[whatwg] Proposal for separating script downloads and execution
bzbarsky at MIT.EDU
Tue Feb 8 13:13:15 PST 2011
On 2/8/11 3:56 PM, Kyle Simpson wrote:
> Yes, it's important to note that it's not even the *execution* of
> the parsing of it (even if invoking of the functionality is deferred
> until later) that causes the biggest slowdown, in most cases.
Here's the thing. Parsing is a black-box behavior. Nothing says a
browser needs to do it right before execution, that it needs to happen
on the main thread, or that it's happening at all.
For example, for Gecko we have had proposals to parse off the main
thread, as well as to store parsed script data structures in the cache
so on a cache hit we just load the already-parsed data. We've also had
proposals to parse eagerly when priming the speculative load cache; this
could be a good bit before the script executes, assuming it ever executes.
So is control over parsing really just working around quality of
implementation issues? Or do you really need to constrain the timing of
the parse so as to preclude it being done on a background thread, while
the parser is waiting for a previous script to load?
> For the purposes of this discussion, we are combining (but safely so, I
> believe) "execution" and "parsing"
See, I'm not sure that combination is safe.
> The reason it's necessary to draw the distinction (and point out that parsing is the
> costly bit) is to defuse the argument that the script author can simply
> change the script to not execute itself until manually invoked at a
> later time.
Yes, but parsing is also the unobservable bit, which a browser therefore
is at liberty to optimize in various ways, even if none do so far.
More information about the whatwg