Re: Net_GSQ API Question

"David Grant" <david@grant.org.uk> wrote in message news:4087C249.8060102@grant.org.uk... > Aidan, > > Aidan Lister wrote: > > If a connection to a server fails, there is a 1000ms delay in the socket > > before it's killed (this is minimum). That happening 4 times is a problem. > > This would only occur if you were creating a socket four times though. > Why not create the socket on instantiation, then test the socket for > signs of life before you throw something down it. Because it's a UDP query, it'll stay happily connected to "null" until you try and read from it. > > >>I would've thought an array to be a more appropriate argument to the > >>query method than multiple length arguments. > > > > > > Yeh, It could be done either way, I think the overloading is nicer. > > You'll be returning in an array, but the design choice is up to you. > > >>Personally I would stick > >>with the seperate methods, as it leaves the API obvious. > >> > > > > Requests can't be concatenated, this is a big problem if you're querying 10 > > servers. It would be 4 times slower (with the gamespy protocol) sending each > > request seperatly. > > How about collecting all the information the implementor requires, then > ask them to call another method to start the retrieval. For example: > > $server = new Net_GamesServerQuery($ip, $port, $game); > > $server->requestFoo(); > $server->requestBar(); > $server->requestBaz(); > > $results = $server->sendQuery(); Hey, I like that! > > >>If someone > >>wants to alter it, then the license allows them to do so. > > > > Well sure, but we're aiming to release top quality stuff, not quality stuff > > that can be improved :) > > That's a fair point! > > Regards, > > David