Backwards compatibility (was: Re: Wwhich RFCs the new work would obsolete, vs update or leave alone)

[Simon Josefsson]

Harald Tveit Alvestrand <harald at alvestrand.no> writes:

>> I'm assuming IDNABIS won't be (fully) backwards compatible with
>> IDNA2003.  That is the impression I have gotten from all discussions so
>> far.
> The only incompatibility so far proposed is that some names valid
> under IDNA2003 will not be valid under IDNA200x, and vice versa. For
> all names that are valid under both proposals, I don't believe any
> change has been proposed.

When you upgrade the Unicode version, some strings that normalize to one
value under Unicode 3.2 NFKC will not normalize to the same value under
Unicode 5.0.  See:

http://www.gnu.org/software/libidn/manual/libidn.html#PR29-Functions
http://www.gnu.org/software/libidn/manual/libidn.html#PR29-discussion

I have also noted the discussion around ß.  If IDNABIS-ToASCII(ß) != ss
then another backwards incompatible change is made.

In my comment on the IDNABIS WG charter I suggested that the charter
should say that the IDNABIS output needs to be strongly backwards
compatible.  Both Lisa and Sam from the IESG disagreed.  My conclusion
is that the WG is entitled to make backwards incompatible changes, and
given the discussions so far that involves using Unicode > 3.2 NFKC and
how ß will be handled, the WG will likely also make backwards
incompatible changes.

I think a useful output from the WG would be to clarify which backwards
incompatible changes are being made.

/Simon