3 commits

Author	SHA1	Message	Date
Jeff Hiner	6c1add8b73	Add support for virtual uinput devices (#37) ... * Prototype support for virtual uinput devices * Tidy uinput a bit * Have all the evtest examples use the same args/prompt code * Switch to libc uinput* structs * Use InputId in uinput, use libc _CNT constants * Don't use align_to_mut * Only check /dev/input/eventN files * Spelling fixup * Fix compilation error * Ah, there we go. Much better. Co-authored-by: Jeff Hiner <jeff-hiner@users.noreply.github.com> Co-authored-by: Noah <33094578+coolreader18@users.noreply.github.com>	2021-03-19 09:19:51 -06:00
Noah	3581aa25e0	Syncing impl (#36) ... * WIP syncing impl * Finish up syncing impl * Reorganize synchronization code + add more docs * Wrap libc::input_id * Remove thread::sleep used for testing * Make EventStream::Item io::Result<InputEvent> * Add RawDevice::empty_state() * Update Device rustdoc * Make raw/sync_stream naming consistent * Update crate docs * Fix missing first event of block * Owned AttributeSet, borrowed AttributeSetRef * Add some basic syncing tests * Add some more syncing tests	2021-03-16 20:38:42 -06:00
Jeff Hiner	cb2f5ef87a	Separate out sane blocking and async (nonblocking) implementations (#33) ... * Default impl should block. Renaming it to avoid API confusion from previous users. * Can't get rid of libc::read because we have MaybeUninit, but we can clean up the match * Remove extra include * Switch everything over to io::Error * Add initial tokio impl+example * Move evtest_tokio to normal example * Add documentation and clarify * Use a VecDeque (ring buffer) instead of repeatedly popping things off the front of a Vec * Looks like we are not using thiserror anymore; removing * Store read buf between calls * Add nonblocking example with epoll Co-authored-by: Jeff Hiner <jeff-hiner@users.noreply.github.com> Co-authored-by: Noah <33094578+coolreader18@users.noreply.github.com>	2021-03-01 10:09:11 -07:00