[aerogear-dev] cordova push plugin simplification

[Burr Sutter]

On Feb 28, 2014, at 5:33 AM, Matthias Wessendorf <matzew at apache.org> wrote:

> 
> 
> 
> On Thu, Feb 27, 2014 at 9:40 AM, Erik Jan de Wit <edewit at redhat.com> wrote:
> Hi,
> 
> I was still thinking about those 2 last points (last night while baking pancakes) it seems that we can’t make it optimal. So I would propose to do something like this:
> push.register(function(event) {
>   alert(event.alert);
> 
> }, 
> function(error) {
>   throw error;
> 
> }, {
>     pushServerURL: "<pushServerURL e.g http(s)//host:port/context >",
> 
>     alias: "<alias e.g. a username or an email address optional>",
>     android: {
> 
>       senderID: "<senderID e.g Google Project ID only for android>",
> 
>       variantID: "<variantID e.g. 1234456-234320>",
>       variantSecret: "<variantSecret e.g. 1234456-234320>",
> 
>     },
>     ios: {
>       variantID: "<variantID e.g. 1234456-234320>",
> 
>       variantSecret: "<variantSecret e.g. 1234456-234320>",
>     }
> 
> });
> Instead of leaving it to the user to find some sort of way to deal with the 2 variants we supply him with a way to configure both. On iOS the plugin will use the iOS part of the config and on android the android part plus a generic part.
> 
> 
> I like that android/iOS config object!
I like this better as well, verse having to put in a if then else block based on mobile OS, which is error prone.

Question about the JS syntax above - it looks odd to have the configuration object below the function(error), I assume I could have declared the config object above and used it as a parameter to push.register().   When JS gets too creatively away from old school (e.g. Java 1.5) syntax it tends to throw me off.  :-)

How else could the block above be written to look more "straightforward"?   keeping in mind that our average user is a Struts/SpringMVC kind of developer.

> 
> 
>  
> 
> And if you cordova app is only for one device maybe we could support this as well:
> push.register(function(event) {
>   alert(event.alert);
> 
> }, 
> function(error) {
>   throw error;
> 
> }, {
>     pushServerURL: "<pushServerURL e.g http(s)//host:port/context >",
> 
>     alias: "<alias e.g. a username or an email address optional>",
> 
>     senderID: "<senderID e.g Google Project ID only for android>",
>     variantID: "<variantID e.g. 1234456-234320>",
> 
>     variantSecret: "<variantSecret e.g. 1234456-234320>"
> });
> This will just take the variantID and secret that are on the ‘root’ level regardless of the device type, so it would only work on one. But maybe this will confuse people again.
> 
> perhaps :) 
>  
> 
> WDYT,
> 	Erik Jan
> 
>>>  
>>> Till now all these changes can be made on the plugin, but we could take it even further. Two things that are still platform dependent the senderId and the variantID/secret. Now senderId is ‘known’ by UPS so why do we need to specify it here? We could make senderId part of the response when registering a device on UPS then the client doesn’t need to specify it and all configuration is in one place.
>>> 
>>> It's optional on UPS, as only really Android clients need it. And I think originally Luke just added the 'senderID' to UPS for having something handy (copy/paste usage).
>>> 
>>> So, I am not sure on that one, as that means we now have to enforce Android Variants always get the senderID attribute. Could be done, yes, but I am not sure.
>>>  
>> 
>> Hmm, right it’s optional then this maybe not the greatest idea 
>> 
>>> 
>>> That leaves variantID/secret and that is the boldest proposal. How about we make it possible to register for an application instead for a specific variant? Then based on the meta information (deviceType, operatingSystem and osVersion) we choose the right variant.
>>> 
>>> -1
>>> 
>>> we should never 'deploy' the masterSecret and the applicationID on clients. If that information is stolen (easy on devices), you can start sending dirty assaults to ALL the devices for that application. Same w/ all the different providers like Parse, Stackmob etc: they all use a MasterKey/Password for the actual send. And they _highly_ recommend NOT applying that to the clients (it's easy to leak that information)
>>> 
>> 
>> Right didn’t think about that, this together with the objections that Sebastien has makes this a definite no go.
>> 
>> 
>> _______________________________________________
>> aerogear-dev mailing list
>> aerogear-dev at lists.jboss.org
>> https://lists.jboss.org/mailman/listinfo/aerogear-dev
> 
> 
> _______________________________________________
> 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
> _______________________________________________
> aerogear-dev mailing list
> aerogear-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/aerogear-dev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/aerogear-dev/attachments/20140228/12445af9/attachment.html