- From: Henri Sivonen <hsivonen@iki.fi>
- Date: Sat, 27 Aug 2005 22:03:32 +0300
- To: Lachlan Hunt <lachlan.hunt@lachy.id.au>
- Cc: www-validator@w3.org
On Aug 27, 2005, at 16:19, Lachlan Hunt wrote: > Henri Sivonen wrote: >> My validation service ( http://hsivonen.iki.fi/validator/ ) now >> includes an experimental and incomplete HTML5 parser ( >> http://hsivonen.iki.fi/validator-about/htmlparser.jar ) which does >> not do fixups like TagSoup. > > Can you please clearly explain what you mean by "does not do fixups > like TagSoup"? The design goal of TagSoup is that it "keeps on trucking" and never gives an error of any kind. The design goal of my parser is to report errors and not expend effort toward recovering from them. The classic example is: <i>foo<b>bar</i>baz</b> The example is not conforming HTML. TagSoup will emit parse events as if the document was well-formed. That *is* a fixup. My parser will report an error and stop. So by "fixups" I mean the features of TagSoup that reshape *non-conforming* HTML into well-formed XHTML. As for things that are *not* fixups: My parser is incomplete because it does not perform tag inference and it does not convert certain attribute values to lower case. These features are not fixup features. They are legitimate unambiguous HTML features that I intend to support in a future version. That is, <p>foo <p>bar</p> is legitimate and equivalent to <p>foo </p><p>bar</p> Supporting the omission of </p> is *not* a fixup. Likewise, <form method="GET" action="/"> is legitimate HTML, so lowercasing the value of the method attribute in conversion to XHTML is *not* a fixup. -- Henri Sivonen hsivonen@iki.fi http://hsivonen.iki.fi/
Received on Saturday, 27 August 2005 19:03:41 UTC