/ReadYourWrites.actor.cpp b/fdbclient/ReadYourWrites.actor.cpp index 070c3456f..25ffe8cf6 100644 --- a/fdbclient/ReadYourWrites.actor.cpp +++...

Found it by manual white-box testing while trying to solve an another (slightly related) issue. Since I failed to find a solution to that other problem, it might be good...

Yes, but the issue is with abstractions that do not have a way to know about the beginning of the transaction. In my case, the abstraction and the code managing...

@josephg Have you been able to think about it? I see it being quite critical because it prevents an entire large class of use cases. Are there any issues with...

I'd like to also address another (related) issue which may also affect the directory layer. What happens if the transaction gets retried while there are still jobs in the queue?...

Is it okay to overhaul the whole LevelDB adapter and related code? This part hasn't received meaningful maintenance since 80c4f2f1 (from 2017) and there are some seemingly significant issues. I've...

> I wouldn't recommend removing the check for this node directly, as it would introduce breaking changes. I didn't see the original change but I think would be great to...

> But we can never know which of the two conflicting rules users are actually relying on. > Removing the node check directly would be a breaking change for users...

The online documentation is for 4.2.0 only. `ignoreTypes` was added in 4.0.0. I strongly suggest migrating ESLint configuration to ESM. It's not as difficult or scary as it's often made...

fwiw, I weren't been able to reproduce a scenario where fix-on-save doesn't work. Tried most combinations of different ESLint versions (8.5.1 vs 9.22.0), Stylistic versions (3.1 and 4.4) and VSCode...