Internet sharing

[PurpleYouko]

It's been a long old time since we really had internet sharing working in darwinbots.

I have come up with a solution which just might work without causing too much of an issue with bandwidth and so on.

My idea is that we utilize something like this forum to allow the client DB programs to log onto a php based program housed on the DB website. the program would be very simple. It would just need to allow registration and log in from the client programs then the organism files can just be uploaded directly to the MySql database, just like posting message in the forum.

We might even be able to simply piggyback this forum directly and allow the clients to read and write files directly into a special thread that we can create.

Each user would be required to manually create an account then put the username and password into the relevent slots in the internet sharing settings, then the program can take care of logging in, posting and reading text files from the thread.

An additional step might be to allow access to a central database of robots so that we don't all have to have our own seperate stables all the time.

Workable idea?
or completely stupid?

What are your thoughts?

[EricL]

Sounds good, but one thing is that I do think we need a file server / file share or peer-to-peer based sharing mechanism as well.  I want to run a several instances on one computer as well as other instances on other computers on my home network and have them all share without having to run a local database server or web server with PHP scripts.  Allowing for a bunch of instances provides environmental locality, barriers to migration, etc. and I hope to see different species dominate the different connected ecosystems.

That said, with a real server, we could provide a means for lots of cool collaborative features such as collecting and presenting global population information accross the connected sims, global dead and born counters for each species, etc.  (Although there are less robust ways to do this using FTP or similar.)  Microsoft's Terrarium did this and the competitive aspects really spurred organism authoring.  That is, seeing at a glance on a web page how many nodes are currently connected and which bots have the highest population numbers in the connected ecosystem motivates one to attempt to author a better bot and become top dog (errr, I mean top bot).

[PurpleYouko]

We thought about doing it by FTP once before but came up against the issue that most site hosts have a severe problem with lots of different people logging on the upload and download files all the time.

That's why I thought of a similar format to the forum itself. Possibly even USE the forum itself.

Good idea about direct connection or peer to peer. Would it be possible to actually use an internet peer to peer system? I think I would prefer to have the option of that kind of thing for a LAN but use a central server for internet. Especially if we can look at (as you say) the number of nodes currently looged into the system and a little info about each one.

I think it would be neat.  

[EricL]

I actually have it working using FTP in 2.42.X.  Not fully tested, but I had it working a couple ov versions ago.  I added FTP password support so that everyone can come under the same FTP account.  Haven't figured out the directory structure that was being used before, but the connectivity with authenticated FTP is working.

Even with peer-to-peer, you still need a rendeveuz server unless we want to get super duper sophisticated.   Terrerium utilized a rendeveuz server for hook-up and central stat reporting, but all the organism transfer was peer-to-peer.  It was all about showcasing the .NET technologies after all...

Do the FTP issue exhibit themselves if everyone comes in on the same account?

[Numsgil]

My friend worked on a bot database for his Comp Sci class.  It's largely incomplete, but a working backbone exists.


Here it is

For internet sharing via a central repository, we'd need:

1.  A master free for all simulation.

2.  Ability to create personal experiments, which are password protected, so people can have a small group of other people help on their experiment.

We'd also need to redesign the organism file format to include things like version number, etc. so older experiments aren't corrupted by newer program releases, etc.

[Henk]

My friend worked on a bot database for his Comp Sci class.  It's largely incomplete, but a working backbone exists.
Here it is

I'm convinced