On Mon Nov 28 11:27:20 2022 +0000, Bernhard Kölbl wrote:
Seems like I just marked some formats wrong so I can just eliminate the broken here, which should make it more simple.
Imho it would still be better to reverse the control flow here. The test control flow should follow what you expect first, then what you got if you want to support potential errors.
ie: If you test a bad format and expect an error, it should be `if (test.bad_fmt) ok(FAILED(notify.error), ...)`, rather than `if (FAILED(notify.error)) ok(test.bad_fmt, ...)`.