After the Sieve interop yesterday evening sort of crumbled over
encoded-char, I implemented the hex side of things for libSieve.
I also noticed that in 2038bis-12, 2.4.2.4, the sentence:
In the following script, message A is discarded, since the specified
test string is equivalent to "$$$".
...should refer to 'message B'.
It's about 50 lines, quick-and-dirty, so when I do Unicode I'll write a
better little state machine loop that should be half as long as able to
do both hex and unicode. It should port nearly verbatim to Cyrus IMAPd.
I do ignore hex 00, however, as I'm not 100% 8-bit clean (and Cyrus is
_really_ not 8-bit clean). I assume this needs to be resolved before we
have a 'working implementation' for standards track purposes?
Aaron