Polyglot markup and authors

Michael[tm] Smith, Tue, 22 Jan 2013 15:09:10 +0900:

Co-chairs: Does Mike and Henri’s[1] unwillingness to create a validator 
even matter? Can't we send the spec to RECOMMENDATION unless we have a 
validator?

Mike: In trying to understand what Henri said about authors, I see that 
he was afraid that A List Apart would write about polyglot. And clearly 
many authors read ALA. But I find only 2 references to Appendix C at 
ALA.[2] Henri seems to forget that it was XHTML 1.0 that spread the 
message that XHTML syntax is OK in text/html. Hey, W3.org uses XHTML 
syntax all over - to this day.

But as told in the Polyglot Recommendation Rationale page:[3] when 
XHTML 1.0 came along, all it brought was a new syntax. And thus, 
naturally the syntax caught attention. Even XML was new then. However, 
with HTML5, we have lots of real new features that take the attetion. 
And with HTML5, the difference and likenesses of XHTML and HTML are 
spelt out in the main spec - HTML5. Not least does HTML5 "bless" many 
XML-isms. Thus we are far from in the same situation. Even XML is not 
as hot as it once was.

Thus, this "history repeat" argument seems not very well founded.

> it's bad for authors because it introduces another perceived
> requirement that a lot of them are going to feel compelled to try to meet
> -- whether they actually understand what it even is or why it's there.

This is up to how you design the interface. Daniel offered a nice 
interface in BlueGriffon.[4] (BlueGriffon 1.6.1 has since then been 
released - with the polyglot option!)

Daniel's design reminds about the options that NU validator already 
has. E.g. NU validator’s image checking tool. As long as they are not 
the default option, then there should be no reason to have the fear 
that authors will be seen as a "perceived requirement". Or do you have 
other experiences with the current extra options?

> What I mean is that many authors are going to try the option, find that
> their documents fail to validate under it, and then feel the need to fiddle
> with documents that are otherwise fine just in order to fix "errors"
> reported by the Polyglot option.

This allows me to bring in NU validator's presets.

NU validator currently has 10 presets, of which 5 are HTML5/XHTMl5 
presets. The "extra option" design that Daniel has in Blue Griffon, 
would probably lead to a polyglot option in each of the five 
XHTML5/HTML5 presets.

But if you fear - or experience - that polyglot as an "extra option" 
fails to be perceived the right way, then you could just change the 
option into a new, preset number 11.

Justification: I doubt that NU validator has any users that try to run 
all the 10 presets on the same page.

> On top of that, there's the additional opportunity for more fun with the
> really negative social dynamic of markup pedants who like to run other
> people's documents through the validator so they can then point out to the
> authors, Haha, you think you know so much about markup, but your document
> doesn't even pass the W3C validator without errors.

See what I said above about presets. If someone does try to run all the 
current 10 presets, then I doubt that it causes the negative social 
dynamic that you describe below if someone discovers a page that 
doesn't pass all the 10 presets … Thus, an 11th polyglot preset, would 
not have these effects that you fear.

> I really don't think authors would be well-served by us adding the option,
> and certainly I don't think that we have a large number of authors who are
> crying out for us to add it.

I believe I hereby have counted your arguments about how polyglot 
validation would be bad for authors.

[1] http://lists.w3.org/Archives/Public/public-html/2013Jan/0059

[2] 
http://www.google.no/search?hl=en&q=%22appendix+c%22+site%3Aalistapart.com

[3] http://www.w3.org/html/wg/wiki/PolyglotRecommendationRationale

[4] http://lists.w3.org/Archives/Public/public-html/2013Jan/0127

-- 
leif halvard silli

Received on Monday, 28 January 2013 04:44:38 UTC