16/05/2004 @10:16:32 ^11:06:07

don't stop baby not tonight

I can never get enough of your love


To snafu I've added a livejournal owned by some guy I don't actually know but who is a constant source of hilarity on the rogue #warwick and so I decided it'd be worth knowing when it gets updated.

I have also reduced its checking frequency. Scans are at 42 minutes past the hours 2, 7, 12, 16, 21, and 23 (presumably that's triv.org.uk local time, I just realised I'm not sure exactly how cron handles timezones) Anyway nobody updates often enough for such frequent checking to be a necessity. Apologies to people who aren't on UK time any more, but if you will go live in strange foreign parts that's your fucking problem I guess. Having said that if you would like any changes please feel free to ask.


Simplistically the majority of the internet works like this: you have two computers, hosts, whatever, call them C and S. Then S opens a port P to which C connects. Effectively S is saying "I have this sort of information available" (where "this sort of information" is a function of P) and C comes along, says "hello, can I have some of this" and off it goes.

In particular IRC works in exactly the same way, you connect to a server, join a channel, and send messages that the server transmits to all other clients inthe same channel. Encapsulated inside IRC there is another protocol called DCC which is used primarily for file transfer between clients. The client which has the file acts like a server, opening a port the number of which is sent to the other client, which connects to that port and gets the file.

I believe this is the wrong way round to do it. Many clients acting as DCC servers are on machines running behind large firewalls over which the owner of the DCC server has no control. The client who wants the file however is more likely to be a home user with either very little firewalling or at least the ability to open ports in it if required. I think DCC would work a lot better if the client opens the port and sends the port number to the server.

I mean, I know I can give files to other people because I set up iptables with the connection tracking modules, but the number of times I've tried connecting to some DCC server and it's just hung there waiting for data is extremely irritating. I guess it's fortunate that I was just messing with this stuff to see if it worked, and there aren't any files that I really want.