- From: James Graham <james@hoppipolla.co.uk>
- Date: Wed, 20 Nov 2013 12:11:01 +0000
- To: Simon Pieters <simonp@opera.com>, whatwg@lists.whatwg.org
On 20/11/13 12:07, Simon Pieters wrote: > On Wed, 20 Nov 2013 12:30:18 +0100, James Graham > <james@hoppipolla.co.uk> wrote: > >> This seems like a nice proposal. There seems to be a minor problem >> that elements created through innerHTML will have the parser created >> flag set and so will not start loading until they are inserted into >> the document. So you probably want to call the flag the "delayed load" >> flag or somesuch, and only set it if the parser isn't in the fragment >> case. > > Yeah, indeed, thanks. > > A separate case I was thinking about is more than one <img>s in a > <picture>, do we want both to work or just the first? The proposal right > now would do both. If we want only the first, that means the selection > algorithm needs to check that there are no previous img siblings. When > an img is inserted to a picture so it becomes the first img, we need to > rerun the selection algorithm on the next img sibling (i.e. the img > element that was previously the first). Similarly when an img element is > removed, the (new) first img child needs to run the selection algorithm. > Although it involves more checks, I think it seems saner to have only > the first img use the <source>s. > I'm not sure that the extra checks buy you much apart from implementation complexity. What are you trying to protect against?
Received on Wednesday, 20 November 2013 12:11:37 UTC