rivenskaye / winmmf-rs

A crate to support Memory-Mapped Files for IPC on Windows

Welcome, Stranger

Mind telling me what the heck you're doing here? I'm not that interesting! I do have a blog for when I do something mildly interesting. I suppose you should enjoy the stat dump while you're here though. Oh and the short list of suggested info too, those seem fun!

• 🔭 I'm currently working on getting a Rust implementation of Named Shared Memory working, adding QoL to windows-rs, and mental preparation to update Zwei again;
• 🌱 I'm currently learning more about Windows internals than I'd like to know;
• 👯 I'm looking to collaborate on anything really; My extensions to windows-rs kinda died because I'm pretty darn busy. Would love to help on projects I like using though!
• 🤔 I'm looking for help with windows-rs-ext mainly.
• 💬 Ask me about what I did last night! I swear it's not a bad joke, I make long days at work so night time is when I do fun stuff.
• 📫 How to reach me: Discord @rivenskaye, email Riven [at] tae.moe, or by telex
• 😄 I'd like to get some more time free to continue playing with the CCITT encoder/decoder I was planning with a friend. No, fax is not dead.
• ⚡ Fun fact: monochrome TIFF images embedded in PDFs are still valid CCITT fax format imagery!

winmmf-rs's Issues

Generalize sizes to native pointer size

It shouldn't be too hard to rewrite current code to generalize the mapping sizes. Though I can't imagine anyone needing more than u32::MAX bytes of IPC space, it's possible and should be implemented.

Fix lock safety

Use compare_exchange for locks to prevent races in the very brief window where we modify lock values as per Discord talks with @xacrimon after his blessed fixes.

Might also want to look into fetch_update after doing some of my own digging; seems like I could keep it relatively smooth that way?

Add guards

@xacrimon pointed this out - currently there's nothing preventing reads and writes from being reordered dangerously. This can cause a funny to happen

Add unit tests

They're missing while they should be there to ensure changes don't cause problems.
Would need to find some manner of running two processes at once to validate things properly though

Add some usage examples

Properly commented examples in docs or an example crate in the repo should make things easier to use. Perhaps this could even help get #2 up and running?