[whatwg] Reporting errors during Web Worker startup

[Ian Hickson]

On Wed, 9 Jan 2013, Kenneth Russell wrote:
>
> http://www.whatwg.org/specs/web-apps/current-work/multipage/workers.html#creating-workers 
> doesn't seem to define what happens if there aren't enough resources to 
> create a separate parallel execution environment.

User agents may impose implementation-specific limits on otherwise 
unconstrained inputs, e.g. to prevent denial of service attacks, to guard 
against running out of memory, or to work around platform-specific 
limitations.

-- http://whatwg.org/html/#hardwareLimitations


> Would it be legal for a UA to consider this as violating a policy 
> decision and throw SecurityError? Or is that step intended to reflect a 
> static decision, such as whether the UA allows workers to run at all?
> 
> If this behavior isn't specified, could some graceful failure mode be 
> specified? Currently some UAs terminate the execution of the page 
> attempting to start the worker, which is obviously undesirable.

We can't really define behaviour in hardware-constrained situations, 
because you might not have the resources to do what the spec requires. (In 
the most trivial case: the spec right now technically requires that when 
you don't have enough resources to start a new worker, that you start a 
new worker anyway. Obviously you can't do that.)

Implementors are encouraged to not have limits, but when you hit the 
limits, you just need to do what seems best.

My recommendation here would be to just queue the worker up and wait for 
there to be available resources. Or, even better, to run workers in a 
thread pool, and when you have more workers than threads, time share the 
threads between the workers.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'