Re: Polyglot: the final thread?

On Tuesday, March 12, 2013, Larry Masinter wrote:

>  I think polyglot is a general important technique for transitioning
> languages, interfaces and protocols, and the TAG could do most good by
> understanding and explaining it.
>

There is a group that is documenting it. What will the TAG add to this
discussion? Are you proposing work in this area?


> Whenever you need to transition a one-to-many system without a flag day, t
> you need polyglot or a variant to allow the transition.
>

This is, again, analogous to the comment hacks I've used in a past life to
embed valid XML in valid JS (parsing as JS when fed to a browser and XML
when fed to an XSLT processor). It may exist, and may have value for some
users, but its importance seems a question to be asked and answered with
data. What incentives are at play? How often do users do this in the wild?

A concrete way to answer this might be to scan, say, the top million web
pages on a periodic basis and attempt to determine the % that are polyglot
and note the direction of that value over time. A double-parsing
script/crawler would not be hard or expensive to write should someone want
to convince the current TAG membership that this is worth starting new work
on.


> *Larry*
> *--*
> *http://larry.masinter.net* <http://larry.masinter.net>
>
>
> Alex Russell <slightlyoff@google.com <javascript:_e({}, 'cvml',
> 'slightlyoff@google.com');>> wrote:
>
>  Hi everyone,
>
>  As promised, I spent time looking into the history of the current
> polyglot work, the TAG's involvement, and the options for all parties.
>
>  There is, I hope, uncontentious background. To recap what I've learned:
>
>
>    - *Polyglot exists in the wild.* It is possible to write documents
>    served as "text/html" that would parse both as HTML5 and as XML.
>    - *Nobody knows how popular it is.* The lack of signage, coupled with
>    default in-browser parsing as HTML means that few on any side of the debate
>    understand to what extent producers are creating this sort of content. It's
>    difficult to draw any conclusions about importance based on a lack of
>    information either way as a result.
>    - *Polyglot markup has little impact either on XML or HTML. *HTML is
>    tightly constrained by DOM coherence between HTML and XHTML DOM
>    serializations. This is a stronger constraint on the evolution of the
>    parsing algorithm than anything Polyglot has come up with or is likely to.
>    XML, as the stricter subset, is unbothered. There's a honeybadger meme in
>    there somewhere.
>    - *The HTML WG may product a Polyglot document with or without the
>    TAG's request. *The specifics of why the TAG decided to jump on
>    issuing a request are fuzzy, but it doesn't seem to matter. The TAG's
>    request (or absence) has no impact on process from here. Polyglot is inside
>    the HTML WG's charter and is proceeding towards publication. Maintaining or
>    rescinding the request will not change that. At some point in the future,
>    the Polyglot document will come up for a vote as to REC or NOTE. This will
>    not be affected *in any way* by the TAG.
>
> Now, as to what the TAG can and should do, I'll editorialize a bit;
> apologies in advance:
>
>
>    - The current and past TAG members do not agree that the outstanding
>    request speaks to any core architectural principle. That the HTML WG has
>    identified the subset and is describing it  may be good; but no better
>    perhaps than naming the comment escaping hacks that would let you nest XML
>    and JS in the same document (as I did for generating
>    Docbook documentation from my very first JS toolkit).
>    - The utility of polyglot is in dispute.
>    - There's worry that if sent to REC (with our without the TAG's
>    request), it will be seen as being being something the W3C *wants* authors
>    to do; not merely something that authors *can* do (or may happen
>    into). Some see the TAG's request as a vote in this direction. A community
>    of people find some value in the subset today, but there's very little data
>    to say that the architecture of the web will be bolstered by creating more
>    Polyglot-published content.
>    - There's no way to know besides double-parsing should that future
>    arrive and nobody is doing this over a large enough body of content to
>    determine if there is more or less polyglot content today vs. yesterday. It
>    does not appear this will change.
>
> As a result of all of the above, having (I hope) fairly weighed the
> arguments, I would like to recommend that we find a way to extricate
> ourself from the request. It doesn't matter to the future of Polyglot, and
> it does not, in my view, serve the TAG to be in the middle of this.
> Polyglot can have whatever future it will in the W3C without our group
> involvement.
>
>  Thoughts?
>
>  Regards
>