[whatwg] WebWorker questions
aa at google.com
Tue Aug 12 11:26:07 PDT 2008
On Tue, Aug 12, 2008 at 4:50 AM, Shannon <shannon at arc.net.au> wrote:
> If a WebWorker object is assigned to local variable inside a complex script
> then it cannot be seen or stopped by the calling page. Should the
> specification offer document.workers or getAllWorkers() as a means to
> iterate over all workers regardless of where they were created?
I don't follow the first phrase here, but I can still answer the
There is currently no API to stop a worker from the outside, only from
the inside. Providing an API to kill a worker from the outside is a
little weird because we also want to be able to share workers between
multiple pages. If we did both of these things then one page could
kill a worker that other pages are relying on. Not saying this is
unreasonable, just something to think about.
As for getAllWorkers(), yes, I think this would be cool (perhaps for a
future version of the API). If for no other reason than that there are
interesting applications for diagnostics and introspection.
> Is it wise to give a web application more processing power than a single CPU
> core (or HT thread) can provide? What stops a web page hogging ALL cores
> (deliberately or not) and leaving no resources for the UI mouse or key
> actions required to close the page? (This is not a contrived example, I have
> seen both Internet Explorer on Win32 and Flash on Linux consume 100% CPU on
> several occasions). I know it's a "vendor issue" but should the spec at
> least recommend UAs leave the last CPU/core free for OS tasks?
You are right that workers and other things that are controllable by
authors but use threads should be pooled and limited. It shouldn't be
possible to DOS a browser using workers. Currently, Gears doesn't do
this, but we definitely should. I agree the spec should have a warning
to this effect.
> more client processing power than a single P4 core?
It's easy to imagine examples, but of course none exist today since
this capability doesn't exist in browsers.
100ms is generally considered to be absolute maximum amount of time a
UI can take to update after direct interaction. After that, users
think it is hung. It's not that hard to imagine applications for JS
computation which surpass this amount of time, particularly when you
consider that the cpu(s) might be busy doing something else!
But I think that computation is the least interesting of the benefits
that workers provide. The others ones are:
a) Workers allow applications to perform blocking IO. Currently this
isn't possible because it would block the UI too. In the current
database proposal, every single SQL statement is asynchronous.
b) Workers allow you to share parts of your application between
windows. So for example, if you open Google Docs in two windows, they
can't share any code today. Google docs is a pretty significant JS
app, so this is already crazy inefficient. But if the user has Google
Docs Offline enabled, then both copies of the app are also both trying
to synchronize to the same local database. The only way to get around
this is to implement a watchdog system where each window tries to
detect each other through updates to the shared database. Shared
workers allow you to solve this in a more natural way. Just put the
synchronization code in a shared worker. All the instances of the app
can access it.
> Shouldn't an application that requires so much grunt really be written in
> Java or C as an applet, plug-in or standalone application?
No. Web applications have many benefits over native apps. We should
try to increase the number of apps that can be written on the web.
> If an application did require that much computation isn't it also likely to
> need a more efficient inter-"thread" messaging protocol than passing Unicode
> strings through MessagePorts? At the very least wouldn't it usually require
> the passing of binary data, complex objects or arrays between workers
> without the additional overhead of a string encode/decode?
We (Google) have proposed the addition of being able to pass complex
JS objects without manually encoding/decoding, and I think that this
will be added eventually.
But even with just strings passing, it is easy to imagine applications
where there is a large amount of background work that needs to be done
given a small amount of input. For example, "download all the
documents for user X and store them in the database".
> itself, or a matter of current interpreters being non-threadsafe?
Threads as they exist in java and c are widely understood to be difficult:
I don't have a problem with coroutines personally, but I think that
there is still a place for full processes in any programming
environment. Sometimes it's just natural to break up an application
into UI and background processes. Unix applications have traditionally
been built with this model, and modern windows applications also use
> The draft spec says "protected" workers are allowed to live for a
> "user-agent-defined amount of time" after a page or browser is closed. I'm
> not really sure what possible value this could have since as an author we
> won't know whether the UA allows _any_ time and if so whether that time will
> be enough to complete our cleanup (given a vast discrepancy in
> operations-per-second across UAs and client PCs). If our cleanup can be
> arbitrarily cancelled then isn't it likely that we might actually leave the
> client or server in a worse state than if we hadn't tried at all? Won't this
> cause difficult-to-trace sporadic bugs caused by browser differences in what
> could be a rare event (a close during operation Y instead of during X)?
I agree that workers should be immediately killed on page unload, just
like XHRs are.
> I just don't see any common cases where you'd _need_ multiple OS threads but
> API, and MessagePorts' limited IO. The only things I can think of are new
> user annoyances (like delaying browser shutdown and hogging the CPU). Sure
> UA's might let us disable these things but then some pages won't work. The
> Working Draft lists a few examples, most of which appear to use non-blocking
> network IO and callbacks anyway. Other examples rely on the ability for
> workers to outlive the lifetime of the calling page (which is pretty
> contentious). The one remaining example is a contrived mathematical
> exercise. Is the scientific world really crying out for complex theorems to
> be solved in web browsers? What real-world use cases is WebWorkers supposed
> to solve?
I hope I've addressed some of these questions above. I'm sure you'll
let me know if you're still unconvinced.
More information about the whatwg