Related
I am using mail2web free service to have my mail pushed to my WM5 device, but I don't like the idea of forwarding my private email to a company I don't know and I don't trust.
Normally I have my email at a home mail server running linux, and I was wondering if there's any open/free solution compatible with directpush technology to replace the need for an exchange 2003 server.
If it does not exist (i have been searching and couldn't find it), what is preventing someone of writing a free replacement? patent issues? authentication issues? I haven't sniffed the protocol yet, but I think it's impossible that no one has thought about it before... wouldn't you like to have your home imap server pushing your inbox to your PocketPC phone? If the protocol is not very obscure it should be easy to write a daemon that does it...
any thoughts or experiences?
Kerio Mail Server has the push compatibility.
Not free though, but it is an alternative to Exchange.
Their latest beta supports it. I reported a bug to them re: contacts getting corrupted but they say they fixed that in the latest build.
Can't retest now since that server is at another site I won't be visiting for a while.
Hmm the directpush "protocol" is pretty trivial it shouldn't be impossible to implement an open alternative.
I installed the latest Kerio MailServer yesterday. It works great!
Especially nice is that it runs on XP as well as Linux. Too bad about the price!
pof, have you had the chance to sniff the protocol?
not yet ivorh, been too busy with many other things, but this is still on the queue
sniffed
Heh, got bored so I setup an exchange account and sniffed the packets.
I'll go through them and post some details when I get a chance.
great ivorh!
Can you attach a capture in pcap format?
update and thoughts
Pof,
Hmm, ok a bit more thinking and digging and I'm not sure implementing an open alternative directly is that useful. Let me explain why:
The direct push only works with the outlook exchange active sync. When the device gets a "direct push" byte, it triggers a sync with exchange - the functionality is tightly bound together, and as far as I can tell you cant dip in and get it to do something else. So to get it working you would need a server providing the exchange http interface. This wouldn't be impossible but would need a lot of effort for little benefit.
I took a look at the open-exchange but that doesn't seem like an ideal solution since it would require a completely different server installed rather than IMAP or POP and as far as I can tell the Outlook connector isn't one of the open-source components anyway.
Now what I'm currently thinking would be a neat workaround would be to implement a custom "direct-push" to basically do exactly the same, have a client app on the device open an http connection to a server running, er, "OpenPush" if you like... use exactly the same technique of a keep-alive connection and occasional heartbeats, but on a message notification on the client get the client to trigger an IMAP pull.
Now this is where I need some advice.... I haven't done any Windows Mobile development yet, so can anyone tell me what sort of API is available to the messaging app? Can you/how do you trigger a mail pull? (oh someone please tell me it's not the same horrible old MAPI interface??).
PS I've just been capturing the data using a simple http proxy actually, I'll make some samples and upload them with descriptions.
Cheers,
Ivor.
http://www.ivor.it
Hi pof,
try funambol. It was formerly known as Sync4j. I once found it when I searched a complete sync solution that I could implement in the mailserver of my company. We are using kolab so I only tried the old Sync4j cause there is a kolab connector available for v2.3.
v3.0 implements (real) Push-Mail. Microsoft Active Push works with a http connection that is opened by the client. Funambol Push-Mail connects to a port the client opens. I didn't want to test any further cause I'm using a Wapflat and thus only get an internal IP and have to use a proxy.
http://www.funambol.com/opensource/
Perhaps this is what you are searching for
What I forgot to say: For funambol you install a java program on your phone wich will insert the received mails in your Pocket Outlook
Yay!
Pof,
Ha! Ok I've whipped up a version 0.0.0 of OpenPush. and it works rather nicely!
Basically it consists of two parts one is an app that runs on the mobile and operates in the same manner as DirectPush. It opens up a socket connection to the server and waits for a notification byte. If it recieves a byte it kicks off a mail retrieval.
The other is a daemon that runs on the server and watches for a change in the users mailbox if it changes (i.e. a mail has arrived) then it pops a byte down the socket.
It just needs finishing now...
Currently the daemon is just an app that listens on a dedicated socket. I plan on turning it into a mod_perl module and using http keep-alive in the same fashion as directpush.
Regards,
Ivor
ivorh, that sounds cool!!
I had a look at funambol but seems too 'bloated' for my needs, I think your OpenPush will be more tight to what I was looking for, so if you want a beta tester just send a link to it
Is there any specific server configuration? I am running a Gentoo server with courier-imap, sendmail and apache2.
Any chance you could extend the daemon so other programs can tie into your new-item notification? It would be cool to write a program that can keep files synchronized over-the-air with a desktop machine using push sychronization.
Pof,
Yeah it's just a prototype at the moment. So I need to write it properly next. At the moment the requirements are simply "perl". It's independent of mail system, it simply monitors a directory/file you give it for changes.
I'm going to write it to be a mod_perl module for various reasons, so the requirements will be just apache and mod_perl.
I'll hack some more tonight and try and get a 0.0.1 ready.
aatreya,
Well I'll keep it simple for now and just doing one job well.
Sound interesting to me.
I am new to this. Pardon me for some questions.
Do I need to buy a server at home?
This server can be any OS?
I am really happy to see some people trying to do just about the same I intended to. Today I started experimenting with Open-Xchange. I also have a Gentoo server that provides an ebuild for Open-Xchange, but after a bit of browsing it just seems like the thing MS Exchange and OX have in common is the similarity of their names. So installing OX and doing all the Exchange stuff with it does not seem to be an option.
So I looked into Funambol - sort of again. About a year ago I already tried to get Sync4J running to sync my SyncML phone. I did not succeed, but that just makes me eager to try it harder this time.
I also have to use a Proxy-Server for my GPRS connection but I want to have it working via WLAN, too, hence without proxy and the whole tunneling disco.
PS: Ivor, I am glad to meet You once again. When our roads crossed the last time, You just figured out how to get the CLE266 MPEG2 stuff working, respect!
A very rough pre-alpha version should be ready this weekend. I can only apologise in advance for the quality of my WM5 app!
But it's "working for me".
CWKJ,
At the moment the "server" is a simply perl app that watches for changes to a directory or file so its pretty portable. I run it on linux since that's what my mailserver is on.
As for needing a server at home... well its entirely up to you, really you want the server running wherever your mail is retrieved from.
If the app/system gets a bit more polished and advanced you might even find independent ISP's willing to add it as a service.
rabinath,
Heh! Small world.
I'd also like to be considered for Beta testing when available,
I run Ubuntu 5.10 Server at home, hosting 5 domains for myself, and would Love to not have to forward them through mail2web anymore as I don't like to reply because it will go through Mail2Web. I know I can create a separate "account" in Pocket Outlook but you can only have 5. I need more than that...
with direct push over the o2 wap proxy all https conections are closed after 2 min. this is becausse all 2 minutes a new sync is needed. This eats much battery. So why is it nnot possible to mak e a ppc client that just sends the current ip adsress to the client on the server. and the server just push the email to the known ip adress. The client on ppc just has to send a new ip in case it changes. this wouldt be much better for battery life.
Thats the reason Exchange-Activesync works the way that it does. The server sends out a text message that is handled at the system level on the PPC, and this is basically an instruction to sync with the server. No unneeded traffic just to check if there is anything new.
Most Celllar service providers are using NAT technology so reporting your IP address to a server and telling it to make a connection to that IP, would just be telling the server to connect to the "gateway" back into the Cellular network. A text message sent to what is called the "SMTP Gateway" for each service provider will get to the phone no matter if the phones IP changes.
Hello everyone,
I come to you for some advice about an application I'm developing for Android 1.6 and higher.
Specifically, this application must be able to query a remote MySQL database and display the retrieved data. It need also be able to UPDATE data on this databse and that is where things get complicated ...
For selection queries, I wrote a small PHP script on the remote server that retrieves the desired data in the database and encodes them in JSON. From my Android application, I have no problem to retrieve, process and display them.
Now I wonder about the solution (if there is one ...) to run UPDATE from my application on the remote database. For now, I see only two solutions:
- Develop a webService on the server with an update method. Several problems with that: it seems webServices on Android are not easy (not native, have to go through external libs). Moreover, as it's UPDATE query, I want to make sure nobody else can use this webService.
- Alternatively, write a PHP script that makes the UPDATE and call it via HTTP. Same problem as webService: I want to secure this call to be sure that only I can use this script. These scripts can be integrated into an application already present on my server and which is secured by login / password. So, if it is possible to manage PHP sessions from Android, it's won! but Ican't find anything about it ...
What do you think? Ideas?
Thx!
These apps allow you to remotely access your phone from a web browser. However, they all run a web server on the phone, and I cannot connect to any of the over 3g (Verizon).
LazyDroid Web Desktop
Remote Desktop
Remote Web Desktop
I want to move the web server off phone, and (hopefully) onto private sites.google.com site. App Engine might be necessary, but I'm hoping this could be done solely in JS.
The hosting site would provide the UI, and interact with the phone using C2DM (the magic that powers Chrome2Phone, GMail, and installing apps from the web Market).
The UI is pretty obvious. It just needs a whiz to create HTML, Javascript, etc.
The C2DM backend is a still a bit mystifying to me... and searching for c2dm and javascript does not yield any obvious working implementations. But it seems plausible. Push a command to the phone, phone returns/uploads data to website, and UI updates.
Then there is the Android end. Well, there are the 3 projects above, Tasker for a quasi-hackish approach, and RPC (promising, but it seems like a WIP).
Thoughts? Volunteers? Geniuses?
Ooo... 2 birds with one stone!
This would also kill 2 birds with one stone.
No more typing in dynamic IP addresses! You get to use DNS to handle the connections. Bookmark your site in your desktop browser (it is always the same!). And set a preference in the Android app.
On lazydroid i've in planning some kind of trick that will let you connect behind firewall ... similar to a vpn...
CloudsITA said:
On lazydroid i've in planning some kind of trick that will let you connect behind firewall ... similar to a vpn...
Click to expand...
Click to collapse
I tried it again last week, and it is still unsuccessful. Webkey is currently the only application that I can successfully use to reach my phone.
Now, I could be wrong, but I believe all of these apps run a web server on the phone. I get a lovely, private 10.x.x.x IP address, which I can't reverse the route to. I have tried and failed to get DynDNS to work.
I have been looking into a solution since my original post. I have not had any time to do code squat, but I have loosely figured out all of the parts.
The big architectural difference I have been seeking is removing the server from the phone. I am not an Android expert, but I don't believe it even requires a running service. (Thank you, C2DM.)
With the app-webservice separation, you can work a "protocol" that reduces the overall bandwidth used... and thus improve battery life. Put all the "hard work" on a webserver, and (things get fuzzy here) possibly push it off onto the client browser (JS).
C2DM Browser Links
I could probably make something like WebKey but with C2DM and some more features. If you want you can give suggestions and I'll start making on saturday (after my exams). It would probably be possible in javascript for the actual sending from server and php just for logging in to your google account. The phone would just be registered on the server and no services (just as you wanted )
nebkat said:
I could probably make something like WebKey but with C2DM and some more features. If you want you can give suggestions and I'll start making on saturday (after my exams). It would probably be possible in javascript for the actual sending from server and php just for logging in to your google account. The phone would just be registered on the server and no services (just as you wanted )
Click to expand...
Click to collapse
I am not sure "more features" is necessarily the direction I'm headed. I am focused on making a "seamless" experience (i.e. less separation of phone and computer).
I was headed to App Engine (Python bias + easy Google integration). I have a project created. I haven't pulled together the various examples to make the core, but it seems <naive>simple</naive>. Stir in some templates, CSS, a sprinkling of JS, and voila!
The big "tricky" part that I can't convert from f***ing magic to a clear approach is the data link in the server. I want to avoid any storage to a Google disk, or otherwise, even temporarily. No stored data = easy privacy policy.
nebkat, if you're really chomping at the bit to code, here's my Android client concept.
- C2DM is a wake-up call. (cheat an borrow ChromeToPhone's ID to begin with)
- Connect to web server, send "I'm here," and wait for further instructions (Channels API/Comet/AJAX/.........)
- make the command set extensible
- each command is blockable in the client. (Permission control is set on the phone, not remotely.)
- After N minutes of no activity, send a "good bye," disconnect from the server, and fade into the background.
Don't worry, I'm very experienced with the server side stuff and I know exactly what you want. The only information stored on the sever side would be google account, the device c2dm registration id and some logging features just for statistics. A password could be set on the phone that would be sha512 hashed on the ajax request and would be sent to the phone. Even if a hacker found the hash, it would be useless without being logged in to the persons google account or knowing the server side auth token.
For now i'll just make the reciever, processor and command output and later on the extra security and ui stuff. It will work exactly the same way as Chrome2Phone except it will have server side php and the different commands. The connection from pc to phone will be something like this.
user command -> ajax request -> php c2dm request -> phone
phone -> php server http request -> controller page status
BTW I'm saving up for a Nexus S, how much would people pay for this type of app? There would definitely be a free version, but I just need to get the Nexus S because I have a Galaxy Spica now and it isn't the best for app development. I'm new to how stuff at xda works, would a donate version get me enoguh for the Nexus?
nebkat said:
Don't worry, ... <snip> ... auth token.
Click to expand...
Click to collapse
Alrighty then. I'm feeling like I can stop contemplating implementing this.
BTW I'm saving up for a Nexus S, how much would people pay for this type of app? There would definitely be a free version, but I just need to get the Nexus S because I have a Galaxy Spica now and it isn't the best for app development. I'm new to how stuff at xda works, would a donate version get me enoguh for the Nexus?
Click to expand...
Click to collapse
Since I was learning the ins and outs of App Engine, I read their quota rules and realized if this were popular it would require funding. I don't know where you are going to your web server, but I assume you'll have to pay someone to keep it running. But I had thought about $$$ already.
"Give away the razor, and sell them the blades."
Make the app free, no feature restrictions.
You get your money through various "membership" levels on the server. (See the account levels at fastmail.fm for an example.) So, you can use the app for free, but you only get, say, 2-3 MB of traffic per day, and only X sessions per day. Need more? See the pricing chart.
user command -> ajax request -> php c2dm request -> phone
phone -> php server http request -> controller page status
Click to expand...
Click to collapse
user command -> php server http request -> phone
phone -> php server http request -> controller page status
user command -> php server http request -> phone
lather, rinse, repeat.
C2DM is not deterministic, and acts up in low signal conditions. So, I made a decision to only use C2DM to initiate a session. Once both ends are connected to the server, everything goes over HTTP.
Oh.... and not that we need another Lookout/Phone Finder, but a shared-secret SMS code for the case where "they" have shut down the data connection.
I have my own server nebkat.com and there is nothing on it anyway.
The only other way to make "push" requests to the phone is with WebSockets. It would probably be better than c2dm because we have full control over what gets sent (google limits some requests). The advantage of WebSockets is that they send no header information which means that we could send our messages in 20 to 30 bytes.
I'll look into more detail on friday.
With web sockets won't you need to ensure the phone has a routable, external IP address? I know, for one, t-mobile does not expose an external IP address for their phones. Unless, of course, if the phone is connected over WiFi. C2DM works great for me (I have used a couple of apps with it and it is really useful).
MrGibbage said:
With web sockets won't you need to ensure the phone has a routable, external IP address? I know, for one, t-mobile does not expose an external IP address for their phones. Unless, of course, if the phone is connected over WiFi. C2DM works great for me (I have used a couple of apps with it and it is really useful).
Click to expand...
Click to collapse
No, WS is server initiated and the ip address' shouldn't make a difference.
MrGibbage said:
With web sockets won't you need to ensure the phone has a routable, external IP address? I know, for one, t-mobile does not expose an external IP address for their phones. Unless, of course, if the phone is connected over WiFi. C2DM works great for me (I have used a couple of apps with it and it is really useful).
Click to expand...
Click to collapse
You need an valid external IP address if you are attempting to initiate contact with your phone, which is why the 4-5 apps I've mentioned do not work on carriers like t-mo and verizon.
But the phone can establish a connection, and the carrier NATs (or whatever) will handle the routing for outgoing and incoming data.
I think the right questions are: Will Verizon/T-Mo allow the ports and protocol for WebSockets? Do Android and desktop browsers implement the draft API correctly and consistently?
I like C2DM. I works well when you have a good connection. But there are 3 issues with it.
1) The message size limit is 1024 bytes. Not ideal for file transfers.
2) In a poor signal areas, since the service retries sending messages, you will get delayed and/or duplicate messages. I work in a large "concrete" building, so I get this behavior often enough that I don't want to rely on it.
3) I believe there is a limit on the number of messages you can send. So, hunting around the filesystem could hit this limit (but unlikely in reality... I hope.)
It would be interesting to see exactly how those apps handle all of the data. Do they only use C2DM, or do they hand over to another protocol?
Ok my exams are over and I am starting with it. I'll give updates on this thread
Hi!
Anyone else bothered by sharing their personal info on social network to the third party? I had thought about developing open source distributed social network app based on https server running on android phones. Each member should have android device as a server of his/her profile.
0) - https protocol could provide possibility to connect to your social network using web browser.
1) - One may be able to deal with extensive upload and download or lost connection issues by allowing distributed profile updates sharing:
I suggest following scenario:
D updates profile. B, C are acknowledged friends of D. B is on quality link - good connectivity to internet and sends broadcast I am sitting on good connection. B downloads update of D. D looses connection. C downloads and update of D from B.
2) - Re-acknowledging of friend ship should be day-wise of few hour-wise. So passwords should change during time using some hashing or whatever.
what do you think?
see ya
mrWax said:
Hi!
Anyone else bothered by sharing their personal info on social network to the third party? I had thought about developing open source distributed social network app based on https server running on android phones. Each member should have android device as a server of his/her profile.
0) - https protocol could provide possibility to connect to your social network using web browser.
1) - One may be able to deal with extensive upload and download or lost connection issues by allowing distributed profile updates sharing:
I suggest following scenario:
D updates profile. B, C are acknowledged friends of D. B is on quality link - good connectivity to internet and sends broadcast I am sitting on good connection. B downloads update of D. D looses connection. C downloads and update of D from B.
2) - Re-acknowledging of friend ship should be day-wise of few hour-wise. So passwords should change during time using some hashing or whatever.
what do you think?
see ya
Click to expand...
Click to collapse
I implemented such a prototype some year ago, although not for mobile clients. You'll have a few issues you'll need to address. You have to use PKI to ensure the identity of the sender in the distributed environment, and you'll have to implement a message routing protocol distributing the messages among the peers. Have a look at my prototype and feel free to steal any ideas: http://whisper.ping.se
Thanks!
Maybe for simplicity one could just copy WIFI WPA2 protocols where router would be one member and internet connection would be connection to the members personal data.
mrWax said:
Maybe for simplicity one could just copy WIFI WPA2 protocols where router would be one member and internet connection would be connection to the members personal data.
Click to expand...
Click to collapse
Huh? I'm sorry, but I can't make any sense of this at all. To achieve what..? I'm missing something here, I guess.
Android is shipped with the BouncyCastle crypto libraries. This is what you'll use to sign the messages with each users private keys to ensure the identity of the author of each message. If you want to encrypt, you'll use this library as well. Since you are working in a distributed store and forward environment, you can not rely on link encryption (e.g. https, wpa2 etc) but have to use point-to-point encryption instead, protecting the message itself instead of the transport channel.
Radio broadcasting resembles similarities to the internet "broadcasting" or: social network resembles similarities to the "person-his friends" virtual private network.
Since it is said that it is hard not to make security flaws when constructing new protocols. i think that adopting whole existing expert made protocol is best solution.
Whole wifi wpa2 protocols could be packed inside ordinary encrypted channel. Where the personal info provider of specific person from which his friends would want to download info - would take place of a "wifi Access Point " in this.
Hi. How to write own authentication server with rest resources? At this moment we have server with Jersey and shiro. Users are stored in mysql database with hibernate.
Now I have problem how to join this with oauth to authenticate android users.
Also this auth/resources server will use vaadin webapp.
On other forum I get suggestion to use spring android framework. http://projects.spring.io/spring-android/
What option will be the best?