[webkit-dev] Merging the iOS port to svn.webkit.org (Re: WebCore and interlocking threads)

Maciej Stachowiak mjs at apple.com
Sun Feb 10 14:06:10 PST 2013


On Feb 10, 2013, at 10:48 AM, Adam Barth <abarth at webkit.org> wrote:

> 
> 
> Ok.  I'm sold.  I suspect we'll want to move the iOS port off of USE(WEB_THREAD), but I can see how it's better to do that work in trunk.

We definitely want to get off the web thread over time. There's two paths by which this will start happening in the short to medium term:
- We'll likely want to offer opt-in for clients to a lockless pure message-passing model (I can't share details or timelines but it's probably a good guess that it would involve WebKit2 abstractions)
- We'll want to pare away as much of the web-thread-related code as we can without breaking clients that haven't migrated.

Unfortunately there are public APIs on iOS that are hard to fulfill with a cleaner concurrency model so it will likely take a while to be fully rid of it.

>  I wonder if it would be worth using a name like USE(DEPRECATED_WEB_THREAD) to discourage others from adopting this execution model.


How about USE(LEGACY_WEB_THREAD) or USE(IOS_LEGACY_WEB_THREAD)? That would correctly convey that ports should not adopt it, without putting it on quite the same level as Deprecated classes that are further on the path to removal.

Any new port that wants to avoid blocking the UI with web content processing should definitely *not* use the Web Thread, and instead should use the WebKit2 or Chromium architectures (possibly including the threaded models offered by those).

(Also it should probably be an ENABLE, not a USE.)

> Thanks everyone for taking the time to discuss this issue.

Likewise.

Regards,
Maciej

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-dev/attachments/20130210/dd9a3247/attachment.html>


More information about the webkit-dev mailing list