OT: XML 1.1 (was Re: ZWJ&XML)

From: Mark Davis (mark.davis@icu-project.org)
Date: Wed Sep 13 2006 - 12:08:30 CDT

  • Next message: Debbie Garside: "RE: New RFC 4645-4647 (language tags)"

    On 9/13/06, John Cowan <cowan@ccil.org> wrote:
    > Mark Davis scripsit:
    > > As I recall, the problem with XML 1.1 adoption was that XML 1.1 was
    > > not fully backwards compatible with XML 1.0: there were XML 1.0
    > > documents that were not valid XML 1.1.
    > In the sense that "XML 1.0" names a countably infinite set of abstract
    > objects, true; in the sense that "XML 1.0" names a set
    > of texts physically fixed in a tangible medium, I venture to doubt it.
    > Specifically, I doubt that any Real World XML 1.0 documents contained
    > any instances of U+007F through U+009F not as character references.
    > In exactly the same sense, Unicode 2.0 was not backward compatible with
    > Unicode 1.1, a fact which does not seem to have seriously impeded its
    > adoption.

    That's not a good analogy. Unicode 1.1 was not well established at the
    time we decided to change Korean. And the UTC debated long and hard
    about whether to make the change, but in the end we did have all the
    major implementers on board to make the change AND move to 2.0.
    Moreover, there were compelling reasons in the rest of what was in 2.0
    for people to move to 2.0. So for that case, we broke backwards
    compatibility. It was an extremely painful decision, and we decided
    never to do it again.

    XML 1.0, on the other hand, was extremely well deployed at the time of
    1.1, and there weren't enough compelling new capabilities in 1.1 to
    outweigh having multiple implementations.

    > The issues with XML 1.1 were in fact political; I say no more.

    I will. I had a number of conversations with people having sizable,
    widely-deployed implementations who were ready to move to XML 1.1, but
    decided against it simply because of the compatibility issues. Had 1.1
    been backwards-compatible, I have no doubt that they would have simply
    upgraded their implementations.


    This archive was generated by hypermail 2.1.5 : Wed Sep 13 2006 - 12:12:37 CDT