Upgrade windows crate to 0.62 #518
Conversation
Signed-off-by: Nico Burns <nico@nicoburns.com>
| numbersubstitution: *mut Option<IDWriteNumberSubstitution>, | ||
| numbersubstitution: OutRef<'_, IDWriteNumberSubstitution>, | ||
| ) -> windows::core::Result<()> { | ||
| numbersubstitution.write(None)?; |
There was a problem hiding this comment.
Not 100% sure this is the correct way to translate this into the new API
There was a problem hiding this comment.
I looked a bit into this and I think the answer is yes, this is the way. However, the surrounding context seems quite arcane.
The COM memory management rules state that all out pointer parameters (including pointer members of a caller-allocate callee-fill structure) must explicitly be set to NULL. This is what the old code was presumably doing with None and this new write call replicates that.
I do find it scary that "safe" Rust code can be written so easily that results in uninitialized memory being read by the OS. I did not find any signs that windows adds any safety for this purpose, but it's possible I missed it. I did find various flavors of the following statement: metadata doesn't currently include information about which fields require automatic initialization. Meaning that the bindgen just doesn't know which parameters need extra safety vs. where it would ruin things.
It's not just the lack of well structured metadata either. Human readable docs are quite limited. GetNumberSubstitution docs have little content with no info about safety. The Rust bindings (1, 2) have literally zero documentation.
A few relevant links to this specific change:
- windows-rs#3025 added
OutRef - windows-rs#3386 started using it for COM interfaces
OutRefsource
There was a problem hiding this comment.
Yeah, I also looked for docs on this and struggled to find any.
2 participants
Upgrades the
windowsandwindows-corecrates to the latest 0.62 versions