| |
| |
| |
|
<Time> I know it's not precisely a ratbox question but what does the ":" mean in a /who, i.e. Time H@ timeless@maximum.sendq.net :0 Glory is fleeting, but obscurity is forever.?
<AndroSyn> its a delimiter
<Time> ah, so its trivial really?
<AndroSyn> the number after the : is the hop count
<AndroSyn> yeah, it makes for parsing a bit easier
<Time> aha, gotcha
<Time> thanks :)
<AndroSyn> np
<anjuta> is there some online info about TS5 and TS6 i can link to ?
<AndroSyn> http://www.ircd-ratbox.org/TS6.txt
<AndroSyn> that is the TS6 stuff
<AndroSyn> https://voting.efnet.net/docs/ts5.txt
<AndroSyn> and that is TS5
<anjuta> great, thank you
<AndroSyn> http://www.iagora.com/~espel/ircd/protocol
<AndroSyn> and that is TS3
<anjuta> TS6 is actually only ratbox extention ?
<AndroSyn> hybrid-7.1 does it as well
<remorse> TS kinda scares me
<anjuta> but as far as I understood from the ratbox's source TS5 and TS6 coexist without any problems ?
<remorse> cause im gonna life to see the 2.038k bug :/
<remorse> live
<remorse> even
<AndroSyn> anjuta: they coexist but you still have the TS5 issues around
<AndroSyn> remorse: hopefully by then most of the 32bit systems will be dead
<AndroSyn> and then time_t will be a 64bit integer
<remorse> i think time_t is signed 32 now....they should just grab the extra bit and give us another 60 years
<AndroSyn> can't do that
<AndroSyn> you'll break things
<remorse> yea i know :/
<AndroSyn> got about 33 years to be moved to 64bit systems :P
<AndroSyn> 32bit time_t will flip over on my 58th birthday :P
<remorse> Tue Jan 19 03:14:07 2038
<jilles> there's no specific way that ircd TS forces 32bit
<AndroSyn> nope
<remorse> time_t is time.h ****
<jilles> ASCII digits can describe 64bit numbers as well
<jilles> lots of stuff is going to break if time_t overflows
<AndroSyn> i'm figuring within 5 years most all new pc's sold will be 64bit
<AndroSyn> easily
<remorse> yey action pack subscription
<AndroSyn> most embedded stuff will have been replaced by then i hope
<anjuta> can't you just use unsigned integers in the servers instead of time_t ?
<AndroSyn> no..
<AndroSyn> time() returns time_t which is a long int
<AndroSyn> time() returns -1 on failure
<anjuta> but still.. ahh
<anjuta> ok
<AndroSyn> and a lot of code ***umes that time() returning < 0 means it failed
<anjuta> it would fail if the clock is broken i suppose..
<AndroSyn> or you p***ed an invalid pointer to time()
<anjuta> yep
<AndroSyn> though gettimeofday() is a bit more useful and i suppose you could probably get away with having it use a unsigned long for the second field..
<AndroSyn> but..that would likely break stuff as well..
<jilles> gettimeofday() is a bit problematic because it pretends to be "accurate" somehow
<jilles> many apps don't need the microseconds but they're rather expensive to measure
<jilles> well, expensive on PCs
<jilles> most other architectures have better clocks
<AndroSyn> well on newer system's gettimeofday() is pretty accurate
<AndroSyn> even on x86
<jilles> but then it's slow
<jilles> and it isn't that accurate on linux
<AndroSyn> well anything recent isn't using the RTC as its primary timer source
<AndroSyn> on x86 if it has a TSC you can easily get microsecond resolution
<jilles> TSC is fast but not really accurate
<jilles> and you can't combine it with speedstep :p
<AndroSyn> that is what the ACPI timer source is for :P
<AndroSyn> which is quite fast and accuate, even when you're playing games with the cpu clock
<anjuta> is it a good idea to extend the /who and possibly /who ! to return the # matches with the end of who numeric ?
<anjuta> or are there better ways to count the matches ?
Return to ratbox
or
Go to some related
logs:
canada
football
osdev
disarm trap in bwl"
|
|