[aerogear-dev] SimplePush Polyfill Changes

Matthias Wessendorf matzew at apache.org
Fri Jul 26 10:25:31 EDT 2013


On Fri, Jul 26, 2013 at 4:15 PM, Kris Borchers <kris.borchers at gmail.com>wrote:

> Hey all,
>
> Before I make a change to this polyfill I wanted to make sure there were
> no concerns or objections. Basically, what I want to do is change the usage
> so that calls to navigator.push.register() or
> navigator.push.registrations() (not yet added) would have to be done after
> a connection is established.


+1 that makes perfectly sense - the users will be 'notified' of an
established connection, right ?



> Currently, a user can call either of those methods at any time which
> causes headaches when trying to account for registrations that happen
> before the connection and also register previously used channels stored in
> localStorage. Just as an FYI, this is not an issue for Mozilla because they
> already know they have a connection when their JS executes because the
> browser creates the connection during startup.


yeah - and in case of FFOS, the connection to their SP server is already
there. They serve all SP apps over one WebSocket connection (similar to
APNs and Google's persistent connection).



> I have done a LOT of work to try to account for race conditions which may
> occur and in the process added a LOT of extra code.


I hear you :-)


> And even with all of that work, I'm still not 100% sure that all race
> conditions have been accounted for once this code is used at scale with
> lots of connec!
>  tions.
>
> So what am I suggesting? I would like to require our users to create a
> connection to the SimplePush server and do their interactions with that
> server in the success (or onConnect or what ever we want to call it)
> callback.



+9001


> After that, all code would be identical to the Mozilla APIs.


that is the key. We are a slightly different environment, so setting these
'requirements' is more than reasonable!



> Then, we could just write a brief guide or tutorial on transitioning code
> between our implementation and Mozilla's to highlight the small difference.
>

yes - makes all perfectly sense


>
> Thoughts?
> _______________________________________________
> aerogear-dev mailing list
> aerogear-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/aerogear-dev
>



-- 
Matthias Wessendorf

blog: http://matthiaswessendorf.wordpress.com/
sessions: http://www.slideshare.net/mwessendorf
twitter: http://twitter.com/mwessendorf
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/aerogear-dev/attachments/20130726/2134ff1f/attachment-0001.html 


More information about the aerogear-dev mailing list