"brian m. carlson" [off-list ref] writes:
I talked to Carlo off-list because he was concerned about the potential
security implications of the patch, and as part of that discussion he
suggested an additional test for our UTF-8 handling. Since adding some
additional tests for that case seemed like a good idea, I squashed in a
patch of his to handle that case.
That's the only change between v4 and v5.
With the polarity flip of the option done in v4, the resulting code
is quite easy to follow. Thanks, will queue.