IRC log started Wed Sep 29 00:00:01 1999
-:- SignOff liar: #TUNES (BRiX [http://www.qzx.com/brix] :: sleep)
[msg(TUNES)] permlog 1999.0929
-:- SignOff td: #TUNES (Ping timeout for td[1Cust11.tnt3.wilmington.nc.da.uu.net])
-:- Connection closed from irc.us.openprojects.net: Success
-:- Use /Server to connect to a server
-:- Connecting to port 6667 of server irc.us.openprojects.net [refnum 0]
-:- BitchX+Deb1an: For more information about BitchX type /about
-:- Welcome to the Internet Relay Network TUNES (from carter.openprojects.net)
-:- Your host is carter.openprojects.net, running version u2.10.05.18.(ipcheck4-5) (from carter.openprojects.net)
-:- This server was cobbled together Wed Apr 28 1999 at 12 02:19 EDT(from carter.openprojects.net)
-:- carter.openprojects.net u2.10.05.18.(ipcheck4-5) dioswkfcg biklmnopstv 
-:- [local users on irc(46)] 11%
-:- [global users on irc(168)] 40%
-:- [invisible users on irc(255)] 60%
-:- [ircops on irc(13)] 3%
-:- [total users on irc(423)]
-:- [unknown connections(0)]
-:- [total servers on irc(33)] (avg. 12 users per server)
-:- [total channels created(154)] (avg. 2 users per channel)
!carter.openprojects.net Highest connection count: 77 (76 clients)
!carter.openprojects.net Welcome to Open Projects!  You are on 4 ca 1(2) ft 14(14) tr.
-:- Mode change [+f] for user TUNES
-:- Mode change [+iws] for user TUNES
-:- JOIN activated by "TUNES #tunes tunes@bespin.dhs.org "
-:- TUNES [tunes@bespin.dhs.org] has joined #tunes
-:- Topic for #TUNES: Free Reflective Computing System <http://www.tunes.org> || http://www.aftersleeves.org/apostle/ || http://www.netpedia.net/sites.html
-:- topic set by hcf [Mon Sep 27 13:28:40 1999]
-:- [Users(#Tunes:4)] 
[ TUNES     ] [ abi       ] [ smkl      ] [@Tril      ] 
-:- Channel #Tunes was created at Sun Feb 28 08:48:06 1999
-:- BitchX+Deb1an: Join to #tunes was synced in 6.586 secs!!
-:- Mode change [-s] for user TUNES
-:- _ruiner_ [nate@ppp159.wi.centurytel.net] has joined #tunes
-:- _ruiner_ [nate@ppp159.wi.centurytel.net] has left #tunes []
-:- smokie_ [tw026024@zaalf05.twi.tudelft.nl] has joined #Tunes
<smokie_> hi
07:10am
-:- Downix [down@d-gnaps-261.ici.net] has joined #tunes
-:- SignOff smkl: #TUNES (Ping timeout for smkl[glubimox.yok.utu.fi])
-:- hcf [nef@me-portland-us128.javanet.com] has joined #tunes
-:- smkl [sami@glubimox.yok.utu.fi] has joined #tunes
-:- core [core@core.suntech.fr] has joined #tunes
-:- SignOff smokie_: #TUNES (BitchX-75+Deb1an -- just do it.)
<Downix> hey
-:- Downix [down@d-gnaps-261.ici.net] has left #tunes []
-:- SignOff core: #TUNES (Ping timeout for core[core.suntech.fr])
11:30am
-:- _ruiner_ [nate@ppp022.wi.centurytel.net] has joined #tunes
-:- SignOff _ruiner_: #TUNES (Leaving)
-:- _ruiner_ [nate@ppp022.wi.centurytel.net] has joined #tunes
<_ruiner_> anybody here?
12:20pm
-:- SignOff _ruiner_: #TUNES (Ping timeout for _ruiner_[ppp022.wi.centurytel.net])
-:- eihrul [eihrul@usr5-ppp80.lvdi.net] has joined #tunes
-:- SignOff eihrul: #TUNES ([x]chat)
-:- eihrul [eihrul@usr5-ppp80.lvdi.net] has joined #tunes
-:- Connection closed from irc.us.openprojects.net: Success
-:- Connecting to port 6667 of server irc.us.openprojects.net [refnum 0]
-:- Unable to connect to port 6667 of server irc.us.openprojects.net: No route to host
-:- Connecting to port 6667 of server irc.us.openprojects.net [refnum 0]
-:- BitchX+Deb1an: For more information about BitchX type /about
-:- Welcome to the Internet Relay Network TUNES (from adams.openprojects.net)
-:- Your host is adams.openprojects.net, running version u2.10.05.18.(ipcheck4-5) (from adams.openprojects.net)
-:- This server was cobbled together Wed Apr 28 1999 at 13 24:47 EDT(from adams.openprojects.net)
-:- adams.openprojects.net u2.10.05.18.(ipcheck4-5) dioswkfcg biklmnopstv 
-:- [local users on irc(38)] 8%
-:- [global users on irc(222)] 44%
-:- [invisible users on irc(284)] 56%
-:- [ircops on irc(15)] 3%
-:- [total users on irc(506)]
-:- [unknown connections(0)]
-:- [total servers on irc(34)] (avg. 14 users per server)
-:- [total channels created(142)] (avg. 3 users per channel)
!adams.openprojects.net Highest connection count: 45 (44 clients)
!adams.openprojects.net Welcome to Open Projects!  You are on 4 ca 1(2) ft 14(14) tr.
-:- Mode change [+f] for user TUNES
-:- Mode change [+iws] for user TUNES
-:- JOIN activated by "TUNES #tunes tunes@bespin.dhs.org "
-:- TUNES [tunes@bespin.dhs.org] has joined #tunes
-:- [Users(#Tunes:5)] 
[ TUNES     ] [ abi       ] [ eihrul    ] [ smkl      ] [@Tril      ] 
-:- Channel #Tunes was created at Sun Feb 28 08:48:06 1999
-:- BitchX+Deb1an: Join to #tunes was synced in 8.394 secs!!
-:- Mode change [-s] for user TUNES
-:- liar [brand@p0wer.qzx.com] has joined #tunes
-:- _ruiner_ [nate@ppp125.wi.centurytel.net] has joined #tunes
<_ruiner_> hmmmmmmm........
<liar> hmmmmm
<liar> damn there is a shitload of email from the bochs mailing list today
<_ruiner_> ha ha, sucker
<liar> and half of it is from core
06:10pm
-:- carlito [sabanmr@137.28.109.92] has joined #tunes
<carlito> what's up
<liar> not much
06:30pm
<liar> lar1?
<abi> it has been said that lar1 is mailto:lar1@bigfoot.com or mailto:lar1@tunes.org
-:- Metaphor [Meta@spc-isp-ktc-uas-8-39.sprint.ca] has joined #tunes
06:40pm
-:- SignOff Metaphor: #TUNES (ircII 2.8.2-EPIC4pre1.001 -- Starting all over again.)
-:- ElGato` [no@209-68-229-51.dialup.cust.tfb.com] has joined #tunes
<ElGato`> my spanish teacher is an airhead you guys...what's the correlation you ask?
<ElGato`>  she's finnish
<ElGato`> :D
<ElGato`> j/k
-:- SignOff ElGato`: #TUNES (:P)
-:- lar1 [LarMan@dialup-209.245.132.174.SanJose1.Level3.net] has joined #tunes
<lar1> Hey
07:10pm
-:- SignOff lar1: #TUNES (Read error to lar1[dialup-209.245.132.174.SanJose1.Level3.net]: Connection reset by peer)
-:- NetSplit: varley.openprojects.net split from irc.linux.com [07:17pm]
-:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [varley.openprojects.net]
-:- Netjoined: varley.openprojects.net irc.linux.com
-:- abi [nef@bespin.dhs.org] has joined #Tunes
-:- _ruiner_ [nate@ppp125.wi.centurytel.net] has joined #Tunes
-:- carlito [sabanmr@137.28.109.92] has joined #Tunes
<liar> anyone awake?
07:30pm
<eihrul> yes
<liar> ha i knew i was right
<liar> beos has a 250us latency on interrupts
<liar> and they think thats kewl cuz windows is 20-100ms
<eihrul> 250 us?
<eihrul> that's a lot
<liar> i couldnt figure out what the hell they could be doing to use up that much time
<eihrul> it's a proprietary OS :)
<liar> qnx is 4.3us
<eihrul> yah... all you need to do is link into scheduler and context switch
<liar> on a p133
<eihrul> and if you reuse a thread
<eihrul> it's just a context switch
<liar> be was probably running their test on a p2-500 :)
<liar> that makes me feel better cuz i know that brix wont even come close to 250us
<liar> so now i just gotta come as close as i can to 4.3us
<eihrul> damn
<eihrul> that's like 33,000 cycles :)
<eihrul> unless my quicky C program is wrong
<liar> qnx is 571 cycles
<eihrul> 571 is a lot
<eihrul> that's probably about 300-400 instructions
<eihrul> maybe actually 300
<liar> ya
<eihrul> my message passing doesn't even take that long
<liar> u sure?
<eihrul> my thread migration path is only 70 or so instructions :)
<eihrul> and it lets you pass all registers
<eihrul> and pass a message up to 4MB :)
<eihrul> and most of the 70 instructions are well paired
<liar> this is for an interrupt
<eihrul> well an interrupt should be less than message passing
<liar> u have to save the current thread and load another
<eihrul> as there is much less to do :)
<eihrul> so, i do that in my thread migration
<eihrul> and if i used interrupt migration :)
<liar> a context switch in qnx is 1.95us
<eihrul> i could probably do both in about the same amount of cycles
07:40pm
<eihrul> at what MHz?
<liar> 259 cycles
<liar> qnx did all its tests on a p133
<liar> qnx (386/33) is 22.6us context switch
<liar> http://www.qnx.com/products/os/qnxrtos.html
<eihrul> 259 cycles is still more :)
<eihrul> way more than i have... 3 times more
<liar> under Performance Profile
<eihrul> QNX tries to implement POSIX
<liar> but qnx A) works B) is more secure than yers cuz C) it works :)
<eihrul> no...
<eihrul> on the contrary, an EMK system has never crashed
<eihrul> nor has it ever had a security breach... :)
<liar> ha
<liar> nor has it ever run
<eihrul> minor detail
-:- NetSplit: varley.openprojects.net split from irc.linux.com [07:42pm]
-:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [varley.openprojects.net]
<eihrul> and even so, my microkernel is much simpler
<eihrul> and thus it can easily run faster, as it makes no attempt to mimic posix
<liar> qnx doesnt try to implement posix, it does
<eihrul> exactly :)
<eihrul> and plus, we both use forms of thread migration
<eihrul> whereas QNX is sync message passing :)
<liar> message passing isnt the question right now
<eihrul> i commented about the interrupt latency
<liar> its context switching and interrupt latency
-:- abi [nef@bespin.dhs.org] has joined #Tunes
-:- _ruiner_ [nate@ppp125.wi.centurytel.net] has joined #Tunes
-:- carlito [sabanmr@137.28.109.92] has joined #Tunes
<eihrul> 260 cycles is a lot :)
<liar> i know
<liar> so it shouldnt be hard to beat qnx
<liar> but our oses need to run just as stable as qnx or being faster doesnt mean jack
<liar> this is what i find funny
<liar> qnx gets 1.1MB on a 10Mbit eth0 card but only gets 7.5MB on a 100Mbit card
<liar> linux also gets 1.1MB on a 10Mbit card
<liar> but shouldnt a 100Mbit be 10 times faster than a 10Mbit????
<eihrul> there is nothing to be instable about my OS
<eihrul> it barely implements anything :)
<liar> ah here we go
<eihrul> just memory objects, activations, and tasks
<liar> With nested interrupts, these interrupt latencies represent the worst-case latency for the highest priority
<liar>     interrupt. Interrupt priority is user-definable and the interrupt latency for lower-priority interrupt sources is
<liar>     defined by the user's application-specific interrupt handlers.
<liar> i still dont see how worst-case is that high
<eihrul> well it's like 60 cycles atleast to enter and exit the interrupt
<eihrul> that still leaves 200 cycles in between however
<liar> right
07:50pm
<eihrul> so 125-150 instructions?
<eihrul> actually... then you have to save and restore segment registers....
<eihrul> and other things, could pile on a lot of cycles
<liar> i dont save segregs
<liar> other than ds cuz i change it
<liar> cs is handled by the int
<eihrul> you still have to write the seg regs
<eihrul> and that still takes cycles
<liar> huh?
<liar> i dont have to save/restore es,fs,gs
<liar> or ss
<eihrul> you still have to load ds
<eihrul> with a kernel selector and then a user selector
<liar> so
<eihrul> and if i remember correctly, thats around 5 cycles...
<eihrul> and then you have to put the selector in a general purpose register
<liar> oh wait
<liar> i dont save ds
<eihrul> you have to change the segments though :)
<eihrul> even if you don't save them
<liar> why?
<eihrul> unless you're only reading data
<eihrul> then it's possible to map user read only the whole kernel space
<liar> oh wait, i load ds and es with the user ds selector
<eihrul> but other than that, you need access to write :)
<eihrul> or you could use an ss segment override
<liar> my kernel space is user read
<eihrul> since ss is automatically loaded for you :)
<eihrul> but takes a cycle for the prefix i think
<eihrul> isn't esp implicitly ss?
<liar> cpl0 can write to a user read segment
<eihrul> then why do you bother loading es/ds?
<liar> esp and ebp use ss
<eihrul> if you could just do that
<liar> incase the stupid user loaded cs into one of them :)
08:00pm
<eihrul> can't you prevent the user from loading cs with one of the privilege level fields?
<liar> or tss
<liar> no
<eihrul> i seem to remember some PL field on the descriptors themselves
<liar> my context switch is 55 instructions best case
<liar> im not even gonna count worst case
<liar> best case is no fpu usage, normal non-system code, and no timers are set
<eihrul> you have to for a real time system :)
<eihrul> worst case is important
<eihrul> 1600x1200... wow
<eihrul> that's near double my old monitor
<liar> and if timers are set but not expired its 57
<liar> i use 1600x1200
<eihrul> is that counting the 60 or so cycles for the timer interrupt? :)
<liar> oops, i counted the iretd
<liar> 54/56 + irq/iretd
<liar> another 9 for a driver that uses io
<eihrul> and then even instructions is a bad heuristic :)
<eihrul> since the pentium is so backwards
08:10pm
<liar> the irq/iretd is 75 cycles 
<liar> from user mode and 58 from kernel mode
<eihrul> ouch
<liar> so we have 184 cycles to beat qnx
<liar> pushad/popad is 10 cycles
<liar> i still dont see how beos uses 25,000 cycles
<eihrul> i calc'd 33 :)
<eihrul> k that is
<liar> 33 what?
<eihrul> brb, going to restart X11 in all of it's 1600x1200 splendor
<eihrul> thousand cycles
-:- eihrul [eihrul@usr5-ppp80.lvdi.net] has left #tunes []
-:- eihrul [eihrul@usr5-ppp80.lvdi.net] has joined #tunes
<liar> At the lowest
<liar>                     level, the kernel provides very low scheduling latency services. When a driver
<liar>                     event [occurs], [the kernel] releases a semaphore which causes a thread to run.
<liar>                     That [elapsed] time is very small, measured on the order of microseconds. In this
<liar>                     upcoming release we've worked to bring this scheduling latency down to the point
<liar>                     where real-time threads' latencies can be [measured at around] 250
<eihrul> this is, ummm, big....
<liar>                     microseconds. 
<eihrul> definitely need larger fonts now
<liar> hahah
<liar> what size monitor?
<eihrul> 1600x1200, 17"
08:20pm
<eihrul> whereas before i was on a 1024x768, 13"
<liar> same here, im using a 14pt for bitchx
<liar> and netscape and pretty much everything else
<eihrul> what's the name of the font?
<_ruiner_> later all
-:- SignOff _ruiner_: #TUNES (Leaving)
-:- carlito [sabanmr@137.28.109.92] has left #tunes []
<liar> -b&h-lucidatypewriter-medium-r-normal-*-*-140-*-*-m-*-iso8859-1
<eihrul> 9x15, 7x14 :)
<liar> what?
-:- kepe [kepe@ci1000097-a.nash1.tn.home.com] has joined #Tunes
-:- kepe [kepe@ci1000097-a.nash1.tn.home.com] has left #tunes []
-:- eihrul [eihrul@usr5-ppp80.lvdi.net] has left #tunes []
-:- eihrul [eihrul@usr5-ppp80.lvdi.net] has joined #tunes
<eihrul> there, 10x20 is good
<liar> heh
<eihrul> still a bit too small too
<liar> did u read that crap i dumped about beos?
<eihrul> nope, been restarting X11, repaste it
08:30pm
<liar> thats probably why it takes 250us
<eihrul> you said 250ms
<liar> no i didnt
<liar> windows is 20-100ms
<liar> beos is 250us
<liar> qnx is 4.3us
<eihrul> ah... ok
<eihrul> foo
<liar> i still dont see how it takes 250us
<eihrul> there... 
<eihrul> ok... was just suffering minor confusion with the 260 cycles and 250 us :)
<liar> 260 cycles is the context switch time for qnx (1.95us)
<liar> 250us (beos) 4.3us (qnx) is interrupt latency time 
<liar> 4.3us is 571 cycles or soemthing
<eihrul> must restart X11 again
-:- eihrul [eihrul@usr5-ppp80.lvdi.net] has left #tunes []
<liar> why?
-:- eihrul [eihrul@usr5-ppp80.lvdi.net] has joined #tunes
<liar> why did u restart?
08:40pm
<eihrul> when i reconfigured it
<eihrul> it overwrote another config file and selected the wrong x server
<eihrul> so had to put the right one back then restart
08:50pm
<eihrul> liar: there?
-:- Netjoined: varley.openprojects.net irc.linux.com
-:- abi [nef@bespin.dhs.org] has joined #Tunes
<liar> ya
<eihrul> is there any way to do something like this in nasm?
<eihrul> %define foo(bar,baz) bar.baz
<liar> uhh
<liar> i think so
<eihrul> it isn't evaluating correctly
<eihrul> it isn't recognized bar or baz
<eihrul> when they're grouped like that
<eihrul> so i need some way to concatenate strings
<liar> %define param(a,b) ((a)+(a)*(b))
<liar> try the () aaround it
<eihrul> this is specifically what i'm trying:
<eihrul> %define emStackOffset(structure, field) (structure.size - structure.field)
<eihrul> so if i put the () around it
<eihrul> then nasm will bitch
<liar> %define emStackOffset(structure, field) ((structure).(size) - (structure).(field))
09:00pm
<eihrul> i tried that
<eihrul> it doesn't like it
<liar> dunno
<eihrul> that's not good...
09:10pm
<liar> have u seen the matrix?
<eihrul> nope... hrmm, going to fiddle with nasm's preprocessor and see if i can get it to do what i want
<liar> u havent seen the matrix? yer missing out man
<liar> i just want to know why they teach em to move so fast they can dodge bullets but they dont teach em to shoot a person with less than 100 rounds
<eihrul> it's a movie...
<liar> so
<liar> it would have been better if they could shoot straight
<eihrul> ah...
<eihrul> very simple, all i have to do is modify the tokenizer
09:30pm
<eihrul> ok...
<eihrul> i think i fixed nasm :)
<liar> heh
<eihrul> damn, almost
<liar> damn, i almost had the resizing working in bfe
09:50pm
-:- NetSplit: varley.openprojects.net split from irc.linux.com [09:51pm]
-:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [varley.openprojects.net]
-:- Netjoined: varley.openprojects.net irc.linux.com
-:- abi [nef@bespin.dhs.org] has joined #Tunes
<eihrul> it worked on half of it...
-:- _ruiner_ [nate@ppp166.wi.centurytel.net] has joined #tunes
<liar> _ruiner_: whats say u help me figure out howto resize my xforms without them lookin like shit
10:00pm
<_ruiner_> sorry chief, I'm working on a java applet
<liar> YES, i got the buttons resizing correctly
<liar> downloaded xfmail and looked at how they did xforms crap :)
<_ruiner_> heh heh
<_ruiner_> thief
10:10pm
<liar> perfect, the structure windows resize perfectly now
<_ruiner_> <g>
<_ruiner_> way to go boss
<liar> stop calling me that
<liar> i get called that all day at work
<_ruiner_> k boss
<_ruiner_> ;-p
-:- water [water@tnt-10-68.tscnet.net] has joined #tunes
<liar> water!
<water> liar!
<water> ruiner, too?
10:20pm
-:- hcf [nef@me-portland-us321.javanet.com] has joined #tunes
<water> hcf!
<hcf> hi
* water/#tunes checks the logs
<_ruiner_> ruiner too
<_ruiner_> whats up water?
<water> trying to get tunes back in shape
<water> and still working out how arrow will develop
<liar> eihrul: u here?
<_ruiner_> tunes is dead
<water> you guys figured out how slow interrupts are?
* _ruiner_/#tunes wonders if tunes was ever alive
<eihrul> liar: wha?
<liar> eihrul: i fixed all the resize problems
<liar> eihrul: even the hide/show button works
<liar> when running on bochs 981102 bfe is perfect and ready for 1.0 release
<eihrul> nasm is still not cooperating...
<liar> kevin is looking into the 990925 problem
<water> damn. i missed tril, fare, AND core. :(
<water> what's all this garbage on the tunes mlist, anyway?
10:30pm
<_ruiner_> I want to get an orb drive
<water> who cares?
<_ruiner_> I care
* water/#tunes suggests #linpeople for those not interested in having their hardware preference logged on a linux server in northern Washington state.
<_ruiner_> eh?
<_ruiner_> those redmondians don't care
<water> i mean bespin
<_ruiner_> though I don't doubt they use linux for all their critical stuff
<_ruiner_> oh.....why should I care about if bespin logs it?
<water> i do care
<_ruiner_> uhhuh.....why?
<water> btw, that means that you Don't care, in which case i recommend #linpeople
<water> "Free Reflective Computing System"
<water> how the hell does it relate?
<_ruiner_> talk about anal...
<water> no, just frustrated
<_ruiner_> how does being frustrated relate to a free reflective computing system?
<hcf> its a meta-project issue
10:40pm
<_ruiner_> I want to get an orb drive
<eihrul> liar: nasm is very loopy... code-wise
<eihrul> liar: some of the comments are scary, even the authors aren't sure of some of it's behavior....
10:50pm
<liar> heh
-:- td [x@1Cust90.tnt3.wilmington.nc.da.uu.net] has joined #tunes
<water> well, the conversation is pretty damn dry and useless (i don't mind saying), so i'll see you guys in about an hour or so
<water> hi td
<td> hello
11:00pm
* water/#tunes -> afk
-:- SignOff _ruiner_: #TUNES (Leaving)
-:- SignOff hcf: #TUNES (Leaving)
<liar> eihrul: bfe is fully functional with bochs 981102a
<liar> eihrul: its up on the page for download
-:- SignOff liar: #TUNES (BRiX [http://www.qzx.com/brix] :: sleep)
[msg(TUNES)] newlog 1999.0930
IRC log ended Thu Sep 30 00:00:01 1999