[palemieux]

Closes #566

[palemieux]

@dheitmueller This is based on d97ea9f, but fixes a longstanding oversight that the background color was not kept as a state variable (the root cause of the issue).

[dheitmueller]

@dheitmueller This is based on d97ea9f, but fixes a longstanding oversight that the background color was not kept as a state variable (the root cause of the issue).

@palemieux So this looks generally good. My only comment is that your test case has invalid parity for the byte pair, which a later patch I am submitting will cause it to skip that packet. The patch for the parity checking I plan to submit also goes through and fixes all the test cases that have incorrect parity, and I'll make sure your new test is part of that.

Separately, it's an interest corner case I'll see about adding it to my conformance suite in the next version.

[palemieux]

@dheitmueller Does f5a16ca fix it?

[dheitmueller]

@dheitmueller Does f5a16ca fix it?

Yes, this fixes it. With parity checking on, the word is no longer dropped and the red mid-row color is kept.