On Oct 09, 2007, at 15:39, Jon Ferraiolo wrote:
> First, should there be a manifest file that lists the
contents of
> the ZIP package? Let's assume the answer is yes. Then,
you probably
> want to have an attribute such as 'type' that defines
the MIME type
> of each particular file such as you describe below. My
personal
> opinion is that it is always better to KISS, which
translates into
> standards groups holding a ruthless sword that battles
against
> feature creep of non-critical features, and in my mind
manifest
> files are non-critical in many common workflows because
the ZIP
> file itself has a directory and because the most common
payload
> format, HTML, works on the Web very well without
manifest files. My
> pragmatic proposal would be to define the file format
for the
> manifest and define a standard location and name for
the manifest
> within the package, but make it optional. In other
words, manifests
> are not required but if present must conform to the
spec. One more
> thing. If possible, to accelerate time to market and
industry
> adopton, I would push the manifest file out of version
1.0, but
> have the 1.0 specification state reserve a future
location for the
> manifest file.
I'm with Jon on this one; I think it's a useful feature in
some
circumstances but I'm not convinced that it makes the 80/20
mark. I'd
be happier with a promise to push a v2.0 out quickly with
that
feature included (and supporting content encoding as well as
type).
--
Robin Berjon - http://berjon.com/
------------------------------------------------------------
------------
"Technology is a word that describes something that
doesn't work yet."
-- Douglas Adams
|