On Mon, May 7, 2012 at 1:58 AM, Tab Atkins Jr. <[email protected]> wrote: > On Mon, May 7, 2012 at 10:34 AM, Simon Pieters <[email protected]> wrote: >> On Fri, 04 May 2012 23:46:46 +0200, Ian Hickson <[email protected]> wrote: >>> What does it do in the case of: >>> >>> var frag = document.createDocumentFragment(); >>> frag.innerHTML = 'bla bla .. 1GB of text .. bla <caption> bla' ? >> >> When hitting non-whitespace text, it seems better to use "in body", I think. > > That's potentially usable, since having non-whitespace inside a table > is erroneous anyway. However, it's less compatible with the <ruby> > modes, as <ruby> can contain raw text. Is that okay?
As long as all tokenization happens the same way both when scanning for the first element, and when the tree building is happening, then it should matter if there is 1GB of content before the element. It would just need to be tokenized once anyway. / Jonas
