Hosting a seed on your {I|Android|Meego}Phone

[Christoph Witzany]

Hi guys!

Who would be interested in a project to package diaspora for Android or
IPhone?

Would be great to put it on the device you always carry with you.

+ Everybody will have a smartphone in the near future
+ You don't need a hosting

- Battery of the phone will suffer
- Not everybody will have a open ip

What do you think?

Christoph

[Daniel Ravenatic]

Great idea I say.

[murth]

Christoph Witzany wrote:
> Hi guys!
>
> Who would be interested in a project to package diaspora for Android or
> IPhone?

I think it's safe to say that most users will want this and it will
happen if it can.

[John Favorite]

For academic reason I think it is great. In practice not so much. 

How about a client (phone) which is an authenticated master client of sorts. In other words your phone is trusted to do anything with your info/server and also acts as a portal/proxy into diaspora. Traffic would appear to come from your installed server (not the phone).

[Sam Phippen]

Hi,

I'm not sure seed hosting on a phone is possible. A seed would be able
to push updates out but not pull them in without polling other servers.
There would need to be some proxying layer between the seed on a phone
and the actual internet.

Thanks
--
Sam Phippen <samph...@googlemail.com>

[Alex Wright]

This is much more likely to work.

A truly portable instance of Diaspora would be great, but it's not going to work on mobile phones. You never have a preditable IP address, if it's even a public/real IP address you get. And with the limited storage, cpu, connectivity, battery.. and the fact that this would never in a million years be approved by Apple.. just not going to work that way.

A dumb-mobile client, akin to the Facebook/Twitter clients on handsets now, that connects to your personal defined Diaspora instance is the way to go. We just need an XML/JSON/Something API to implement.

Alex.

[Christoph Witzany]

I'm replying to Alex bc he brings all the arguments against it that came up before.

On 09/16/2010 04:27 PM, Alex Wright wrote:

A truly portable instance of Diaspora would be great, but it's not going to work on mobile phones. You never have a preditable IP address,

That can be mitigated by using a dynamic dns service. Depending on the protocol also by publishing a changed ip address to other nodes if discovery is handled by a distributed hash table.

if it's even a public/real IP address you get.

Yes. That would be bad. But maybe there would be an easy way to provide a proxy service for this. Also I expect that issue to tune out gradually. And finally with the advent of IP6 (now that Duke Nukem Forever will be released even that doesn't seem out of the quesion ...)

And with the limited storage, cpu, connectivity, battery..

I don't see one seed to see that much traffic. Except you really have a high profile seed, then you should definitely go for some more powerful hosting (or by the next generation smartphone with a quadcore ;) ).

and the fact that this would never in a million years be approved by Apple.. just not going to work that way.

Well I would not count on Apple. Right. But this problem is not an issue with Android.

A dumb-mobile client, akin to the Facebook/Twitter clients on handsets now, that connects to your personal defined Diaspora instance is the way to go. We just need an XML/JSON/Something API to implement.

Sure. that will also be needed. But that solves an entirely different problem.


Christoph

[eveclatrel]

It's a good idea, and I'd like to see it happen. Branches for ARM
would need to be found for all of the dependencies. Might also be a
good idea to limit the number of concurrent users and redesign the
storage setup to something a little more simple and flat, but more
battery friendly. Something like this would go well with a dynamic dns
client of some sort. It's also true that Apple would very likely never
approve this, but it's not too bad to jailbreak a phone to get
something like this installed. Plus Android is on the rise and
probably a better platform for this anyway. Y'know, with a swappable
battery and all ;)

(to head off any flames I use an iPhone and a Droid Inc simultaneously
and will be happy to test on both)

[John Favorite]

I am replying to the replying :)

On Thu, Sep 16, 2010 at 10:37 AM, Christoph Witzany <chri...@crofting.com> wrote:

I'm replying to Alex bc he brings all the arguments against it that came up before.

On 09/16/2010 04:27 PM, Alex Wright wrote:

A truly portable instance of Diaspora would be great, but it's not going to work on mobile phones. You never have a preditable IP address,

That can be mitigated by using a dynamic dns service. Depending on the protocol also by publishing a changed ip address to other nodes if discovery is handled by a distributed hash table.

Actually I would say the ip change is not an issue as long as the client is authenticated properly. Two way authentication is pretty easy for web servers, or even ssh. Broadcasting that change is a different issue. 

And with the limited storage, cpu, connectivity, battery..

I don't see one seed to see that much traffic. Except you really have a high profile seed, then you should definitely go for some more powerful hosting (or by the next generation smartphone with a quadcore ;) ).

I would make sure you are thinking about bandwidth of pictures... and all the multiple connections from users to get updated info, pushes of status updates etc. What happens when your phone goes to sleep? If I upload 40 pictures and 20 of my friends click through them I am possibly sending 800 pictures out of my phone! 

 

A dumb-mobile client, akin to the Facebook/Twitter clients on handsets now, that connects to your personal defined Diaspora instance is the way to go. We just need an XML/JSON/Something API to implement.

Sure. that will also be needed. But that solves an entirely different problem.

Agreed.

[shadowfirebird]

I've got a Nokia N900 ( = Maemo) and I certainly like this idea. I
think it's doable there.

But maybe we should have a reference desktop implementation first?

On Sep 16, 2:53 pm, Christoph Witzany <christ...@crofting.com> wrote:

[Steve Dekorte]

Why do you need a predictable IP? Any data can be shared with other nodes so it doesn't matter if your node is coming on and off the network. If the data is private, it can be shared in encrypted form, etc.

[Alex Wright]

If two users each had dynamic IP addresses on private networks, like
say a cell handset, how would that work?

[John Favorite]

dynamic dns update?

[Alex Wright]

Very few mobile phones get a real as in public IP address. At least here that is the case, and I think the UK is fairly ahead of the curve on mobile data. That was my point re: private networks. Dynamic DNS does little to help you when you're both behind NAT.

[Alex Cline]

Dynamic DNS works for systems that have IPs that change but do so infrequently.  I can imagine the DNS system isn't fast enough to handle a cell phone in a car on the highway jumping from wireless tower to wireless tower getting a new address each hop.

I think once there is a Diaspora API, that a client app for mobile phones would be nice -- I can login to my Diaspora install that is hosted on my web server.  Hosting a whole installation on a mobile phone just would pose too many problems.

--
http://alexcline.net

[John Favorite]

On Thu, Sep 16, 2010 at 1:49 PM, Alex Cline <alex....@gmail.com> wrote:

Dynamic DNS works for systems that have IPs that change but do so infrequently.  I can imagine the DNS system isn't fast enough to handle a cell phone in a car on the highway jumping from wireless tower to wireless tower getting a new address each hop.

good point. dns propagation would be a problem!

 

I think once there is a Diaspora API, that a client app for mobile phones would be nice -- I can login to my Diaspora install that is hosted on my web server.  Hosting a whole installation on a mobile phone just would pose too many problems.

This is my thinking as well. 

[John Favorite]

I am on Sprint here in the US and get a real IP. I have no idea what they filter though...

[Steve Dekorte]

As long as they have a sufficiently large list of nodes that where online the last time they were on the network, the chance of none of them being online when they reconnect approaches zero. Also, a set of open root nodes that were very likely to be online could be used for falling back on in extreme cases.

[John Favorite]

sounds like you are talking about DHT :)