—
Tadeas Kriz

On 23 Jul 2014, at 09:56 pm, Lucas Holmquist <lholmqui@redhat.com> wrote:


so there is this issue,  , about not being able to delete a simplePush installation.

using the UPS js client,  i registered the “deviceToken” with the push server, but when trying to use CURL or even the JS lib,  i was getting a 404, and i noticed that it would never actually hit the server( i was in debug mode in IntelliJ )

this was a demo that i had working back in april https://github.com/lholmquist/mobileweek/blob/master/mobileweek-ffos/js/app.js#L17

i noticed that when i registered the deviceToken, i was doing a encodeURIComponent before sending it to the UPS.

i modified the UPS JS client to do it this way, and then when i tried to do the remove,  i actually hit the breakpoints i set.  but the server could not find the installation doing 

Installation installation =
                clientInstallationService.findInstallationForVariantByDeviceToken(variant.getVariantID(), token);


I changed the JS client to only do 1 encode instead of 2 but it still couldn’t find the installation.

I’m not really familiar with JPA and all that,  so i don’t know how the query stuff works in this case


But i do think we should be encoding the deviceToken( which is the pushEndpointURL ) when registering with the UPS, or else i don’t think we can ever get access to that endpoint


I’d rather not require this. I mean, if there is no other way, then yes, let’s encode, but I can’t see the reason why would that be a problem. Basically, when you put it into the JSON when registering, it’s just a string and string can have any possible character, no matter what. Then the request. When it’s encoded one time, so it can be put into the URL, it should be enough for the server to use it and to search an installation in the database.

Is is possible that there are some Hibernate’s restrictions for the “deviceToken” field?


On Jul 14, 2014, at 9:52 AM, Lucas Holmquist <lholmqui@redhat.com> wrote:


On Jul 11, 2014, at 6:33 AM, Matthias Wessendorf <matzew@apache.org> wrote:

Hey Luke,

I created this JIRA:

As I plan working on AGPUSH-532 next week, to make sure we have the 'right' model in place for 1.0.0. I think once the server bits are merged, we should address AGPUSH-802 

kk


-Matthias


On Thu, Apr 10, 2014 at 5:22 PM, Lucas Holmquist <lholmqui@redhat.com> wrote:
so after testing this thing out with encodeURIComponent,  the delete does work,  however,  we need to "double up" the encode.

this works:



this doesn't:




On Apr 7, 2014, at 2:09 PM, Lucas Holmquist <lholmqui@redhat.com> wrote:


On Apr 7, 2014, at 12:52 PM, Matthias Wessendorf <matzew@apache.org> wrote:

Ok,

So as token, we use the endpointURL, and on client we perform  the encodeURIComponent() function ?

i think so( as suggested 2 months ago  :) )   i just need to see how this will affect how we store stuff in LS.




-Matthias

On Monday, April 7, 2014, Lucas Holmquist <lholmqui@redhat.com> wrote:
so i went back to look at what i had,


i don't think we need to get to complicated here,

reading the spec stuff, and this example


they show sending the pushEndpoint to the "App server",  so i think we could just use and keep it simple

it is also recommended that the channelID is never exposed to the application.


On Apr 1, 2014, at 3:34 PM, Lucas Holmquist <lholmqui@redhat.com> wrote:

i had something,  now i forgot what it was,   need to go back and check
On Apr 1, 2014, at 3:05 PM, Matthias Wessendorf <matzew@apache.org> wrote:


On Fri, Feb 14, 2014 at 2:06 PM, Lucas Holmquist <lholmqui@redhat.com> wrote:
still exploring

:-) any recent thoughts on 'encodeURIComponent()' ? 
 

On Feb 13, 2014, at 3:39 PM, Sebastien Blanc <scm.blanc@gmail.com> wrote:




On Wed, Feb 12, 2014 at 2:30 PM, Lucas Holmquist <lholmqui@redhat.com> wrote:
i might have a couple thoughts, but i need to try some things out first

Any update on that or does the solution proposed by Matzew (using  encodeURIComponent() ) could be enough ?

On Feb 12, 2014, at 3:53 AM, Sebastien Blanc <scm.blanc@gmail.com> wrote:




On Tue, Feb 11, 2014 at 7:15 PM, Sebastien Blanc <scm.blanc@gmail.com> wrote:
Ok,
I've been doing some tests by using the PushEndpoint as device token. For registration it works but I just faced an issue by trying to unregister because the URL for the DELETE looks like : 


And the REST endpoint get a bit crazy by the extra "/" present in the endpoint URL. Therefore, I think we must just use the last URL fragment as deviceToken.
 
Ok answering to myself ;) That won't work neither since if we do that UPS won't have the compllete push endpoint URL. 
So how do we deal with that ?


On Thu, Feb 6, 2014 at 3:27 PM, Matthias Wessendorf <matzew@apache.org> wrote:



On Thu, Feb 6, 2014 at 3:11 PM, Sebastien Blanc


-- 
Sent from Gmail Mobile
_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev

_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev


_______________________________________________
aerogear-dev mailing list
aerogear-dev@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@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev

_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev

_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev