I think it's pretty simple to do in FreeBSD, I'll look into that tomorrow. man 8 mknod has some details, but it'll take a bit of hacking to get the device driver working (especially since kernel rebuilds take forever on my dev machine).
Name:
Anonymous2008-05-08 20:21
especially since kernel rebuilds take forever on my dev machine man 8 kldload
Name:
Anonymous2008-05-08 22:14
write small driver, then mknod
Name:
Anonymous2008-05-09 0:28
How would I do this in Windows? I want a drive letter with an infinite file whose contents is DESUDESU.. etc
>>5
From the sound of it, the driver type numbers are statically compiled in. I honestly haven't looked into it enough to know whether my original assumption is correct, or if you can bullshit it with a dynamic kernel module.
when the /dev/desu is read, have Suiseiseki pop on screen and say DESU! DESUDESUDESUDESU~~~! and doll thrown etc etc and the Suiseiseki icon grows in size as the DESU progresses. win?
This better? Yes, I'm too lazy to test it. mkfifo("/dev/desu",0644);
while(1) {
int fd = open("/dev/desu", O_WRONLY);
while(write(fd, "DESU", 4) > 0);
close(fd);
}
>>36
The question was in userspace, fagball. Even if it weren't, why bother with kernel modules when I can have a PORTABLE TURKEY SOLUTION up in minutes?
>>39 The question was in userspace, fagball.
How so, you gigantic homosexualist? Because OP suggested a possible solution that sucks balls? He just said he wants a desu device, you can safely ignore the rest because he doesn't have a clue what he's talking aboout anyway.
Even if it weren't, why bother with kernel modules when I can have a PORTABLE TURKEY SOLUTION up in minutes?
Because you could have a Tarportuble Kurtey Solution that doesn't suck and actually works in a few minutes instead.
Also, kernel space is cooler. And this whole thing is just about the PENIS anyway.
>>48
It's actually very trivial to do >>1's program in Hurd. A desu translator that does nothing but return a string of desu.
Name:
Anonymous2008-05-10 8:33
>>52
Good to know useless crap can be done easily in Hurd while useful things are virtually nonexistant.
Name:
Anonymous2008-05-10 8:38
>>53
Just like the world's leading purely fictional language Haskell rofl xD.
Name:
Anonymous2008-05-10 8:39
>>53
There isn't any non-trivial program in existance that will write itself to fit user requirements. Even Haskell programs do not write themselves - they always require a programmer. The same is true for Hurd. The Hurd needs more programmerse.
>>33
Ok, this version is tested. Production quality code! #!/bin/sh
mkfifo -m644 /dev/desu
while true; do ( while true; do printf DESU; done ) > /dev/desu; done
>>58
Seems to only be a problem if a client doesn't request a multiple of 4 bytes, and even then it's non-critical. If it causes real problems for you though, you can report it in the Bugzilla. Please attach your Xorg.conf, and output of dmesg, stat /dev/desu, and dpkg -l. Thanks for helping!
Name:
Anonymous2008-05-10 15:16
>>60
My coreutils distribution does not include ``stat,'' good Sir. I emply the desktop-ready Lulznix distribution created by Cudder, which is powered by the highly regarded anoncoreutils project.
>>74
Fifo is faggotry, but making it work right is trivial. I refer you to my first post, >>20. You can even do it without leaving an extra process around, à la >>22. The only real differences between a character device and a socket are 1) devices require potentially-dangerous always-resident unportable kernel code, 2) devices support custom ioctls, which we don't use, and 3) devices require fewer context switches, making them desirable for heavily-used services, which this is unlikely to be.
Name:
Anonymous2008-05-11 6:41
>>76
You go girl! Also, you should probably write to Linux Turdballs about implementing /dev/zero and possibly others in userspace, since userspace is teh awesome.
Name:
Anonymous2008-05-11 6:43
PS: character device for the win, at least that thing only produces DESUs when it's asked to.
Name:
Anonymous2008-05-11 6:45
>>76 3) devices require fewer context switches, making them desirable for heavily-used services, which this is unlikely to be.
This service is unlikely to be of any use at all, which means we shouldn't implement it. But since we do, we might at least do it right.
Name:
Anonymous2008-05-11 7:03
>>77
Missing the point is some kind of sport for you, isn't it?
1) Very simple device, accepts only read requests. very hard to do wrong. Portability not required.
2) Then don't. Might make one to switch it into BOKU mode one day though.
3) See >>79
tl;dr your FIFO shit is boring. We don't do boring things on /prog/.
local variables of the box Compatibility with other operating systems are best written in C to handle much of it As for Linux see for yourself Bullshit You fucking moron And.