The greedy match was including whitespace if more than one space character
separated the "keyswitch" token and the first coordinate.
Signed-off-by: Michael Richters <gedankenexperimenter@gmail.com>
Even if a testcase shouldn't generate any HID reports, we should verify that it
didn't. If it does generate a report when there are none expected, that test
should fail.
Signed-off-by: Michael Richters <gedankenexperimenter@gmail.com>
This replaces the `CHECK_EXPECTED_REPORTS()` macro with a new `CheckReports()`
method. The new method verifies both the content and timestamps of expected
keyboard reports, and provides more output on failures (including details of any
unexpected reports).
Signed-off-by: Michael Richters <gedankenexperimenter@gmail.com>
class for each generated test file. removing
this makes test definitions simpler
Also, grow support for multiple tests within a script
Signed-off-by: Jesse Vincent <jesse@keyboard.io>
This change adds a set of functions to the `VirtualDeviceTest` class to make it
possible to write simpler testcases involving timed keyswitch press and release
events along with corresponding keyboard HID reports.
Other outputs (Consumer & System Control HID reports, LEDs, et cetera) are
not yet included.
Instead of ignoring the return value of RUN_ALL_TESTS, exit with the same status
code. This will make failing tests actually fail the build, instead of logging
the error, and then exiting successfully.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
This change adds the keycodes for keyboard modifier keys to the vector returned
by KeyboardReport::ActiveKeycodes(). Also, a simple method for getting the
modifiers byte from a keyboard hid report.
Signed-off-by: Michael Richters <gedankenexperimenter@gmail.com>
Because the Arduino build system doesn't -really- interface well with Makefiles, there's no other way to pick up Kaleidoscope or sketch changes from Make. This -does- slow down sim builds / test runs a bit.
Fixes#942
For now, this is just a wrapper around RunCycles(n), but the two could diverge
in the future. RunMillis(t) should always cause `t` milliseconds to elapse in
the simulator, regardless of how many cycles that takes.
Signed-off-by: Michael Richters <gedankenexperimenter@gmail.com>
Jesse Vincent
Michael Richters
Gergely Nagy
Gergely Nagy