Henri Sivonen On 09-09-22 10.30:
On Sep 22, 2009, at 10:31, Julian Reschke wrote:
The design principles also say "pave the cowpaths", yet HTML5
defines an entirely new syntax.
Another spec isn't a cowpath. Significant existing usage is. Microdata
has no prior usage. However, HTML+RDFa has very little existing usage
in the grand scheme of things. Moreover, to the extent syntax that
looks like RDFa in text/html is used already, it is processed in a way
that the draft doesn't describe:
http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2009Sep/0124.html
It seems to me that neither Microdata nor HTML+RDFa paves a cowpath.
(Specifying Microformats with well-defined authoring conformance
criteria and processing model would be paving the cowpath.)
The principle isn't called "pave the cowpaths". It is called
"consider cowpaths". Or perhaps we should say "consider cowpaths,
and pave them if possible". It is a principle that is about
looking at existing ways of doing things before inventing
completely new or paralell ways. It is also a principle that adds
weight to "the common way to do it", in competition with "the
ideal way to do it". There is no lower limit for when something
becomes a cowpath.
It would be possible to come to the conclusion that Microdata "has
to be invented" even after considering the existing solutions to
the same/similar problem. Then you could say that you had been
following the principle - to "consider cowpaths". But you couldn't
then say that you, as a result, were paving a cowpath.
It feels a little silly to say that a spec isn't cowpath - because
then it sounds as if you say that only un-academic, in-the-wild
solutions to problems should be considered cowpaths.
It also seems like the idea that we should investigate whether
deviations from a spec points to problems with the spec - or if
they even represent useful cowpaths, is a different thought -
although not completely unrelated - from the one that is expressed
in the "consider cowpaths" principle.
RDFa in XHTML seems like a natural cowpath to consider for HTML 5,
since we know that XHTML is, for the most part, interpreted as
HTML. From that angle, RDFa is already in use in text/html.
Since "consider cowpaths" involves looking at existing solutions,
it isn't unrelevant that something is common. However, whether it
is so common that it can be labeled "common" isn't the thing, in
itself, I think. The real thing is to consider the trouble one
would have - and the trouble _one creates_ - by introducing
something new, compared with the option to follow an existing path.
So, we should consider cowpaths because it is often useful to
follow cowpaths. But it isn't necessary to follow any of the
cowpaths in a particular field in order to claim that you followed
the "consider cowpaths" principle.
--
leif halvard silli
