IRC log started Fri Aug 27 00:00:00 1999
[msg(TUNES)] permlog 1999.0827
-:- SignOff eihrul: #TUNES (Leaving)
-:- SignOff Tril: #TUNES (changing servers)
-:- Tril [dem@bespin.dhs.org] has joined #TUNES
-:- mode/#Tunes [+o Tril] by ChanServ
-:- fare [fare@quatramaran.ens.fr] has joined #Tunes
-:- Linatux [Linatux@ns.security.co.nz] has joined #Tunes
-:- Linatux [Linatux@ns.security.co.nz] has left #Tunes []
-:- overfien [overfiend@m240.telcomplus.net] has joined #Tunes
<overfien> yoh
02:30am
<fare> hoy
03:10am
-:- smoke [smoke@13dyn235.delft.casema.net] has joined #tunes
<fare> hi smoke
* fare/#Tunes is now reading "Metamagical Themas"
03:40am
<smoke> hi fare
03:50am
<fare> if I could find a better ending, I could make it KLUGE
<fare> what does KLUGE mean in german, already?
04:10am
<fare> abi: cora is at http://www.cora.justresearch.com/
04:20am
<Crimson__> a TNT2-server is being developed right now...
-:- Crimson__ is now known as Crimson
05:20am
<Plundis> Crimson: it will support both Riva 128, TNT and TNT2
-:- NetSplit: fontana.openprojects.net split from varley.openprojects.net [05:23am]
-:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [fontana.openprojects.net]
-:- Netjoined: fontana.openprojects.net varley.openprojects.net
-:- smoke [smoke@13dyn235.delft.casema.net] has joined #Tunes
<Crimson> Plundis: very cool... and very wool too.
05:30am
-:- AlonzoTG [Alonzo@209-122-216-176.s176.tnt3.lnh.md.dialup.rcn.com] has joined #tunes
-:- SignOff fare: #TUNES (Ping timeout for fare[quatramaran.ens.fr])
-:- fare [fare@quatramaran.ens.fr] has joined #Tunes
-:- fare is now known as Fare
-:- hcf [nef@me-portland-us147.javanet.com] has joined #tunes
-:- SignOff smoke: #TUNES (One day sheep will rule the world)
-:- hcf has changed the topic on channel #Tunes to: Free Reflective Computing System <http://www.tunes.org> | http://cybercom.net/~rbjones/rbjpub/logic/jrh0100.htm | http://www.cb1.com/~john/thesis/thesis.html | http://www.cs.cornell.edu/slk/ | http://www.cora.justresearch.com/
-:- hcf has changed the topic on channel #Tunes to: Free Reflective Computing System <http://www.tunes.org> | http://cybercom.net/~rbjones/rbjpub/logic/jrh0100.htm | http://www.cb1.com/~john/thesis/thesis.html | http://www.cs.cornell.edu/slk/ | http://www.egroups.com/group/os-edge/10.html? | http://www.cora.justresearch.com/
<hcf> damn
-:- hcf has changed the topic on channel #Tunes to: www.tunes.org | cybercom.net/~rbjones/rbjpub/logic/jrh0100.htm | www.cb1.com/~john/thesis/thesis.html | www.cs.cornell.edu/slk/ | www.egroups.com/group/os-edge/10.html? | www.cora.justresearch.com/
07:40am
-:- SignOff AlonzoTG: #TUNES (Read error to AlonzoTG[209-122-216-176.s176.tnt3.lnh.md.dialup.rcn.com]: Connection reset by peer)
<Fare> hum
<hcf> Fare: hoy, sup?
<Fare> back at work
08:10am
-:- AlonzoTG [Alonzo@216-164-233-212.s593.tnt8.lnh.md.dialup.rcn.com] has joined #tunes
-:- smkl [sami@MCCCXXV.rdyn.saunalahti.fi] has joined #tunes
-:- augspies [augspies@ci419747-a.ashvil1.nc.home.com] has joined #tunes
<augspies> Hey all.
<hcf> hey aug
<augspies> Does anyone here listen to icecast streams with xmms?
<augspies> hcf: hi
<hcf> augspies: look at tunes.org and u'l see this isnt the channel u thot it was
<augspies> heehe, whoops sorry
* augspies/#tunes blushes
<hcf> common mistake
-:- augspies [augspies@ci419747-a.ashvil1.nc.home.com] has left #tunes []
09:00am
<Fare> hum
<hcf> why hum?
* Fare/#Tunes likes humming
<Fare> (and whistling, too)
<hcf> Fare: how about when u get a sec, u update the HLL page
09:20am
<Fare> update it?
<Fare> I'd rather update the HLL sources...
<Fare> unless you have specific issues...
<hcf> Fare: isnt there any new info?
<hcf> nothing has changed?
<Fare> dunno, haven't even globally looked at them for a long time (although fixed a few spots)
<Fare> I'd guess they have to be redone completely
<hcf> which will never happen
<Fare> I need a mind-scheduler
<hcf> no pims have been ported to wetware yet
<Fare> pims?
<hcf> personal info managers
<hcf> wetware=brain
09:30am
-:- hcf_ [nef@me-portland-us245.javanet.com] has joined #tunes
-:- SignOff hcf: #TUNES (Leaving)
-:- hcf_ is now known as hcf
-:- lar1 [LARMAN@1Cust25.tnt2.sfo3.da.uu.net] has joined #tunes
-:- ElGato` [no@209-68-229-175.dialup.cust.tfb.com] has joined #tunes
<lar1> If I make an enrty in then GDT that goes out of the 1MB range, and don't turn on the A20 line... will that cause somesort of a fault when I access dataselector:b8000?
<Fare> not a software fault, but maybe wierd hardware faults -- your choice
<Fare> depends also if you use paging
<lar1> Hmmm, then if I triple fault when I switch to Pmode, that points to the GDT as the source of trobule then?
<Fare> sounds like a wierd question. What don't you understand about either physical memory access, linear memory paging, or segmenting?
<Fare> if you triple fault, it reboots
<lar1> What I don't understand is why my code insists on triple faulting when it swithces to protected mode.
<lar1> Mabye you could take a look?  I rewrote it too, leaving out anything non essenial
<ElGato`> :D
<ElGato`> i love you hcf
<Fare> lar1: dichotomy-boot?
<lar1> Fare:  What?
<Fare> lar1: why not just take a project that does that part right, and improve it? (say, retro)
11:00am
<Fare> dichotomy boot is a way to locate any single faulting bug, by rebooting with different values
<Fare> hum, could even be automatized!
<Fare> i.e., by querying you at boot time about the value of the previous attempt...
<lar1> Fare:  I would need to write to disk to do that
<Fare> anyone cares to automate dichotomy-boot?
<Fare> lar1: write to floppy would be ok; or even using BIOS services
<Fare> (i.e. before going PM)
<Fare> all you need is hardwire a sector address (on either floppy or harddrive), and read a bit sequence from it...
<lar1> Fare:  But I know where it faults, I just don't know _why_ it faults
<Fare> oh, so where does it fault?
<Fare> did you disable interrupts?
<Fare> did you setup linear memory correctly? segments correctly?
<lar1> I diabled interupts
<Fare> what about separating those two issues?
<Fare> did you enable A20?
<lar1> No I left out A20 to see if that was the problem, and it wasn't
<lar1> it faults in the code mov eax, cr0 inc cr0 mov cr0, eax
<lar1> Thats why I belive it to be the GDT
<lar1> But it looks good to me
<Fare> lar1: did you check for X86 bugs? Did you read QZX's docs? did you compare to retro's code?
<Fare> lar1: what code is yours?
<Fare> btw, there is no such thing as inc cr0
11:10am
<lar1> I meant inc eax
<lar1> typo
<lar1> I doubt this is an x86 bug... its pretty basic stuff, I'm not quite sure what docs you are talking about on QZX but I have looked at most of them, and I have read some of retros code, difference is they use macros to butli the GDT and I use hex
<Fare> did you   testb $1,1f ; mov %eax,%cr0 ; 1: jmpil *pmcode   ?
<Fare> what processor is yours?
<Fare> did you setup the GDT correctly?
<lar1> No I didn't jump in this version, aren't you suppossed to beable to keep the realmode value in CS for a while?  I am on a p233 mmx, but I have tested on some 486es and 386s
<lar1> I think my GDT is right... I don't know for sure, it looks good to me
<Fare> does it fault at the cr0, or at the jump that follows?
<Fare> did you locate exactly the fault, with hlt?
<lar1> There is no jump
<Fare> which is the faulty instruction?
<lar1> No, I just commented out the three line above and it stoped faulting
<Fare> did you forget to .code32 ?
<Fare> my, I mean, did you do a .code32 ?
<Fare> lar1: pay me consultancy, and I'll do it for you!
<lar1> My segments are set to 16 bit...
<lar1> Heh
-:- SignOff ElGato`: #TUNES (Leaving)
<Fare> 16-bit PM sucks
<lar1> I don't intend to say with the 16 bit selectors
<lar1> I am just tyring to get the basics working, then I worry about 32 bit, paging, etc
<Fare> compare with the way the old tunes code did
11:20am
<Fare> ftp://ftp.tunes.org/pub/tunes/obsolete/oldtunes.tar.bz2
<lar1> Is there a version of bzip for dos?
<hcf> lar1: yes
<lar1> hcf:  thanks
-:- SignOff overfien: #TUNES (Ping timeout for overfien[m240.telcomplus.net])
<hcf> lar1: http://www.rahul.net/dkaufman/
<Fare> alternatively, you can see how retro does it
<Fare> +
<lar1> I don't like the way retro does it
<lar1> Its too modular
<Fare> why not? (I haven't looked at that part of retro)
<Fare> *too* modular????
<lar1> I'd like my GDT to be made in hex, just a handful of lines at the top of my code, not descr <BIGLIM, SMALLLIM SHORTLIM BIGCODE>
<lar1> Besides, it saves space and cycles to do the gdt in hex
<Fare> why not???
<Fare> no, it doesn't save zilch -- all is done at compile time!
<lar1> It is?!?
* lar1/#tunes feels stupid
<Fare> of course, it is!
* Fare/#Tunes hits lar1 with macroprocessing 101
11:30am
* lar1/#tunes hits himself with NASM for dummies
<Fare> lar1: stop doing useless redundant stuff, and come hack retro!
<lar1> I can't make heads or tails of the old TUNES code
<Fare> then come here and ask about it
<Fare> and/or make the code beter
<lar1> Whats this HLL and LLL?
* lar1/#tunes thinks he should lean forth
<Fare> just go to src.25/LLL/i386/
<hcf> lar1: things that'll never exist
<Fare> that's the thing you want to read, and nothing else
<lar1> ok
<Fare> then see in the makefile that the system is split in three parts: boot record, loader, and base system
<Fare> the former are of little interest
11:40am
<lar1> I am reading the boot record
<Fare> so what you are interested in is biosapi.S
<Fare> oh, and it uses m4 as macroprocessor (these were pre-nasm times)
<Fare> (and m4 is more powerful than nasm macros, too)
<Fare> (but m4 is HELL)
<lar1> How long has the tunes idea been raound for?
<Fare> 4000 years
<Fare> uh, I mean, 15 billion years.
<lar1> It hasn't come too far in the past 15 billion years
<Fare> of course! It has already created mankind, and computer!
<hcf> tunes' ideas predate the big bang
<Fare> that was the hardest task.
<Fare> "too make an apple pie from scratch, you must first create the universe" -- Carl Sagan
<Fare> well, it was a hard universe creation, but now, the apple pie is nearly done!
<lar1> Uh huh
<lar1> I don't understand 90% of tunes theory.. reflection and all that... ahhh!
<Fare> did you RTFF?
<lar1> I read, but I always get lost
<Fare> in the FAQ?
<Fare> then you should submit a bug report!
* Fare/#Tunes hits lar1 with collaboration 101
<lar1> A bug report? The docs arent code... they don't execute
<Fare> hcf: of course it does -- ideas are eternal!
<Fare> lar1: they DO execute -- in wetware!
<lar1> hmm
* Fare/#Tunes hits lar1 with memetics 101
<lar1> I think my brains copy of wetware must be too young
<Fare> forget about bootrec.S and loader.S -- they are just LILO-compatible loaders for Tunes.
<hcf> lar1: read the faq, find a spot that doesnt work for u, send a report that states such
<Fare> or rather, setup code that's compatible with being loaded by LILO
<lar1> Fare:  Where is the real boot code?
<Fare> depends on "real"
<lar1> hcf: Ok
<Fare> lar1: a good exercise for you will be to make retro LILO-compatible...
<lar1> Fare:  I can't do that
<Fare> why not?
<lar1> Fare:  I wouln't know where to begin
<Fare> get a look at the linux boot code and/or at the oldtunes boot code, that are compatible
<lar1> Fare:  Uhg
<Fare> (except that lilo and linux evolved, and not the oldtunes code)
<lar1> There aren't any docs on this kina thing are there?
<Fare> understand the interface between lilo and linux (man rdev)
11:50am
<lar1> I understand _Very_ little about the internals of Linux
<Fare> of course there are docs: 1) source & docs for lilo and linux 2) source and docs for other loaders (LOADLIN comes to mind) 3) source and docs for rdev et al
<Fare> lar1: the internals of boot loading and of the kernel are very independant
<Fare> or 4) read the GRUB source and docs
<Fare> or the SOLO stuff
<Fare> gotta go
<lar1> Ok, thanks
-:- SignOff hcf: #TUNES (zzz)
<lar1> I got to eat
-:- lar1 is now known as lar_eating
12:00pm
-:- raffi [user7684@viper.datatrac.net] has joined #Tunes
-:- raffi [user7684@viper.datatrac.net] has left #Tunes []
<lar_eating> Is forth an interpeted language?
-:- lar_eating is now known as lar1
12:40pm
-:- binEng [Anders@j141.ryd.student.liu.se] has joined #tunes
-:- SignOff lar1: #TUNES (Ping timeout for lar1[1Cust25.tnt2.sfo3.da.uu.net])
-:- SignOff binEng: #TUNES ( <k!14>)
<Fare> my!
<Fare> this guy is sooo ignorant!
<smkl> ?
<Fare> abi: C?
<abi> C is the portable assembly language :) or an interpreted language (see EiC, UPS)
<Fare> EiC?
<abi> EiC is a C interpreter at <http://www.pobox.com/~eic>
<Fare> UPS?
<abi> well, UPS is a set of C development tools including a C interpreter at <http://www.concerto.demon.co.uk/UPS/>
01:40pm
<Fare> CL?
<abi> CL is, like, CommonLISP, a fast compiled language (see CMUCL)
<Fare> CMUCL?
<abi> well, CMUCL is the free optimizing CommonLISP compiler originated at the CMU SPICE project <http://www.cons.org/cmucl/>
01:50pm
-:- binEng [Anders@j141.ryd.student.liu.se] has joined #tunes
-:- pyro [tcn@cci-209150250101.clarityconnect.net] has joined #tunes
>>> pyro [tcn@cci-209150250101.clarityconnect.net] requested PING 935789172 277385 from #tunes
-:- SignOff pyro: #TUNES (pyro has no reason)
<Fare> damn, he left!
02:30pm
-:- Downix [down@d-gnaps-64.ici.net] has joined #tunes
-:- eihrul [lee@usr5-ppp151.lvdi.net] has joined #tunes
-:- ElGato` [no@209.68.229.210] has joined #tunes
<ElGato`> you guys go to http://setiathome.ssl.berkeley.edu/
<ElGato`> join my team :D
<ElGato`> get the client
<ElGato`> spare some extra clock cycles
-:- Closing Link: TUNES[bespin.dhs.org] by lackey.openprojects.net (Ping timeout for TUNES[bespin.dhs.org])
-:- 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 bear.openprojects.net)
-:- Your host is bear.openprojects.net, running version u2.10.05.18.(ipcheck4-5) (from bear.openprojects.net)
-:- This server was cobbled together Wed Apr 28 1999 at 12 02:19 EDT(from bear.openprojects.net)
-:- bear.openprojects.net u2.10.05.18.(ipcheck4-5) dioswkfcg biklmnopstv 
-:- [local users on irc(27)] 6%
-:- [global users on irc(191)] 42%
-:- [invisible users on irc(269)] 58%
-:- [ircops on irc(13)] 3%
-:- [total users on irc(460)]
-:- [unknown connections(0)]
-:- [total servers on irc(33)] (avg. 13 users per server)
-:- [total channels created(135)] (avg. 3 users per channel)
!bear.openprojects.net Highest connection count: 47 (46 clients)
!bear.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: www.tunes.org | cybercom.net/~rbjones/rbjpub/logic/jrh0100.htm | www.cb1.com/~john/thesis/thesis.html | www.cs.cornell.edu/slk/ | www.egroups.com/group/os-edge/10.html? | www.cora.justresearch.com/
-:- topic set by hcf [Fri Aug 27 07:38:37 1999]
-:- [Users(#Tunes:12)] 
[ TUNES     ] [ ElGato`   ] [ eihrul    ] [ Downix    ] [ binEng    ] 
[ smkl      ] [ AlonzoTG  ] [ Fare      ] [@Tril      ] [ abi       ] 
[ Crimson   ] [ Plundis   ] 
-:- Channel #Tunes was created at Sun Feb 28 08:48:06 1999
-:- BitchX+Deb1an: Join to #tunes was synced in 7.474 secs!!
-:- Mode change [-s] for user TUNES
<Downix> uh... ok
<Downix> what is it for?
<ElGato`> help search for ET :)
03:30pm
-:- hcf [nef@me-portland-us739.javanet.com] has joined #tunes
<eihrul> el: they don't want help though
<ElGato`> D:
<ElGato`> this gets the work done...im processing  something from january 23
<ElGato`> they recieve more than they can process
<ElGato`> you guys : http://setiathome.ssl.berkeley.edu/
<ElGato`> er
<ElGato`> ooops
<ElGato`> http://setiathome.ssl.berkeley.edu/cgi-bin/cgi?cmd=team_lookup&name=Bueno+Labs
-:- NetSplit: lackey.openprojects.net split from sterling.openprojects.net [03:35pm]
-:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [lackey.openprojects.net]
<eihrul> i thought it was other way around?
<ElGato`> don't read the discription...bueno labs really isn't a group
<eihrul> there was lots of fuss on slasdhot about them processing way too fast relative to how they receive
-:- Netjoined: lackey.openprojects.net sterling.openprojects.net
-:- smkl [sami@MCCCXXV.rdyn.saunalahti.fi] has joined #Tunes
-:- binEng [Anders@j141.ryd.student.liu.se] has joined #Tunes
-:- Downix [down@d-gnaps-64.ici.net] has joined #Tunes
<ElGato`> well gee i still am processing something from 8 months ago
<ElGato`> i think i can see a problem
<ElGato`> of course who knows where seti was 2 years ago
-:- SignOff binEng: #TUNES (Ping timeout for binEng[j141.ryd.student.liu.se])
<eihrul> also on slashdot
<eihrul> there was fuss made about them distrubiting work MULTIPLE TIMES
<eihrul> because of the above
<ElGato`> D:
<ElGato`> heh no wonder
<ElGato`> well you never know if something screws up on one computer
-:- binEng [Anders@j141.ryd.student.liu.se] has joined #tunes
03:40pm
<Fare> hum
<ElGato`> hi fare :D
<ElGato`> i just discovered how limited C macros are
<ElGato`> how bout you?
<Fare> EG: I discovered that long ago
<Fare> happily for me, I used macroassembly before to use C, so I've not been brainwashed
<ElGato`> m4?
<abi> i guess m4 is a horrible macro system
<eihrul> you aren't forced to use CPP with C
<eihrul> it's just standard is all... you can use a non-standard macro processor
<ElGato`> ...
<Downix> hmm
<ElGato`> what if someone with gcc wants to compile some source with weird ass macros?
<eihrul> don't use weird ass macros
<Fare> abi: no, m4, is a powerful macroprocessor that makes you understood why macroprocessing is inherently hell
<Downix> lol
<eihrul> in most cases, you don't need macros at all in C/C++
<eihrul> their only good use is to prevent multiple inclusion
<Fare> eihrul: in most cases, you don't need C/C++ at all
<eihrul> otherwise, using them to augment the language reeks of hack
<Fare> eihrul: you don't know the magic of metaprogramming, do you?
<eihrul> fare: when i find a better language that compiles half as well as C, i'll jump ship
<Downix> eihrul:  Same here
<Fare> macroprocessing is a lame way of metaprogramming that makes a few easy things simple, and all the rest horribly complex
* ElGato`/#tunes is making a bitchen metaheader...in C
<eihrul> right now, that's assembly :)
<ElGato`> but it's really tough
<eihrul> but i'd preferrably like something higher level
<Fare> eihrul: OCAML
<eihrul> i like C syntax somewhat more than the prospect of that
<Fare> eihrul: it compiles within 70% of C speed, and is a much better language
<eihrul> perhaps i'll take a look
<Fare> eihrul: with camlp4, you can choose your favorite syntax for OCAML
<eihrul> but people have said the same about Java... 
<Fare> (although I believe no one was stupid enough to implement a C-like syntax yet)
<Fare> eihrul: not me
<eihrul> i like C's pseudo-tersity specifically
<eihrul> not hungarian notation however
<Fare> First, they said Java was good for app-lets; then they recoiled, and said it was good for serv-lets; actually, what it's good for is toy-lets.
<Downix> eihrul:  Ok, something as fast as C that's as portable as C
<Fare> C is a pseudo language, indeed
<Fare> Downix: C is NOT portable; it's ubiquitous.
<eihrul> but for a pseudo languages, it makes some pretty fast code :)
<Downix> Fare:  True.
<eihrul> fare: is ocaml garbage collected?
* ElGato`/#tunes gets the urge to stop programming in c and give up to scheme
<Fare> abi: C is also not portable but ubiquitous
<abi> okay, Fare.
<ElGato`> no!
<ElGato`> must not!
<ElGato`> grrr...
<Fare> eihrul: yes, and it has perhaps the best GC currently in a language
<Fare> implementation
<eihrul> but can i do explicit memory management as well?
<Fare> EG: try OCAML
<Fare> eihrul: why would you do that? it manages memory better than you can!
<eihrul> perhaps so... perhaps not
<Fare> eihrul: now of course, you can either link to C, or allocate a huge vector...
<eihrul> that too, depends also on the compiler implementation
<ElGato`> fare: what is ocaml like?
<eihrul> one compiler may have an awesome GC, one may have absolutely horribly performing GC code
03:50pm
<Fare> eihrul: typical benchmarks show that GC is as good as malloc/free, and much less bug-prone.
<eihrul> malloc/free themselves are not supposed to be heavily relied upon
<eihrul> you can usually make quite faster special case allocators
<Fare> which leads to library hell!
* ElGato`/#tunes is sick of a huge amount of little things in a language
<ElGato`> c++ is not clean
<Fare> because with multiple C libraries, you never know who must free what
<eihrul> only if it's done naively
<Fare> EG: go OCAML!
<eihrul> el: whoever would think it was?
<ElGato`> c++ es sucio
<Fare> eihrul: no, C memory allocation accross libraries is hell, period
<eihrul> that's why you stay away from that :)
<Fare> eihrul: you can't!
<eihrul> you can't
<eihrul> but you can avoid it as long as possible
<ElGato`> fare: tell me why i should go ocaml over scheme?
<ElGato`> i like scheme
<Fare> eihrul: postponing problems only make them worse when you cannot avoid them anymore
<eihrul> well, if not explit memory management, can i force an the GCing of an object?
<eihrul> -an, s/explit/explicit
<Fare> EG: because 1) it's better for end-users 2) it helps you be more disciplined 3) it has wonderful support for compiler-writing, even if you don't use either ocamllex nor ocamlyacc
<Fare> 4) it DOES have ocamllex and ocamlyacc
<Downix> hmm
<Fare> eihrul: why would you???
* Downix/#tunes is reading up on a companies Amiga clones...
<Fare> eihrul: mu
<eihrul> there's some cases where it's required
<ElGato`> fare: i forgot that url you gave me on ocaml
<Fare> caml.inria.fr
<Downix> nice, someone actually did it
* Downix/#tunes gryns... I KNOW these specs
<Downix> anyone else here an old Amiga user?
<eihrul> fare: curiously, what specifically makes ocaml slower than C, being a static language?
<ElGato`> mabye i shall write it in forth :D
<Fare> eihrul: not having thousands of people trying to superoptimize it, but only a handful of developers
04:00pm
<eihrul> but in theory, in could be nearly as efficient as C?
<eihrul> so long as the promise of better compilers is there for the future, i'll try it :)
<Fare> also, some dynamic and modular design of the implementation
<Fare> in theory it could be *more* efficient than C -- see Stalin, for Scheme
<eihrul> modularity shouldn't impose much overhead in itself
<eihrul> and i though Stalin compiled to C?
<Fare> of course it does; see the way OCAML handles polymorphism
<eihrul> wouldn't that be classified under dynamicism?
<Fare> eihrul: Stalin does compile to C, and gives better code than you'd write in C
<smkl> in theory, you could get huge performance advantages using modular designs
<Fare> eihrul: of course, you couldn't maintain the Stalin output, but that's precisely the point
<ElGato`> fare: how does poplog interface with it's stack
<eihrul> better code than the intermediate Cer, or better code than the advanced C-head?
<ElGato`> is it's stack is execution environment?
<eihrul> fare: so why not compile to the gcc back end at that point?
<Fare> smkl: huge performance advantage in terms of programmer productivity, not in terms of low-level implementation (except in as much as keeping code small to fit in cache/RAM counts)
<eihrul> outputting unmaintainable C code at that point is just pointless
<Fare> EG: mu
<ElGato`> mu?
<abi> mu is at http://charts.unicode.org/unihan/unihan.acgi$0x83ab
<Fare> eihrul: better than any non-crazy Cer
<smkl> Fare: see the paper about optimizing ensemble with prl
<ElGato`> do be honest with you fare i don't get it
<Fare> eihrul: Stalin wins because it writes routines with hundreds of global and static variables, and tens of gotos
<eihrul> which is precisely why Stalin would be better off outputting gcc constraint language
<Fare> eihrul: not something anyone would write in C, but extremely efficient code.
<eihrul> or outputting assembly or perhaps C-- at that point
<Fare> no, Stalin is portable, and takes advantages of compilers better than GCC.
<eihrul> how so?
<Fare> like, the DEC C Compiler, on Alphas.
<Fare> or the Sun Workshop Compiler, on Sparc64's
<eihrul> then why not give stalin itself a back end? :)
<eihrul> that outputs for the appropriate compiler or back end
<Fare> eihrul: are you nuts or what?
<eihrul> generally speaking, maybe
<Fare> Stalin leverages the power of the thousands of lemmings who optimize C compilers.
<Fare> and it's automatically ported to all platforms that have a C compiler.
<smkl> and nobody uses it
<Fare> the backend is whatever cc.
<eihrul> but C itself is not exactly made for being a backend language
<Fare> smkl: at least the author uses it to do number crunching.
<Fare> and authors do, as well.
<Fare> eihrul: then convince people to spend time optimizing C-- instead of C!
<eihrul> could you perhaps send me a choice snippet of outputted stalin C code?
<eihrul> fare: that would be impossible :)
<Fare> eihrul: why don't you just go to http://www.neci.nj.nec.com/homepages/qobi/ and see by yourself?
<eihrul> well, i didn't have that url for starters
04:10pm
<Fare> abi: no, stalin is a _very aggressive_ optimizing scheme compiler at http://www.neci.nj.nec.com/homepages/qobi/software.html
<abi> okay, Fare.
<smkl> i can show code from ghc:
<smkl> IFN_(c16y_entry) {
<smkl> EC_(PrelNum_zdfNumInteger_closure);
<smkl> FB_
<smkl> STK_CHK_NP(4,1,);
<smkl> UPD_BH_UPDATABLE(R1.p);
<smkl> PUSH_UPD_FRAME(R1.p,0);
<smkl> R1.p=(P_)&PrelNum_zdfNumInteger_closure;
<smkl> Sp[-4]=(W_)((P_)&c17a_info);
<smkl> Sp=Sp-4;
<smkl> JMP_((P_)(ENTRY_CODE((D_)(*R1.p))));
<smkl> FE_
<smkl> }
<eihrul> that is scary...
<eihrul> not only does it look hard to read, but amazingly suboptimal :)
<ElGato`> it makes me want to cry
<Fare> eihrul: why suboptimal? that's typically the kind of thing that the C compiler can optimize easily!
<eihrul> that code, at first glance, looks as if it's emulating mechanisms that you can express natively in assembly
<Fare> eihrul: and of course, stack checking might be done by page-trapping, depending on the macros...
<Fare> eihrul: again, who cares if the C code is suboptimal, when this suboptimality is wiped away by the C compiler?
<Fare> you're just being childish
<eihrul> gcc does a really poor job of that, unfortunately
<eihrul> sure, other compilers may optimize it away, but i don't quite think gcc would do the best job at it
<eihrul> er, excuse the redundant clause there
<Fare> so what? the interest of producing C code is precisely to be as portable and efficient as C, not just as portable and as inefficient as a particular C implementation
<Fare> get DEC's cc
<Fare> or VC++
<Fare> or Sun's cc
<eihrul> A) i can't afford the, B) i can't afford the systems they run on
<Fare> or Watcom C
<eihrul> er them
<Fare> or CodeWarrior
<Fare> or whatever!
<Fare> anyway, GCC is not the be-all end-all of C compilers.
<Fare> even though it is a honorable one.
<eihrul> it's the only one easily available to me
<Fare> no, there's lcc, also
<Downix> lcc
<Downix> ?
<Fare> or pcc
<eihrul> lcc admits to compiling worse than gcc
<smkl> i can paste the asm code also
04:20pm
<Fare> so what? I contradicted your statement, that's all that counts.
<Fare> smkl: only if it's short
<eihrul> you've not contradicted my statement
<Downix> eihrul:  what is lcc?
<eihrul> you've managed to contradict something, however
<Fare> or put it on one line, separated with ; 
<Fare> <eihrul> it's the only one easily available to me
<smkl>         leal -16(%ebp),%eax
<smkl>         cmpl %eax,88(%ebx)
<smkl>         jbe .L161
<smkl>         jmp stg_gc_enter_1
<smkl>         .align 4
<smkl> .L161:
<smkl>         leal -12(%ebp),%eax
<smkl>         movl $Upd_frame_info,(%eax)
<smkl>         movl 84(%ebx),%edx
<smkl>         movl %edx,4(%eax)
<smkl>         movl %esi,8(%eax)
<smkl>         movl %eax,84(%ebx)
<smkl>         movl $PrelNum_zdfNumInteger_closure,%esi
<smkl>         movl $c17a_info,-16(%ebp)
<smkl>         addl $-16,%ebp
<smkl>         jmp *(%esi)
<smkl> ....
<hcf> smkl: thanks for not listening to Fare
<ElGato`> yuck!
<smkl> this is short for ASM
-:- SignOff Downix: #TUNES (Read error to Downix[d-gnaps-64.ici.net]: EOF from client)
<Fare> smkl: what was the source for that procedure?
<eihrul> fare: perhaps i should have said best one easily available
<smkl> this seems to be somekind of helper procedure
<Fare> "best" is a relative term
<eihrul> and it's relative to my meaning of the term
<Fare> anyway; these compilers writer do not work for your eyes only
<Fare> as strange as may seem
<Fare> about mozilla: what's the use of a display engine that's twice as fast, if it must draw every page three times before it stabilizes?
<eihrul> i'd have to say the twice as fast thing
<eihrul> bugs can be cured, performance penalties usually can't
<Fare> x*2/3 = x/1.5 < x
<eihrul> 2 + 2 == 4
<Fare> zéro plus zéro égale la tęte ŕ toto...
04:30pm
<eihrul> the point of such being?
<Fare> unlambda is a silly language based on call-by-value combinatory logic at http://www.eleves.ens.fr:8080/home/madore/programs/unlambda/
04:40pm
<ElGato`> bye
-:- SignOff ElGato`: #TUNES (:D)
-:- Downix [down@d-gnaps-88.ici.net] has joined #tunes
-:- SignOff smkl: #TUNES (sleeping ....)
04:50pm
* hcf/#tunes is away: (afk 1hr)
-:- lar1 [LARMAN@1Cust47.tnt2.sfo3.da.uu.net] has joined #tunes
-:- liar [brand@p0wer.qzx.com] has joined #tunes
<lar1> Hey liar
<liar> hey lar1
<lar1> Could you mabye take a look at some code?  Real short :
<liar> after i eat
-:- SignOff eihrul: #TUNES (Ping timeout for eihrul[usr5-ppp151.lvdi.net])
<lar1> ok, _thank you_
<liar> www.egroups.com/group/os-edge/10.html  <-- whose page?
-:- eihrul [lee@usr5-ppp151.lvdi.net] has joined #tunes
05:20pm
-:- lar1 is now known as lar_away
-:- SignOff binEng: #TUNES (Ping timeout for binEng[j141.ryd.student.liu.se])
-:- SignOff AlonzoTG: #TUNES (Have Nice Day :))
<hcf> liar: the tao os dude's mlist
-:- AlonzoTG [Alonzo@216-164-233-235.s616.tnt8.lnh.md.dialup.rcn.com] has joined #tunes
06:10pm
<Fare> what's the name for guys that the mafia send to clean your shop when you don't pay protection money?
<liar> hcf: oh is that an mlist archive?
<hcf> liar: yeah, the tao dude started the os-edge mlist
<liar> ah that would explain why it looked like the dumbest web page around
<hcf> liar: thats egroups.com in general
<hcf> as sucky as all the other mlist archive sites
<liar> well ya but i thought it was a web page and couldnt figure out why he had it separated into 16 pages without an index and random comments scattered around
<liar> being an mlist archive its ok
<hcf> abi: tao?
<abi> tao is, like, an expression of the Water or tao, the (so-called) ultimate os at http://www8.pair.com/mnajtiv/tao.html
<hcf> thats the tao page
<liar> ya been there
06:20pm
<liar> Have you ever sniffed your finger because you couldn't remember if you used it the night before?
<Downix> uh... no
<liar> wtf kinda question is that?
<Downix> it's a lezbian one?
<liar> must be
<liar> Have you ever smoked Mary Jane, herb, pot, dope, a pinner, a blunt, a freakin' cannon, puff puff pass, marijuana,
<liar>        or hash?
<Downix> nope
<liar> Have you ever seen a pornographic movie?
<Downix> yes
<liar> are rated-R shows with breast shots pornographic?
<Downix> nope
<liar> Have you ever performed fellatio?
<Downix> you mean in the past 24-hours?  sure
<Downix> 8)
<liar> Have you ever performed fellatio (oral sex on a man)?
<liar> yer sick
<Downix> I'm joking
* Downix/#tunes doesn't exactly swing that way
<liar> hmmm have i ever cheated on an exam
<liar> would making the instructor change the answers be cheating?
<Downix> where are you getting this from?
<liar> purity test
<liar> http://test.thespark.com/puritytest/purity.cgi
<Fare> is haschisch the same as mary jane, or related, or completely different?
<liar> ive talked a few instructors into changing "the correct answer" to what i had
<liar> actually i had one instructor that used my answers to grade everyone else
06:30pm
<liar> pissed the class off one time when i had one wrong
<Downix> lol
<Fare> hashish is hemp, actually
<liar> Please type in the meanest word you've ever called someone:
<liar> thats a toughy
<Fare> is this #stupidtest ?
<liar> it measures how pure u are
<Downix> meanest word:  Limbaughian!
<liar> Do you regularly not make it to the bathroom?
<Downix> eww!
<liar> Have you ever played dress up... in that way?
<liar> are u taking it?
<Downix> actually, no
* Downix/#tunes should, but he'd flunk it
<liar> Have you ever had the kind of sex where you put something in the "out" hole? (If you don't know what that
<liar>        means, move along, move along)
<liar> Have you ever kissed someone in your family... in that way?
<liar> Have you ever fantasized about a family member?
<liar> i dont know if i wanna be graded on this test
<Downix> hehe
<liar> it was made for fags and shit
<Downix> liar:  if it makes you feel better, I definately don't want to.
<liar> Do you know what urine tastes like?
<Downix> liar:  that is disgusting!
<liar> Have you ever walked in on your parents having sex?
<liar> Did you join in?
<Downix> kinda impossible
* Downix/#tunes smiles
<liar> u joined them?
<Downix> no, impossible to walk in on them, and even more impossible to join in
<Downix> my parents were divorced before I was born
<liar> how is it impossible?
<liar> uhh
<Downix> and didn't see each other again until last year
<liar> so yer a bastard child?
<Downix> right
<liar> Have you ever intentionally burned yourself?
<liar> damn i had to answer a yes
<liar> thats gonna hurt my grade
<Downix> hey, I would have to answer yes too
06:40pm
<liar> i cover my hand with vaseline and then light it on fire
<Downix> same here, I did it for a special effect in a student film I was making
<liar> if u do it right so no one sees the vaseline then they will think u burned up yer hand
* Downix/#tunes also set gas-soaked tennis balls on fire and threw them around, looked awesome on tape
<liar> i have screwed up a couple times tho
<Downix> oops
<liar> i lit my shirt on fire once to for halloween
-:- Downix_ [down@d-gnaps-88.ici.net] has joined #tunes
<Downix_> oops
<Downix_> I hate that cntr+x is right next to cntr+c
<liar> Have you ever licked someone's eyeball or had yours licked?
<Downix_> there
<Downix_> uh, I've had mine licked
<liar> yer eyeball?
<liar> or eyelid?
<Downix_> eyeball
<liar> damn
<liar> i cant even wear contacts
<Downix_> hurt like a sonofabitch
<Downix_> it was a cat which did it
<liar> aint no one gonna stick their tongue in my eye
<Downix_> smart man
-:- SignOff Downix: #TUNES (Ping timeout for Downix[d-gnaps-88.ici.net])
<liar> Have you in the past five years defecated in your pants?
<Downix_> no way dude
-:- Downix_ is now known as Downix
<liar> Have you in the past five years intentionally defecated in your own pants?
<Downix> no way dude
<Downix> gross
<liar> i wanna meet these ppl that intentionally shit their pants
<liar> what the hell is the purpose of that
<liar> if i made this test it would have 1 question: do u think yer a fscking luser or not?
<liar> Have you ever been turned away at a blood drive because of your own dirty past?
<liar> ok i was turned away twice cuz of low T cell count
<liar> does that count?
<Downix> I was turned away due to blood poisoning
<Downix> My appendix burst
<liar> i have a very weak immune system but that cant count as a dirty past
<Downix> nope
<liar> heh
<Downix> I have a strong immune system, but I'm allergic to a ton of foods and medications
<liar> i have no allergies
06:50pm
<Downix> be glad
<Downix> eating out sucks for me
<liar> well i do have a joint disorder
<Downix> fun
<liar> imagine having arthritis at the age of 18
<Downix> topic of the day:  bodily disfunctions
<Downix> ow
<liar> the best is when my jaw joint swells and pushes on my ear drum
<Downix> oh man, I've had that happen
<Downix> if I eat sea-food my joints all swell
<AlonzoTG> om
<liar> for about 8 hours everyday of yer life?
<Downix> no, for about 4 days straight
<liar> i have to take a gram of naproxen everyday
<liar> and 2 grams in the winter
<Downix> I'm glad my problems are caused by external elements
<Downix> I just have to avoid a lot of things
<liar> mine are caused by external elements too
<Downix> oh?
<Downix> oh, the reduced immune system, right
<liar> cold, temperature change and excessive usaged of the joint
* Downix/#tunes can avoid his allergies if need be
* Downix/#tunes nods
* Downix/#tunes thinks liar needs to be placed in a bubble
<Downix> 8)
<liar> if i sit in a warm room all day and do nothing than it doesnt swell
<liar> ofcourse chewing my food is excessive jaw usage
<liar> so i only eat twice a day
<Downix> oh man
<liar> 86% Pure!
<Downix> Well, they say no matter what you problems are, there's someone else that has it worse
<liar> if only that could comfort me :)
<liar> out of 3,254,958 ppl only 4% are more pure than i
<Downix> fun
07:00pm
<liar> if anyone sees eihrul tell him im going hunting and will be back monday
<liar> cya
<eihrul> i'll let him know
-:- SignOff liar: #TUNES (BRiX [http://www.qzx.com/brix] :: gone)
-:- liar [brand@p0wer.qzx.com] has joined #tunes
<liar> :)
<eihrul> netscape was not cooperating with me
<liar> eihrul: scared ya didnt i :)
<liar> uh ya right
<eihrul> doth it suck too much?
<liar> uhh
<liar> u want to trap ALL memory access?
<liar> that will make it DAMN slow
<liar> nt traps all io that is why nt is slow
<eihrul> good? bad? say something :P
<liar> if nt trapped all memory access then it would be unusable
<eihrul> it doesn't trap all memory accesses
<eihrul> it only traps the first access
<eihrul> after the address is loaded into the TLB, you won't get traps
<liar> well a faster way would need 3 pages
<liar> no that wont work
<eihrul> why 3?
<liar> when u change cr3 u also need to reset all the values in the page table
<liar> so u have a really high context switch
<eihrul> no you don't
<eihrul> the state of every entry in the table
<liar> and what if program A accessed a byte at 4meg and then a byte at 20meg?
<eihrul> when not in handler is "not present"
<eihrul> exactly what should happen...
<liar> is there a way to set a tlb entry?
<eihrul> unless you're using 21MB pages, wouldn't be a problem :)
<eihrul> well, the way i described should basically set a tlb entry
<eihrul> it's not the most optimal, but from the docs i've read, should hackishly do it :)
07:10pm
<liar> oh u reset the page entry after setting it
<eihrul> yessir
<liar> hmm
-:- hcf_ [nef@me-portland-us333.javanet.com] has joined #tunes
-:- SignOff hcf: #TUNES (Leaving)
-:- hcf_ is now known as hcf
<liar> hmm
<eihrul> thas what i was thinking :)
<liar> i might try that in brix
<Downix> speaking of brix...
<eihrul> i'd be interested to know how it effects performance, so that'd be a good thing
<liar> Downix: sunday
<Downix> ok
<liar> eihrul: but i need to finish 0.0.1 first
<Downix> I have sunday off
* Downix/#tunes sighs, finally, a day off
<eihrul> if you try it at all i'd be happy :)
07:20pm
<liar> i will be writing some stuff to test kernel speed for 0.0.1 and then i can use that to compare another kernel using software tlbs
<hcf> abi: CRML is Compile-time Reflective ML at http://www.cse.ogi.edu/~sheard/crml.html
<liar> if the software tlb kernel isnt much slower than ill use it
-:- infotek [infotek@fol-tco.ametro.net] has joined #tunes
<liar> or i can always have 2 kernels and let the user pick: speed or memory
<Downix> liar:  I do that with Linux already.  I have 3 lilo options:  network, dial and media
<liar> Downix: thats dumb :)
<Downix> hehe, saves me loading up useless functions
<infotek> does ne body know of a player for mpeg 4 ?
<liar> i load my kernel once every few months so load time is irrelevant and i have plenty of memory so the 100k or whatever u might save doesnt matter
<liar> infotek: there is one at thespark.com
<infotek> are they gonna revamp mpg123 ??? 
<Downix> I reboot once a month, but I also like to have the options JIC
<liar> they already did, its called mpg1234
<infotek> where can i get it?
<liar> thespark.com
<infotek> i'll just shutup now...
<infotek> ;>
<liar> good idea
07:30pm
<infotek> hehe
<infotek> i got it 
-:- SignOff infotek: #TUNES (infotek has no reason)
-:- SignOff Downix: #TUNES ([BX] I theenk I need a beeger box!)
07:50pm
-:- pyro [tcn@cci-209150250081.clarityconnect.net] has joined #tunes
<pyro> hey
<abi> what's up, pyro
<hcf> pyro: hoy
08:00pm
<pyro> is anyone else here?
<liar> hey pyro
<pyro> howdy
<abi> what's up, pyro
08:10pm
<pyro>  /kick abi
<liar> heheh
<AlonzoTG> om
<pyro> I was thinking of getting into radio again
<liar> AlonzoTG: u got any relatives named Rachel Grimes?
<liar> pyro: why would u do that?
<pyro> mostly to mess w/ radio networking.. if anyone else does it anymore
<AlonzoTG> nope =\
<pyro> you think the internet killed it?
<liar> no
<liar> AlonzoTG: do u have a lot of criminals in yer family?
<AlonzoTG> nope
<liar> dont lie to me
-:- SignOff AlonzoTG: #TUNES (Have Nice Day :))
<liar> i knew i could make him leave :)
<pyro> no I did
<pyro> a couple weeks ago I told him he irc's way too much
08:20pm
<liar> i wonder if he types that message wrong each time he quits or if he just typed it wrong into the script and is too dumb to fix it
-:- SignOff lar_away: #TUNES (Ping timeout for lar_away[1Cust47.tnt2.sfo3.da.uu.net])
<liar> er not dumb just not imformed cuz no one likes him so they wont tell him
<pyro> haha
<pyro> hey, between a 350, 400, and 450, which is best?
<liar> eh?
<pyro> cpu
<liar> uhh faster is better
<pyro> no bus speed problems like with 133/150/166?
<liar> no
<liar> they all use a 100mhz bus
<pyro> what about the rambus?
<liar> what about it
<pyro> is that 200mhz?
<liar> uhh
<pyro> ram ain't that fast yet is it
<liar> no
<Fare> pyro!
<liar> the k7 supports a 200mhz bus
<Fare> pyro: what's your threading model for retro forth?
<Fare> is that call-threading?
<pyro> Fare: yeah, or subroutine-threaded
<Fare> : A B C D ; will be compiled into A: call B ; call C ; call D ; ret ?
<pyro> yup
<Fare> ok
<Fare> I'll be writing a kernel lisp
<Fare> trying to complete the acronym into KLUGE
<pyro> hehe
08:30pm
<Fare> but having a hard time at it
<pyro> what have you got so far?
<pyro> i know more english than u :)
<Fare> not much: the overall design, and the decision that my cross compiling platform would be CL rather than Scheme
<Fare> (so as to have more infrastructure on which to rely)
<Fare> I'm beginning with a simple pattern matcher, using CLOS to define a "pattern" class
<Fare> pyro: Kernel Lisp U. G. E.
<Fare> (optional D. between U. and G. accepted)
<Fare> Unlike GNU Elisp?
<pyro> is it supposed to run in retro?
<eihrul> is your lisp supposed to be reactionary?
<Fare> yup, one backend for retro, another for a linux process and/or for clementine
<Fare> reactionary????
<Fare> what's that?
<pyro> my 'retro - the reactionary os' thing, on my web page
<Fare> ok
<eihrul> as in, elisp is not good enough, so KLUGE was made as a reaction to that fact :)
<Fare> yeah, pretty reactionary -- at least to begin with!
<eihrul> then the name could work... but what about if GNU Elisp fades into obscurity?
<pyro> so what?
<eihrul> you'd end up with an FVWM situation
<pyro> what's FVWM stand for?
<eihrul> noone would really know what the acronym stood for completely
<Fare> what's the expression? offer and demand? supply and demand?
<pyro> supply
<eihrul> the latter
<pyro> hey-- what's eihrul mean?
<eihrul> absolutely nothing
<pyro> sure
<pyro> how about fvwm.. what's that mean?
<eihrul> the vwm being virtual? window manager
<eihrul> but even the author doesn't know what the F is for anymore
<pyro> oh
<pyro> is it descended from fwm?
<eihrul> shouldn't be
<pyro> FORTRAN window manager? heh
<pyro> anyway, fare..
<pyro> I was thinking of a way to blend forth and LISP..
<pyro> have them share the same stack & heap
08:40pm
<pyro> in Forth, ( ... ) will no longer be a comment, but a S-EXP..
<liar> heh the athlon is in a slot1 style case
<Fare> feeble virtual window manager
<Fare> franz' vwm
<Fare> free vwm
<Fare> foo vwm
<pyro> your ISP sucks fare
<Fare> pyro: have you seen poplog?
<Fare> abi: poplog?
<abi> poplog is now freely available at ftp://ftp.cs.bham.ac.uk/pub/dist/poplog/freepoplog.html
<Fare> it's precisely a blend of FORTH and LISP, although it was invented at the same time as FORTH, and at a time LISP was much more primitive than it is.
<pyro> cool
08:50pm
<Fare> Note that its stack is GCed.
<Fare> I'd like a slightly different, reflective approach:
<Fare> for a "same" language, have several slightly different backends, that have different memory management
<Fare> so that KLUGE could be used to produce transient, non-GCed, "real-time" code suitable for device drivers,
<Fare> as well as persistent, GC, "non-real-time" code for normal userland, at the other spectrum
<Fare> and anything in between
<Fare> (that's at least 8 combinations)
<pyro> ahh, ok
<eihrul> fare: no gc? are you okay?
<Fare> eihrul: "no gc?"? are you okay?
<eihrul> that was the question
<Fare> which? that was two question.
<eihrul> the first one was rhetorical
-:- SignOff liar: #TUNES (BRiX [http://www.qzx.com/brix] :: sleep)
<eihrul> just n'm already
<pyro> what are priority heaps/queues good for, besides scheduling?
<eihrul> aren't circular lists better for scheduling?
<pyro> if you want all tasks to be equal
09:00pm
<eihrul> my scheduler maintains several circular lists
<eihrul> for each priority level
<Fare> gotta go sleep
<pyro> see ya
<Fare> pyro: you can schedule something else than processes
<Fare> pyro: or something more local than "global processes", at least
<pyro> scheduling network traffic..
09:10pm
<pyro> I really need to get a copy of AOCP.. it's a good reference..
<eihrul> aocp?
<pyro> art of computer programming
<eihrul> sounds vaguely familiar for some reason, isn't that a knuth book?
<pyro> yup
<pyro> nothing else comes close
<eihrul> what does it cover?
<pyro> lots of algorithms & data structures
<pyro> lots of tips, too
<eihrul> does it cover more than sedgewick?
<pyro> oh yeah
<eihrul> hrmm
<pyro> i don't think sedgewick even covered anything that wasn't around in 1973
<eihrul> sedgewick was pretty thorough nonetheless
<pyro> knuth goes into more depth
<eihrul> will have to check the book out once i get some cash
<pyro> it's like 3000 pages
<eihrul> didn't it come in a set?
<pyro> he only published volumes 1-3 in 1973, afaik
<eihrul> so i'm assuming you're talking about a more recent volume?
09:20pm
<pyro> no, that's it
<pyro> :)
<eihrul> so... then why is sedgewick covering algorithma as of 1973 bad? :)
<eihrul> if that's all knuth explains anyway?
<pyro> sedgewick was about 15 years later, but the algorithms didn't change.. still haven't..
<eihrul> yeah... but i just wanna know what knuth explains beyond sedgewick :)
<pyro> cooler things :)
<eihrul> such as?
<pyro> well, sedgewick devotes a section to sorting & searching.. knuth wrote a whole volume on it
<pyro> besides, Knuth took a more realistic approach.. he used a fictional assembly language, whereas sedgewick used pascal
<eihrul> i have the C++ version
<pyro> that's better
<eihrul> and seeing as C++ is a high level assembly language
<eihrul> but the example code he gives really, really stinks anyway
<eihrul> the explanation were the real relevent parts of the book
<pyro> it's a decent book, and less intimidating..
<pyro> to your brain and your wallet :)
<eihrul> well, i can't just buy one volume
<eihrul> i have a book fetish
<eihrul> i MUST get the whole set
<pyro> I wish you luck finding vols. 4 thru 7
<eihrul> ack, there's 7
<eihrul> there goes my $400
<pyro> you found it?
<eihrul> at the worst, amazon should have it
<eihrul> barnes and nobles might
<pyro> I checked amazon a couple months ago.. it was about $135 for vols. 1-3
<pyro> 4-7 are vapor
09:30pm
<pyro> he planned them, listed them in the table of contents, but gave up actually writing them.. they were about things like compilers, which keep changing
<hcf> abi: knuth?
<abi> i think knuth is at http://www-cs-staff.stanford.edu/~knuth/index.html
<pyro> how nice :)
<eihrul> pyro: that's like a fundamental law of computing science it seems
<eihrul> half of everything is vapor...
<eihrul> but a good 3-4 books on compilers would have been cool
<pyro> heh.. he says he's still working on 4 :)
<pyro> he needs help
<pyro> hey i know someone who speaks Telagu!
<eihrul> what kind of language is that?
<pyro> no wonder 4 isn't done yet, he's trying to list everyone's name in their native language, not latinized.. :)
<pyro> Indian
<pyro> east indian
<eihrul> names schmames, dude needs to just put the algorithms in... a little latinization couldn't hurt
<pyro> I wouldn't mind helping him design a new fictional machine
09:40pm
<eihrul> why? he could just use the merced
<pyro> nononon.. RISC/MISC
<eihrul> there be not a machine more fictional :)
<pyro> haha
<pyro> a fasicle?
<eihrul> no, wait... he can use transmeta or amiga :)
<pyro> oh
<pyro> you can download some of the text
<eihrul> cool, he pays for errors
<pyro> haah
09:50pm
<pyro> you download any .ps files yet?
<eihrul> nope
<eihrul> still looking around
<pyro> hmm.. Catch-22 quotes
<pyro> my god.. 4 more years!
<eihrul> for what?
<eihrul> the dude writes some really ugly code, however
<pyro> I like the old edition better :)
<eihrul> the concept of variable names is kinda lost on him, it seems
<pyro> huh? you mean he uses  short variable names, like a real person?
<pyro> :)
<eihrul> umm, he uses 1 letter variable names
<eihrul> then uses way too many concepts
<eihrul> er comments
<pyro> hmm.. I prefer a compromise
<eihrul> to compensate for the fact
<eihrul> i prefer to try and make my code self-documenting, and totally neglect comments >:)
10:00pm
<pyro> i don't like this new MMIX fictional machine
<eihrul> hrmm, which ps are you look at about it?
<pyro> fascicle 1
<pyro> heh, I should write my own treatise on hacking after I've been around awhile :)
<eihrul> but, the dude taught at stanford
<pyro> make it 
<pyro> ..
<pyro> i wonder if all that programming knowledge couldn't be condensed more, without leaving things out
<pyro> drop the fictional assembly language, use pseudocode and boolean algebra
<pyro> leave little seeds of knowledge but make the reader work hard to figure things out
<eihrul> that would prolly help
<eihrul> hard is relative to reader though
<pyro> these are the kind of ideas you can't just spoon-feed.. the reader has to do a lot of thinking no matter how much explaining the author does
10:10pm
<eihrul> sure, they have to do some thinking
<eihrul> but if you're relatively familiar with lots of other concepts, you can associate much of the laborious previous thinking
<eihrul> and avoid lots of new thinking like a lazy bastard all the same :)
<pyro> hey.. you know where I can find a collection of clever hacks? (probably in assembler)
<eihrul> seeing as much of the conceptual stuff is recycled/mismash of other stuff
<eihrul> the Obfuscated (Insert Language Here) contests are pretty good for that
<pyro> maybe an assembler contest?
<eihrul> obfuscated c one had some goodies
<pyro> oh.. I bet those little graphical demos are good
<pyro> you know, the ones with light-shaded 3D and music, in 700 bytes..
<eihrul> jah
<pyro> haha
<eihrul> perhaps maybe look at Ritchie's C compiler
<eihrul> that was supposedly a monster hack
<pyro> you mean the one that put a trojan in the 'login' program?
<eihrul> nah
<eihrul> the one posted on /. not to long ago
<pyro> hehe
<pyro> "In the 1960s I would probably have chosen Algol W; in the
<pyro> 1970s, I would then have had to rewrite my books using Pascal; in the 1980s, I would surely have changed everything to C; in the 1990s,
<pyro> I would have had to switch to C++ and then probably to Java."
<eihrul> that knuth?
<pyro> yeah
<pyro> explaining why the fictional machine
<eihrul> x86 is very backwards compatable
10:20pm
<pyro> "Therefore I will continue to use English..."
<eihrul> he could have just used it 
<pyro> x86 is UGLY
<eihrul> just think of it as enforcing your earlier spoken principle
<eihrul> force the reader to think harder :)
<pyro> I don't have a problem with the x86's style.. but it's got too many quirks
<eihrul> and that's half the fun
<pyro> it's not organized like the 68000
<eihrul> it's like some really fscked up game
<eihrul> epic is going to be even more quirky
<pyro> speaking of fsck.. why does it take so fscking logn?
<eihrul> it has to do lots of stuff, it might even do a few passes
<pyro> DOS does it pretty quickly
<eihrul> to verify state of filesystem
<eihrul> DOS also uses a diff file system
<eihrul> elf versus fat
<pyro> I just wish it would do the long part (looking for bad sectors) in it's spare time
<pyro> not immediately after you boot it
<eihrul> it only does it if you don't shutdown properly
<eihrul> so it isn't that bad
<pyro> shit happens
<eihrul> i can stomach waiting 4-5 minutes for a boot every half a year or so
<eihrul> just take a piss or something meanwhile
<pyro> I want a <5 second boot time
<eihrul> tough luck
<eihrul> mount FS read-only or something
<pyro> oh I'll do it in my OS
<eihrul> boot time is irrelevent though
<pyro> well, code size makes a huge difference
<eihrul> but that's a general optimization independent of boot time
<pyro> DOS 3.1 on a 4mhz PC, vs. Win NT 4.0 on a 400mhz P3
10:30pm
<eihrul> that's a bad comparison
<pyro> the sad thing is, those old XT's and AT's did what we needed, quicker, more reliably
<eihrul> sure... but DOS was also single-tasking, no memory protection, 16 bit, etc 
<pyro> it was more stable
<eihrul> they did what you needed quicker, so long as you only needed to do one thing
<pyro> they didn't pretty it up so much
<eihrul> you could make a single tasking operating system nowadays
<eihrul> and it would probably be as fast if not considerably faster (relatively speaking)
<pyro> jesus christ! this guy has a pipe organ in his house!
<eihrul> hey, the dude writes books... that's good enough for me
<pyro> hehe
<pyro> the thing is huge!
<eihrul> personally, i'll trade a little speed for ability to run 2 irc clients
<eihrul> you could probably make things even faster by relegating memory protection to the compiler
<pyro> i'm talking about careful programming, not sacrificing the advantages of task switching and paging
<eihrul> you mean speed/memory hacks? :)
<pyro> assembly
<eihrul> jah, speed/memory hacks
<pyro> not trade-offs
<eihrul> not that those are a bad thing
<eihrul> i don't mean trade offs either primarily
<eihrul> but trade offs are to a lesser extent part of it
<eihrul> if you can have your cake and eat it to, that's a good thing
<eihrul> but rarely can you
10:40pm
<pyro> speed/memory vs. programming time :)
<eihrul> but if fare were here, he'd probably say go for the second
<pyro> what if you program it in java (pretend you're a moron :), but that's too slow, so you reprogram it in C, still too slow, so you finally rewrite parts of it in assembler
<pyro> or write better algorithms instead of using library routines
<eihrul> the latter being way better
<pyro> for example, why use a generic searching/sorting routine when you can fit specialized machine code in the same space required for calling the generic ones..
<pyro> maybe you could just write everything in a high level language, if AI wasn't bullshit
<eihrul> but generally, optimizing the conceptual algorithm itself will yield much greater speed boosts
<eihrul> than a clever implementation of a suboptimal algo
<eihrul> actually, the two could be the same in ways :)
<pyro> and focus your optimization where it's gonna count
10:50pm
<eihrul> also da fact that once you have an optimal assembly algorithm
<eihrul> there's no where to go but conceptualization
<pyro> what if you come up with a great concept that doesn't *have* an efficient implementation in assembly?
<eihrul> it ain't efficient then
<eihrul> i'd say efficient is relative to it's applicability to the hardware
<eihrul> maybe it could be simple/viable/etc, but certainly not efficient
<eihrul> and failing that, there's hardware acceleration :)
<pyro> hey, is this what you do for a living?
<pyro> not irc :)
11:00pm
<eihrul> is what what i do?
-:- SignOff pyro: #TUNES (Ping timeout for pyro[cci-209150250081.clarityconnect.net])
-:- SignOff eihrul: #TUNES (Leaving)
-:- eihrul [lee@usr5-ppp151.lvdi.net] has joined #tunes
11:10pm
-:- SignOff hcf: #TUNES (Ping timeout for hcf[me-portland-us333.javanet.com])
-:- SignOff eihrul: #TUNES (Leaving)
-:- lar_away [LARMAN@1Cust153.tnt22.sfo3.da.uu.net] has joined #tunes
-:- lar_away is now known as lar1
<lar1> Fare:  I want to submit a bug report.  To what address do I send this?
11:40pm
-:- eihrul [lee@usr5-ppp91.lvdi.net] has joined #tunes
<lar1> Hey eihrul
<eihrul> hey
<lar1> Do you know where to submit TUNES bug reports?
<eihrul> nopers
<lar1> My code doesn't work and I don't know why and I don't know how to fix it and this is a bad sentence
<eihrul> that's a bad thing
<eihrul> ya needs to know why your code works :)
<eihrul> so the inverse can be true
<lar1> I know how its _supposed_ to work
<eihrul> so then find out why it doesn't
<lar1> I have tryed
<lar1> Its really annoying because noone around me can help, mom doesnt know dad doesn't none of my firends do... 
<lar1> sucks
[msg(TUNES)] newlog 1999.0828
IRC log ended Sat Aug 28 00:00:01 1999