Ben Clayton
Ben Clayton
WGSL used to have the `[[stride(n)]]` attribute. See: https://github.com/gpuweb/gpuweb/issues/2493, https://github.com/gpuweb/gpuweb/pull/2503 The group decided to remove this attribute for the reasons in https://github.com/gpuweb/gpuweb/issues/2493. We also have a long standing desire to...
Seems reasonable. My initial thoughts / questions: * May I suggest `.` for accessing inside a namespace instead of `::`? `::` is very C++ / Rust, and `.` is more...
> revert order-independent module declarations for V1, since I believe that order-independent module declarations provide little benefit without a scope/namespaces tool We (Google) have found a number of use cases...
Happy to see that this aligns with golang's behaviour. To quote [the spec](https://go.dev/ref/spec#Operators): > Except for shift operations, if one operand is an untyped constant and the other operand is...
Naming wise, what you're proposing seems to better fit "multiple return values" than tuples, as tuples usually represent a new aggregate type. Introducing a new tuple type to the language...
I've put together some slides for the meeting later today, which covers the current limitations around pointers, these proposals, and includes a new proposal which may be of interest.
In my career, I've encountered countless bugs due to developers assuming that `std::numeric_limits::min` returns the lowest representable and finite value for the type `T`. `std::numeric_limits::min` has two distinct behaviours between...
Hi @zhufangda , No, there's currently no such primitive. I'd like to understand the use case a little more. Please can you describe the problem you are trying to solve?...
Hi @zhufangda, Thank you for the additional context. So I assume you're constructing the `marl::Scheduler` with multiple threads, and not using marl in fiber-only mode? I'll assume this is the...
Hi @Guillaume227, Marl doesn't really make any guarantees about order of execution - scheduled tasks that are unblocked can execute in any order. Pushing an unblocked task back on the...