[dcdev] Developing
Carl-Adam Brengesjö <[email protected]>
2004-01-15 11:57
[email protected]

What I've seen you have done so far is quickly going down to technical stuff, I'm used to a more .. discussion of some sort were you dont decide _how_ to do things, but _what_ it should do it. Yes, this may be a "mickey$oft" way of doing it, I dont know. But atleast you wont be stuck on a subject for a couple of weeks just to decide how a single thing would look like in the end (we are far from a complete protocol)
It's called 'development', you set up guidelines in the beginning to develop a finished product (in this case the protocol is a product, even if it isn't end-user - if you get me). A example of this:

1) the description of _what_ it does:
"When a user connects and is accepted by the server, the server should respond with a message indicating that it is ready for authentication information. This message should (optional) include the client's reverse look-up dns. The client in return responds with it's proposed nickname and password (if any, optional)..."
2) the text explaining _how_ it does _what_:
Client connects and is accepted by the server. The server (hub) initially sends `AUTH dns=the-client-dns.example.com', telling the client can send auth information. The client responds with `NICK foo bar', telling the hub it would like to use the nick `foo' with password `bar'.

note! this is _not_ a proposed protocol, just a way to make a concept of one.

I know that discussion of _how_ is sometimes required to decide _what_ should be included in the protocol, and that `what' is already defined in the current protocol.

We are currently discussing how a search should look like when we even hasn't decided how a handshake should look like. Yes - the handshake is not a big deal, but it can tell what information a user should have (like the current with email, description, connection speed, sharesize and so on).
Yes - the search part is the biggest one, and therefor more important.
But it would be good if we could get anywere with the protocol instead of discussing details (I have myself to blame aswell, as of my last mail). Like I said, comeon.. not even have we decided what info should be included with the user.

Not even the choice of binary or text protocol affects the `what' concept of the protocol, nor does it define the syntax used. So I say a `what' concept should be done as the number 1 priority.

I may be a bit revolutionary, but most of you seems just eager to showoff with teq and programming skills. I just like to get the concept done (I may have called it `definition' or `description' earlier in this mail, but I figured the out the word [concep] ;)

as fnord stated in the beginning of this mailinglist: "Let's get this show on the road, then. :)"


ps. i originally wrote this and my reply to fnord and fredrik's mails about searching as one mail, but they kinda got off-subject from eachother, so I splitted them.

DC Developers mailinglist