ComplexSpaces

Hi, thanks for opening this issue. Can I ask what OS (and desktop environment, if applicable) you are seeing this issue in?

We recently switched away from homegrown `BMP` parsing to using the `image` crate, but the same issue remained. I've opened https://github.com/image-rs/image/issues/2547 upstream to see if this can be resolved outside...

This is now fixed upstream in the `image` crate via https://github.com/image-rs/image/pull/2552, pending a release to `crates.io`. Once that's out I'll bump `arboard`'s default version in the `Cargo.toml`.

That's great to hear, thanks for testing to confirm!

It seems like the best generic interface we could expose is `impl Iterator]` because its the most typesafe, but that would require a re-collection (`.collect`) on Windows because of the...

That sounds reasonable to me; the Win32 code for this isn't super bad. We already have some primitives for clipboard memory allocation due to image support. I pretty heavily value...

It doesn't look like Chromium does, so I don't think we need to. They create a `STDMEDIUM` structure [here](https://source.chromium.org/chromium/chromium/src/+/main:ui/base/clipboard/clipboard_util_win.cc;l=610-654?q=CF_HDROP&ss=chromium%2Fchromium%2Fsrc) but then when actually writing the data [only the `hGlobal`](https://source.chromium.org/chromium/chromium/src/+/main:ui/base/clipboard/clipboard_win.cc;l=798-802;drc=10d39cde3307a7249d2dd106684cf5467692f8de;bpv=0;bpt=1) part...

Sorry, missed your comment. I learn towards not wanting to introduce UNC paths because since we're very reliant on other applications reading the pasteboard to know how to handle them...

The new set of changes look good to me. I'm happy to approve this and merge once the last thread is resolved.

This is now released in [v3.6.1](https://crates.io/crates/arboard/3.6.1) on crates.io. Thanks again for the new feature! I hope `servo` continues to make good progress :)