text/wiki vs dozens of text/vnd.*.wiki

[Mario Salzer]

(over-quoting for a cross-list-posting)

Linus Walleij wrote:
>
> On Mon, 23 Feb 2004, Mario Salzer wrote:
> > Now the question: Does it make any sense on attempting
> > standardization such an text/plain-alike "hypertext format"
> > (would become text/wiki) or should the over-registration of
> > MIME types (which only were minor variations of each other!)
> > just happen?
>
> For the first question: YES it makes sense! DO IT. The Internet needs a
> Wiki interchange format standard. There are two possible fora:
>
> 1. Workgroup at W3.org to get this done. (W3 people are at this list and
>    may tell their view on the subject). I *think* W3 would very much
>    recognize Wiki as their baby.
>
> 2. Workgroup at IETF to get this done, if W3 don't want it.
>
> Both approaches will cost incredible amounts of time and energy :-)
>
> We cannot stop people from registering vnd.xxx and prs.xxx types. However,
> from experience a very few people go through the (very simple) process of
> actually registering such MIME types anyway. Their interchange needs are
> not big enough.

There exists already a discussion group on the overall issue, but there
hasn't been so far any agreement on an interchange format, nor much
work on defining it. There were however a few research and format
consolidation efforts in the past.
(namely http://www.usemod.com/cgi-bin/mp.pl?WikiMarkupStandard)

However, if you say there _could_ be interest for it, then it makes
sense to reach for a working group (or at least an research group
would help a lot).


For the vnd.xxx MIME types, my main concern was, that it would
heavily enlarge the IANA registry, when that MIME type registration
would actually only be for identification purposes. (They were also
used for interchanging, but not primarily).

So if there are now only around 50 entries in the open text/* MIME type
registry today, there could likely be added 100 MIME types to it -
wouldn't this hurt the whole idea?

Also, in the face of a _potential_ standardization - was it even senseful
to register any "text/vnd.*.wiki" MIME types, if it was (hopefully)
only for a limited timeframe. Or should the WikiWorld simply stick with
an internal list / registry of "text/x.*.wiki" MIME types meanwhile?


Martin Duerst wrote:
>
> [This is just a personal opinion!]
>
> >[...]
>
> I'm not sure at all about it. I see the various Wiki conventions
> as a hack to get around the sad lack of tools for easy editing of
> Web pages and parts of Web pages. At the W3C, we use HTTP Put
> together with an editor that supports this. That's how the Web
> was designed, and that's how it should work. What's particular
> for a Wiki is that a) it's a page that is very explicitly open
> to a wide range of people to write to; b) it's usually not the
> overall page, but just the main area, that's written to.
>
> In my view, the right interchange format is something that already
> has a mime type, namely HTML. For the 'parts of Web pages' part,
> you can use some HTML subset, e.g. just everything that can be in
> a paragraph, or everthing that can be inside a dif,... There are
> technologies that allow you to specify the subset, such as DTDs,
> XML Schema, and RelaxNG.
>
> Having a 'styled text' forms control would also be great.
> In my personal opinion, it should be standard in XForms.

Interesting, that you mention. There has lately been a heated
discussion about using XHTML as interchange format. The conclusion
was, that most Wiki installations already export HTML and using
tools like 'tidy' those could be transformed into valid XHTML, what
indeed was necessary to embed some meta information pure HTML
did not allow.

For example links in Wiki often have some meta data associated,
so that it would often read
<a href="..." xwiki:href="urn:wiki:WardsWiki:WelcomeVisitors">
or even
<a href="..." xwiki:href="urn:isbn:1-3989-3219...">

But in the end this interchange format decision was also meant to
clean Wiki pages from the various extensions some implementations
offered (->information loss, but often acceptable). However, this
will make it as interchange format only for the major implementations,
because it is again more effort to backtransform HTML into wikisource,
even with XSLT.
A simple regex-engine could more easily transform one Wiki format
into another, especially if they only differed in a few points.
Hence my question for special Wiki MIME types.

Regards,
mario