Eduard-Mihai Burtescu

The `README` in this repo currently states: > The `std` crate relies on [`backtrace`](https://github.com/rust-lang/backtrace-rs), which depends on [`libunwind`](https://github.com/libunwind/libunwind) in the default builds for unix. To work around this, `cosmopolitan.a` currently...

v0 mangling should avoid backrefs when they're not shorter than their target.

2

While working on #61486 / #87194 I came across backrefs in byte array values (with repeating elements), which I hadn't thought about (in terms of implications on optimizing the length...

In order for the scripts to run at all on as many systems as possible, this is ideal: ```patch -#!/bin/bash +#!/usr/bin/env bash ``` The only other hurdle I hit trying...

`"rust.all_targets": false` still builds cfg(test) targets.

1

See #1714 for context, but I suspect this can be repeated on even simple crates, they just build much faster so it's harder to tell. (Just checked on another project...

There should be a way to filter which packages in a workspace are built.

1

I work on [Rust-GPU](https://github.com/EmbarkStudios/Rust-GPU) nowadays, and specifically the `rustc_codegen_spirv` crate in it, but that has a bunch of downstream users in the same workspace, so right now I have to...

The first commit is a pretty large refactor that ended up a bit worse than I expected. We might just want to have this instead (and call it in a...

Cargo build scripts have the ability to produce output through Cargo via [`cargo:warning=MESSAGE`](https://doc.rust-lang.org/cargo/reference/build-scripts.html#cargo-warning) *even when successful*. We should probably make `spirv-builder` transform `stderr` lines from its nested `cargo build` (of...

There's this comment I left in the structurizer (and the `do`-`while`->`while` conversion that follows it): https://github.com/EmbarkStudios/rust-gpu/blob/03f89e8ba6f236218b3c5f9b18fe03c25a4d6a5c/crates/rustc_codegen_spirv/src/linker/structurizer.rs#L307-L310 Which seemed reasonable until I realized that `do`-`while` makes more sense for structured control-flow,...

We should ensure entry interface types have reliable layout.

13

Sadly, it's hard/impossible to account for `#[cfg]`s in e.g. `glam`, and vector types in general tend to have weirder alignment rules than scalars (and aggregates thereof). But we should at...

The [description from the spec](https://www.khronos.org/registry/SPIR-V/specs/unified1/SPIRV.html#_memory_operands) of `NonPrivatePointer` is: > **NonPrivatePointer** > The memory access obeys inter-thread ordering, as specified by the client API. > > Missing before version 1.5. My...