[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: PaceMustBeWellFormed

On Wed, 07 Jul 2004 10:23:46 +0200, Danny Ayers <danny666@xxxxxxxxxxx> wrote:
> Generally looks very good, thanks, though I'm a little puzzled by the
> following line:
> Mark Pilgrim wrote:
> >Therefore, this proposal recommends that all Atom-enabled publishers
> >"assume the worst" and only emit ASCII-compatible XML.

I think I explained this better in the "Impacts" section.  Some
publishing programs (such as Movable Type) don't always know *and
can't even discover* whether the feeds they generate will be served as
"text/xml" or "application/xml"-equivalent.  Such programs MUST
generate ASCII-compatible output, otherwise they risk running afoul of
RFC 3023.

This is Martin's worst nightmare [1], that specs would start
recommending ASCII-compatible encoding and eventually everyone would
do that without knowing why.  And he's right, it's the old "cat tied
up during meditation practice" problem [2].  But I don't see any
alternative.  Programs like Movable Type are

(a) our early adopters
(b) publishing static files that are later served by the underlying web server
(c) able to run in extremely unprivileged environments where the end
user is unable to override the default server configuration
(d) generally run on default Apache installs where "text/xml" is still
the default (yes, I know this was changed late last year)

These people *can use Atom*, and they *will use Atom*, and Movable
Type *already supports Atom*.  This proposal spells out the steps they
MUST take to ensure that they are always doing it correctly, even in
unprivileged environments.

Atom is XML, generally served over HTTP.  That is ruled by the
well-formedness provisions and draconian error handling of the XML
specification, and RFC 3023.  I don't make the rules, and (apparently)
I can't change them either.  Failing that, the best I can do is
provide guidance so people can live within the rules.

[1] http://www.imc.org/atom-syntax/mail-archive/msg05928.html
[2] http://www.rider.edu/~suler/zenstory/ritualcat.html