I am trying to figure out how to auto-reconnect a ejabberd socket connection. Like whatsapp, Facebookmessenger etc
For example, when the app is closed, and i receive a push notification to wake the app up and that calls to connect the connection, or if i put the app to the background for an hour, and the connection has been idle too long and get disconnected,then i put the app back to the foreground.
I can only think of 2 ways to automatically connect to the ejabberd server.
1.) To have the JID and Password stored(but this is not good for safety reason)
2.) To have the ejabberd idle connection set to never disconnect a connection(But this uses quite a lot of resources)
Is there any other possibility to automatically re-connect a user to the server?
You cannot stay connected all the time on mobile. This is the case today on iOS and is going to be the same on Android with Android Marshmallow.
So, you need to authenticate and store some form of credentials. It does not have to be a password. If you have a custom auth module in ejabberd for example, it can be a token.
Please, note that you can also store sensitive data encrypted on mobile. Both Apple and Google provide a keychain API that is design to protect the credential. It is not accessible in cleartext from a backup for example.
Related
We are developing a consumer hardware product. Each device is registered on a central webserver and the owner also have a user account to which the device is linked. The owner may also choose to share the device with other users.
Now, to solve the problem of getting through firewalls etc we are using XMPP: the user access his/her devices using an iOS/Android app. The app connects to the XMPP-server and so does the hardware devices. So the app can access the devices by sending custom XMPP stanzas.
Currently the device and the mobile app use the same JID, so the device will allow messages only from the same bare JID as itself uses. To allow for sharing devices we are planning to use the roster instead: the device will get its own JID ("hw381983829#thexmppserver.com") and will accept stanzas from all JID's in its roster.
The problem I'm having is that the users, devices and device-sharing data are stored on the webserver. I would like to use this same information on the XMPP-server: all users and devices on the webserver are allowed to login to XMPP and the roster of a device is the same as the users that may access it. This information can be accessed through a JSON API.
One way would be to mirror changes as they happen, but I don't like that idea since there are too many steps that could go wrong.
The best solution I can think of is to let the XMPP server use the JSON API instead of its builtin database. It would be read-only, but that is not a problem since all registration and sharing should be done on the webserver.
Any ideas on how to proceed? The functionality described above is more or less all that we need: we don't need S2S, offline messages, etc. We are currently using Ejabberd, but Prosody or Openfire are perhaps better alternatives?
For authentication, it looks like this ejabberd contribution does exactly what you need:
https://github.com/processone/ejabberd-contrib/tree/master/ejabberd_auth_http
For roster, it is easy to write a custom roster module that will be hitting your HTTP backend instead of query the database thanks to ejabberd API.
You can have a look at mod_roster as a guide to implement the methods: https://github.com/processone/ejabberd/blob/master/src/mod_roster.erl
I have an app server and several iOS apps that communicate with this app server over the internet. I'm trying to understand if there are any other frameworks, other than Apple Push notifications that allow me to send a message to a specific user of my app while the app is open?
A typical scenario: I type a message at a console on the app server and the message box pops up on a specific device.
One possible approach that I thought of is to periodically poll the server for messages for this particular device. Another approach would be to keep a socket open and stream messages over that socket from the server.
I have an ios app.
In which i have multi user account.So multiple user can Login.
When one user see other user profile and if that user is also online i want to notify him.
I can't get how to implement it.
If i have to continuously check some flag value on server if someone has checked their profile or how can i handle it?
Remote notifications is the way to do it. The logic on when and who has to be notified can be handled at the server.
You may have seen lots of apps asking for permissions to send notification. The downside being if the user denies that permission.
Polling from client side can also be one of the methods which can be used and but would require the client to do most of the computation.
So all in all, its a call you got to take considering what you are building.
We are developing an iPhone application that allow users to send messages to others via Apple Push Notification Service when the target user have installed our application or SMS when haven't.
We want to get notified immediately when our app is uninstalled so that we can decide how to send the message to the target user.
We find the APNS feedback server have a long time delay that doesn't agree with our requirement. So we use another way: when our server recorded the target user have installed our application, we send him message via APNS, if he haven't read the message in 30 minutes, we believe that he has uninstalled the application, so we send the message via SMS.
Is there any way better?
As Oleg said, there is no way to accurately detect if your app was uninstalled.
The APNS feedback service returns a timestamp and a push token for messages it was not able to deliver. Sometimes, this indicates an uninstallation but it can also simply indicate a user that was simply offline at that moment. The Feedback service does have some lag so can not be used for time-sensitive intel gathering.
Based on your requirements, I'd say you're doing it right.
One suggestion that may or may not work for you would be to include a link (via url handler) to your app when you resort to sending an SMS. If it makes sense for the user to return to your app, clicking on that link should launch the app and you'll have a trace on your server if you make a simple call. If, however, after sending the SMS the user is not detected as coming back into the app, chances are highly likely that the app was indeed uninstalled or that the user is offline for an unusually long amount of time which may require some other type of action on your part.
I’m looking at the designing and developing a service for Push notifications, and am trying to understand the order of processing in the Apple App when it comes to Registering for Push Notifications and storage of the Token.
What I’m developing currently is a service that allows subscription to individual changes that happen to particular processing on our server.
When they happen, the phone user may receive a message of type “A”, “B”, or “C”.
The phone user has to “subscribe” to each of these types individually if they want them. Otherwise they can ignore it and not subscribe.
So User Fred, on Phone #4, starts our App, logs in, and then has the ability to turn on or off subscriptions to events that are sent as Push Notifications.
So I need to tie details from Fred’s login, to Phone #4’s device Token, with particular subscriptions.
So my particularly important questions are.
When the Phone connects to the APNS server to get its device token, is this automatic on app start? Or can this be initiated at a later step? Ie, after going through a loging screen on our app.
Can we (are we allowed to) store the device token on the phone in the App's data store?. Or, should the App be connecting to the APNS server every time the app is run?
How does the App know if it as already called the APNS server and retrieved a token, or as above, should it call the APNS server time the app is run?
Can we (are we allowed to) store the token in the App’s memory as it runs, so we can properly subscribe and unsubscribe for particular messages?
We need also to be able to list all the subscriptions that a particular user may have across all their devices so the user can remove old devices (if they change phones). Or can we rely upon data back from calls to the APNS – when we attempt to push a notification - to inform us that a device token is no longer valid?
Or is there some better way of tying this all together?
When the Phone connects to the APNS server to get its device token, is
this automatic on app start? Or can this be initiated at a later step?
Ie, after logging into our app.
After the app has started, the app gets the token by calling registerForRemoteNotificationTypes. This will prompt the user for permission, and call a callback with the device token if permission is granted.
Can we (are we allowed to) store the device token on the phone in the
Apps data store?. Or, should the App be connecting to the APNS server
every time the app is run?
You'll need to build an APN provider, which is a web server that calls apple to send the pushes. The thing to do with the token is post it to your server that uses the APN provider. The app doesn't connect to APNS, your provider does, and it does it when it has pushes to send.
Can we (are we allowed to) store the token in the App’s memory as it
runs, so we can properly subscribe and unsubscribe for particular
messages?
You can keep the token on the client, but you don't really need to. It's your web service that calls APN, so it needs to be kept aware of your users' subscription prefs.
We need also to be able to list all the subscriptions that a
particular user may have across all their devices so the user can
remove old devices (if they change phones). Or can we rely upon data
back from calls to the APNS – when we attempt to push a notification -
to inform us that a device token is no longer valid?
APN also provides a feedback service that you call in batch which returns the device tokens that are no longer valid. Not only can you use this service, but you must. Apple will get mad at apps that repeatedly send to no longer valid devices.
Or is there some better way of tying this all together?
Yes! Parse.com provides a nice wrapper on the client code, does the provider and feedback service, abstracts the idea of single devices to the idea of a "channel" which sounds like just what you need for multiple notification types A, B, C, provides a super-easy step by step setup, and loads of other useful cloud services for iOS. (I'm not affiliated, but a big fan).