Safer DirectX: Types, fns, lifetimes, tests, real docs, intellisense, examples, debug visualizers, ...
🦀 | When | DirectX APIs |
---|---|---|
📦 | examples • rust ⮀ c++ • test coverage • lib.rs • docs.rs | |
✔️ | Now | thindx::{d3d9, d3d::Compiler, xaudio2, xinput} |
Soon™ | d3d11, d3d12, dxgi, dxcompiler, dinput | |
Eventually | d2d, dcompute, dsound, dstorage, dwrite, dxr, xact3, uwp::input? | |
❌ | Never? | d3d10, d3dx*, ddraw, dplay |
I'm exposing a huge legacy C++ API to Rust. Mistakes will happen.
That said, soundness is a very high priority goal. thindx
will add things like extra bounds checks, parameter
validation, extra init, etc. if need be in order to ensure soundness in safe fns whenever possible. When it's not
possible to validate unsoundness away, the fns in question should be marked unsafe
. This crate strives to be sounder
than whatever manual FFI you'd write yourself would be, and that's a high bar.
But there are some practical limits to this. If a background driver thread invokes UB if it fails to allocate memory, without any direct correlation to specific API misuse like a large integer overflowing, that's a bug I can't sanely mitigate via safe fns. I mean, theoretically I could write a pure-software clone of the entire DirectX runtime... but no.
Additionally, while I'm seeking to validate my APIs via testing, older implementations of the APIs in question may have more bugs / unchecked parameters / ??? that I'll fail to mitigate due to being unable to trigger them myself. While I'm happy to investigate, accept pull requests, expand test coverage, etc. it's worth assuming this crate is unsound on older versions unless you've tested yourself.
0.0.0-yyyy-mm-dd
doesn't follow semver.
Individual fn
s are likely to gain/lose unsafe
, traits, etc. in a neverending attempt to make DirectX access sound.
As such, thindx
itself will likely always suffer from major version churn.
This isn't too much of a problem until two crates wish to share / pass thindx
types between themselves.
It might be possible to somewhat stabilize some types by exiling them into subcrates, but this has not yet been done.
Additionally, individual extension traits / functions / methods will likely never get the same treatment (no need?)
Licensed under either of
- Apache License, Version 2.0 (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)
- MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)
at your option.
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.