Forwarding this to the list for future reference ;-) Emmanuel ----- Original Message ----- From: "Jean Schwerer" Sent: Monday, September 10, 2001 12:38 PM Subject: SETI@home > Hi all, > > I guess you know what SETI@home is: It's hopelessly watching for signs of life from > people who don't exist :o) Now Zenja don't worry this list is NOT SETI@home, at least > someone is listening and replying :P > > Ok, enough fun. First things first: I'd like to apologize for the already long silence. This has > been motivated by a couple of things, I guess you'll understand (if you're not interested > skip to the next paragraph ^_^): > - I have been living in a big house with my family and now they are all moving to different > places. I am not moving yet but had to pack some things and help packing other persons' > things, plus my room serves as a repository for things that don't move (read it's become > an ugly mess, in which it takes me up to 3 minutes to go from my computer to my bed, > although it's only 5 meters away). And they are moving with the furniture. I don't know if > you can imagine the mess of 4 people moving to 3 different locations and 1 person > staying on site for a couple of more weeks. This is both tiring and time-consuming. > - I work for a service company that "rents" me to other companies for given projects > (put it differently I am a contractor). My previous contract was for Nortel Networks and > if you've followed what happened to Nortel you can pretty easily guess what happened: > all contracts have been ended a bit abruptly by Nortel, so I am "between contracts", > read I'm paid for waiting for my service company to find a new position for me. > Unfortunately in France service companies don't look for opportunities for you, they > have been created in a context where companies would come to them to offer them > opportunities, and they never did a single prospection effort. They cannot handle any > kind of recession and it's likely they won't find anything interesting for me in a short > future. It's also likely that if they can't find anything interesting for me they won't keep > me longer than a few months. Therefore I am also looking for a new employer in the > telecommunications or IP networks area (heard any good opportunity lately? Feel free > to let me know, I'm willing to move out of France for at least a few years :P). > > Enough with the reasons for a long silence, now let's get back to what is really > interesting :o) > > - It would be nice to have a dedicated mailing-list or at least a generic email address > that targets us all. the "reply-all" function is not working properly in scooby. I don't know > how things are moving on the obedn side but it seems to be quite slow. Eventually we > might create our own dedicated space at sourceforge? Meanwhile I can set an alias on > my own server, e.g. a "netserver_team@xxxxxxxxxxxxx". > > >I've been > >notified by Nathan Whitehorn who also wanted to attempt a netserver rewrite, > >and after hearing about OpenBeOS is interested in joining the project. > > I think it would be great to have Nathan in our group. He's a very good developer, and > deeply interested in networking. He's shown his ability with well-known projects such as > pppoe :o) We definitely NEED him :^) > > >Also, any comments in regards to the use-case diagram I sent out. We > >seriously need to start designing. I need to hear if people are still > >interested in the project. > > Comments below :^) > > >- once initialised, it is usually blocked waiting for an event to awaken it. > >- a user process can awaken the server with a CONNECTION_REQUEST message > >- a user process can send a TRANSMIT message to the server > >- a user process can inform the server that it wants to RECEIVE messages > >- a user process can send a DISCONNECT message to the server > >- the kernel/driver can awaken the server with a incoming data message (READ) > >- the netserver sends the kernel/driver a message (a stream) (WRITE) > > Sort of. I wouldn't "say" the netserver is blocked or sleeping :^) It's actually watching > everything that comes from the kernel/driver, and for security reasons it would be good > that the netserver be able to catch ALL messages on the network, not only those whose > IP is the current machine's. > > >Internally, what does the netserver do? > > > >- it buffers data streams received from user processes > >- it shuffles the data stream and appends data (in accordance to TCP/IP specs) > >- it forwards the new data stream to a device driver > > > >- it buffers data streams received from the device driver > >- it determines who is the receipient of the data (user process, netserver process) > >- it strips and reshuffles the data (according to TCP/IP specs) > >- it forwards the stripped data to a user process or a netserver process. > > I think here 2 is coming before 1. First the netserver recognizes the recipient of data. If > the recipient is invalid (possible attack) log a warning and discard what comes next. > > > > >What are the initial requirements? > > > >- the netserver should be able to serve an arbitrary number of user processes > >- the netserver should be able to link to a set number of network cards > >- the netserver should have selectable shuffling algorithms (eg. IPv4, IPv6, TCP, > AppleTalk etc) > > I think the most urgent is to look for a way to link the netserver to network cards. It's > important to be able to catch what is going on on the network. Implementing the > protocols is a long task but it's far from being the most complex one. > We have an issue as far as the kernel is concerned here: should we build on top of > NewOS's kernel or on top of BeOS's one? Because it's likely that NewOS's kernel is more > advanced than BeOS's kernel from a TCIP/IP point of view. Any thoughts? > > Jean >