W3C home > Mailing lists > Public > www-svg@w3.org > June 2010

Re: Minutes, SVG WG Brussels f2f day 4 (Monday)

From: Henri Sivonen <hsivonen@iki.fi>
Date: Tue, 1 Jun 2010 01:10:55 -0700 (PDT)
To: robert@ocallahan.org
Cc: www-svg@w3.org, Henri Sivonen <hsivonen@iki.fi>, Chris Lilley <chris@w3.org>, abarth@webkit.org, eric@webkit.org
Message-ID: <154205569.319007.1275379855045.JavaMail.root@cm-mail03.mozilla.org>
"Robert O'Callahan" <robert@ocallahan.org> wrote:

> > shepazu: MSIE considered novel syntax... HTMl5 is some way from
> being done
> > and the parsing algo has only 1 implementaitn thats incomplete
> > ... from W3C process perspective its far from done
> >
> I believe Henri Sivonen's implementation (which powers validator.nu,
> which is enabled by default in Firefox nightly builds, and which we 
> plan to ship as the HTML parser in the next Firefox release) is 
> quite complete.

It indeed is quite complete. If you run the Java version through the html5lib test suite, the tokenizer failures are all cases that test implementation details of html5lib itself. The two tree builder failures are spec violations, but they are deliberately unfixed violations for now and it might be worthwhile to change the spec. (See http://wiki.whatwg.org/wiki/ParserIssues#Possible_other_issues ) In any case, the spec violations around reconstructing formatting elements on space characters aren't relevant to SVG support.

I wonder what incompleteness Doug was referring to. Even though mozilla-central is lacking a couple of patches that the htmlparser repo has due to pending reviews, the problem with SVG </a> has been fixed on mozilla-central. The only SVG-related incompleteness that I can think of is the SVG load event, which hasn't landed yet, but you can get builds from http://ftp.mozilla.org/pub/mozilla.org/firefox/tryserver-builds/hsivonen@iki.fi-293982a921b7/

In any case, whatever incompleteness there might be is so small that it doesn't make any sense at all not to treat HTML5 as completely SVG-relevant as-is right now. You can see that http://wiki.whatwg.org/wiki/ParserIssues is pretty short. I realize that REC might be a bit further away, but it would be folly for the SVG WG to treat HTML5 as vaporware until it's "done" from the Process perspective.

> Adam Barth and Eric Seidel are working on an independent
> implementation for
> Webkit (I don't know if they're going to tackle the SVG support,
> though).

CCing Adam and Eric in case they can comment on what their plans for SVG support are.

Unfortunately, it seems that the SVG-in-text/html support in IE9 Platform Preview isn't HTML5-compliant. The nodeName for SVG nodes is wrong, getAttributeNS doesn't match xlink:href properly and the behavior of the HTML parts suggests the parser is a patched version of IE's old parser. (See http://hsivonen.iki.fi/test/moz/detect-html5-parser.html ) 

The Python html5lib is a non-browser independent implementation of the parsing algorithm.

> <AlexD> Are there any tests for HTMl5 parsing conformance?
> >
> Yes, here:
> http://gsnedders.html5.org/html5lib-tests/runner.html
> Those tests include tests of SVG-in-HTML5 (in test10.dat and
> test11.dat).

The upstream repository for the test suite is

Henri Sivonen
Received on Tuesday, 1 June 2010 08:11:29 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:29:43 UTC