On 1/13/12 2:50 AM, Roman Rudenko wrote:
True. In its current form, beforeload is not very useful for partial processing. What if we had 'beforedownload' event specifically for resource fetching, and constructed stub elements to feed it as event.target when load is readahead-induced?
That still has the fundamental problem of serializing subresource processing on the "main" JS thread that page JS is running on. I believe that as hardware becomes more parallel and browsers strive to make use of that parallelism having such a requirement in a specification will be actively harmful.
Maybe the benefits outweigh the harm, but they'd have to be some pretty serious real benefits, not just hypothetical ones, if they're going to constrain the web platform in a major way like that. Hence me trying to understand the actual use cases here.
So, do we expect to need to block loads from _part_ of a document, in the long term?
-Boris
