IRC log started Mon Jul 19 00:00:00 1999
[msg(TUNES)] permlog 1999.0719
-:- pooh0 [pooh0@un-152-2.university.indiana.edu] has joined #Tunes
-:- pooh0 [pooh0@un-152-2.university.indiana.edu] has left #Tunes []
-:- SignOff Fare: #TUNES (Ping timeout for Fare[quatramaran.ens.fr])
!wichert:*! WARNING: verne.openprojecst.net will go down for harddisk-change in 5 minutes. Scheduled downtime is 5 minutes
!wichert:*! verne.openprojects.net going down now
!wichert:*! verne.openprojects.net going down one last time
!wichert:*! verne is back online and working fine now. Our apologies for any inconveniences
<Crimson> Hi guys!
06:20am
<Crimson> abi: chaos?
<abi> i heard chaos was yet another micro/nano-kernel based os at http://www.chaosdev.org or small, fast and secure, blah blah *snore*
06:50am
<Plundis> fett..
07:00am
<Crimson> Plundis: in english please... ;)
07:20am
<Plundis> yeah, yeah.. :)
<s_rr> ?
07:30am
-:- s_r [sr@phila-dialup379.nni.com] has joined #tunes
-:- SignOff s_rr: #TUNES (Ping timeout for s_rr[phila-dialup396.nni.com])
<Crimson> s_r: hi.
08:20am
<s_r> hey
<abi> niihau, s_r
<s_r> how's work on chaos?
<Crimson> s_r: pretty hard work right now. working on the threading routines.
<Crimson> implementing support for process threading and kernel threading.
<s_r> what kind of problems are you having with the threading?
<Crimson> nothing special, just that i have to write some code and have to think... ;)
<s_r> hmm
<Crimson> i dont know if storm will make any use of the kernel-threads, but it was automatically supported.
-:- smoke [smoke@14dyn193.delft.casema.net] has joined #Tunes
<smoke> hello
<Crimson> abi: Tunes?
<abi> i heard Tunes was http://www.tunes.org, a free reflective computing system or for programming languages what the internet is to networks or like flypaper for hallucinogenic programmers or discombobulated or obnubilated
<Crimson> abi: alim
<Crimson> abi: slim?
<abi> i think slim is a setl relative at http://birch.eecs.lehigh.edu/slim/ or quite free and small
<Crimson> abi: rkt?
<abi> wish i knew, crimson
<s_r> why wouldn't storm make use of the kernel threads?
<Plundis> how come my script had automatically ignored most everyone in the world? something seems to be seriously wrong...
<Crimson> I haven't found out a way to take advantage of them yet... ;)
<Crimson> Any ideas?
08:30am
<s_r> well you could have a kernel thread for each process
<s_r> BeOS has kernel threads for processes
<s_r> that do system calls
<s_r> like the system call is called by the interrupt handler
<s_r> and then the system call is added to a queue
<s_r> for each kernel process thread
<s_r> and the thread then executes it
<s_r> or you could just have threads for all system calls
<Crimson> hmmm...ok... but that would be slower than the current system, i think...
<Crimson> i was more thinking about background swapping in and out of unnused memory...
<Plundis> s_r: I think Crimson is right.. our current system is probably better.
<Crimson> s_r: a kernel thread that checks for memory belonging to a sleeping process or something like that, and if found, throws it out on disk...
<Crimson> maybe?
<s_r> yeah
-:- eihrul [lee@usr5-ppp111.lvdi.net] has joined #Tunes
<Plundis> Crimson: yeah, that's a really good idea.
<Plundis> Crimson: the problem is that it would probably be too time consuming on regular IDE systems..
<eihrul> by chance, anyone here knowledgeable of x86 task switching?
<Plundis> eihrul: but of course. :)
<Plundis> what do you want to know?
<eihrul> okay... i've been trying to fud around with building a kernel
<eihrul> i've look at about 7 kernels (source code), and none of them use the TSS
<Plundis> eihrul: they all suck. :-)
<eihrul> is it truely faster to manually load/save the registers than use the TSS?
<Plundis> eihrul: check out www.chaosdev.org to find one that does.
<Plundis> eihrul: I don't think the difference is very big nowadays.. but it probably was once upon the time.
<eihrul> well, what about on a 486?
<Plundis> eihrul: dunno.. I'd recommend you do your own benchmarking or check if Intel has any.
<Crimson> eihrul: using TSSs is easier....
<eihrul> way easier
<eihrul> but if the speed benefit is big enough, i'd prolly go with manual
<Plundis> Crimson: I don't think it's very slow anyway.
<Crimson> i dont think so...
<Crimson> Plundis: is doing the syscall_thread right now. Have to extend som process_structures to keep track of threads... kinda simple, but timeconsuming...
<eihrul> actually, nah, i'll just go with TSS already then...
08:40am
<Plundis> Crimson: np.. but I really can't wait much longer. :)
<Crimson> Plundis: is there nothing you can do in the meantime?
<Plundis> Crimson: like what?
<Crimson> Plundis: some server?
<Crimson> Plundis: comport?
<Plundis> Crimson: does bochs support it?
<Crimson> Dont know...
<Plundis> wait, I'll check..
<Plundis> PS/2 mouse would be great otherwise.
<Crimson> Plundis: fix the keyboard-server to take arguments...
<Plundis> Crimson: it already does.
<Plundis> but I can't test that in bochs, it generates a GPF.. :-/
<Crimson> Plundis: why is that?
<Plundis> it works good on a real computer though...
<Crimson> take Gustavs then... ;)
<Plundis> Crimson: no idea. bochs seems to be seriously screwed, since it doesn't do other stuff as well. but why on earth are we writing this in #tunes? :)
<Crimson> eel and I will write a ps/2 server...
<Plundis> great.
<Plundis> bochs seems to support serial port.
<Crimson> Plundis: lets take internel chat elsewere...
<Crimson> ok? ;)
<Plundis> Crimson: you're not very active on the other server.. that's not my fault. :)
<eihrul> plundis: lots of useful stuff in your kernel, thanks :)
<Plundis> eihrul: hey, no problem. just remember that it's GPL:ed, so don't borrow stuff if you don't like the GPL.
<eihrul> my stuff is GPL too
<eihrul> so that's no problem
<eihrul> i'm just writing it to learn how, so it's not like licensing is a big issue
<Plundis> eihrul: okay, great. I just wanted you to know so you don't wet your pants. :)
<eihrul> no way, that's why i have a toilet.
<Plundis> =)
<eihrul> you hang out here often, cause i might need you again? :)
08:50am
<Plundis> eihrul: I will, from now on.
<eihrul> cool, thanks
<eihrul> been looking for like 4 days for someone who knew anything about x86...
<Crimson> eihrul: if plundis is not around, you can always ask me, cause i wrote chaos together with him...
<eihrul> aye
<Plundis> eihrul: well, you certainly found the right people. :)
<eihrul> well, after 4 days i had to find something :)
<Plundis> I guess so. :)
<s_r> eihrul how far are you in your kernel work?
<eihrul> not very far... i've been spending last two weeks wandering around the intel manuals
<eihrul> i have barely anything written yet, just a memory manager for now
<s_r> i too wonder what the performance difference between manual switching and TSS switching is
<s_r> BSD is a good example of kernel engineering. i think i'll check to see what it uses
<eihrul> BSD doesn't use it i believe
<eihrul> i've looked at mach, linux, minix, fiasco, fluke, and so on
<eihrul> not any of em seem to ues the tss
<s_r> fluke :)
<s_r> fluke is cool
<eihrul> yeah, i might use parts of the oskits once i get microkernel done
<eihrul> and hack together some device servers
<s_r> eihrul take a look at Chaos code
<eihrul> looking at it right now :)
<Plundis> s_r: it's "chaos", not "Chaos". :-)
09:00am
<eihrul> and "chaos code" and "Chaos code" are still missing appropriate articles of speech, i.e. "the", what's the point? :)
<Crimson> you could say chaos' code though...
<eihrul> so he could have really meant "look at that person, Chaos, code"
<eihrul> in which case, plundis is in err...
<Crimson> Plundis sucks... ;)
-:- AlonzoTG [Alonzo@209-122-199-29.s283.tnt4.lnh.md.dialup.rcn.com] has joined #tunes
09:10am
* Plundis/#tunes is never in err.. :)
<eihrul> okay, then s_r committed two errors of speech...
<eihrul> s_r: you can going to take that from plundis, man?
<s_r> haha
<s_r> Crimson: do you know everything about x86 architecture?
<Crimson> almost... ;)
<Crimson> no, dont EVERYTHING, but a lot...
* s_r/#tunes is looking at Alpha Architecture Handbook
<Crimson> noy...
<s_r> i'm thinking of developing my OS strictly for Alpha :)
<Crimson> argh.... mistyping sucks!
<eihrul> er, -can
<eihrul> irony sucks
<s_r> eihrul: eh?
<eihrul> there was a grammatical error in my statement :)
<s_r> ehh
<s_r> heh
<s_r> eihrul: is your OS going to be revolutionary?
09:20am
<eihrul> no, just small
<eihrul> won't be using flat addressing either
<eihrul> that's a counterrevolution :)
<s_r> you're going to use segments?
<s_r> yikes
<eihrul> well... i'm doing it in such a way as to make it mostly transparent from C
<eihrul> the data and stack segments reside in a single segment (same address space)
<eihrul> the text segment is it's own dedicated segment however
<eihrul> so you can't do runtime code generation in C, but oh well :)
<Plundis> s_r: we were thinking of doing like that in chaos earlier too..
<eihrul> i could put the text segment in the one segment too
<Plundis> s_r: actually, we already are, since we're using virtual adressing. :)
<eihrul> but the text segment should be non-writeable
<eihrul> and i don't want bad stack coding to suddenly mess up text segment
<Crimson> is going to test the ps2 mouse server now, hold on to your sofas pals...
<eihrul> and paging looked too ugly to me
<eihrul> too much work for a microkernel since i'm not implementing swappage to disk
<s_r> paging has high granularity
<s_r> err whatever
<eihrul> so does segmentation
<Plundis> eihrul: paging is very nice..
<Plundis> eihrul: we haven't got swapping either, but paging is just so neat.
<s_r> Plundis: you are using multiple segments?
<Plundis> and besides, ELF required paging.. and ELF is great.
<s_r> paging is good except for the high granularity
<eihrul> it does?
<s_r> plundis: XCOM
<eihrul> how does it require paging?
<s_r> XCOM is a new object exec file format
<eihrul> paging just seemed more compact mechanism to me
<eihrul> er segmentation :)
<Plundis> s_r: not really.. but they're not flat, since we're using paging.
<eihrul> slip of tongue
<Plundis> s_r: where can I find docs? is it good?
<s_r> xcom?
<abi> it has been said that xcom is a portable component system at http://xcom.tunes.org
<eihrul> and it should be conceivably faster since you have no address translation
<s_r> there still is a lot of work to be done on it
<s_r> but it has promise
<Plundis> eihrul: yeah. or, actually, Linux ELF:s are built to load at 128 megs..
<Plundis> and that can't work without paging or more than 128 physical megs.
<eihrul> ack... i might have to hack egcs then, what fun :)
<s_r> haha
<eihrul> but i'd have to port it anyhow to get a self-developing platform
<s_r> eihrul no, hack ld
<eihrul> well, i'll worry about that when i finish microk...
<eihrul> i don't have anything complete to compile yet :)
<AlonzoTG> linux sucks
<Plundis> eihrul: it's not very difficult, but we needed paging anyway.. for memory protection.
<Plundis> AlonzoTG: yeah, but chaos isn't ready yet.
<eihrul> segmentaiton does do memory protection though
<Plundis> eihrul: we needed 4 gigabyte segments.
09:30am
<Plundis> and it's hard to do memory protection with 4g segments without paging. :)
<Crimson> A flat address space for each and every process rules...
<eihrul> how is it hard?
<eihrul> but, a process really only needs to access memory within it's address space, anything beyond that is strange
<Plundis> eihrul: think about and you'll figure it out. :)
<eihrul> i don't see why it's hard
<Plundis> eihrul: in chaos, each process has always access to 4 gbyte memory.
<eihrul> each segment can have a maximum of 4GB of memory
<eihrul> and since each offset address within the segment is relative to 0
<eihrul> you can move the segment around in memory to allocate more space
<Plundis> yeah, but all segments should be zero-based to allow access to 4 gigs.
<eihrul> no
<eihrul> only the ones that need 4 gigs of memory should be in a place where it's possible to have that much
<eihrul> and it's very easy to relocate segments :)
<eihrul> but if i do start running into severe problems, there's always paging...
<Plundis> paging is Good, IMO. and I don't think it gives much less performance.
09:40am
<eihrul> well, it's always an option
-:- SignOff AlonzoTG: #TUNES (Have Nice Day :))
<s_r> eihrul you'll have to start coding the GUI soon :P
<eihrul> ack, no way
<eihrul> it's going to have a console and that's it...
09:50am
<s_r> i wonder what the chaos gui is going to be like
<eihrul> actually... would be way easier to make to make an x-server like thing than a multiplexed console
<eihrul> especially in a microkernel model
<Plundis> s_r: it's gonna be like a mix of GNUstep, Amiga Workbench and Windows. (we've removed the bad parts of all those. :)
<Crimson> that AlonzoTG guy is one talkative guy, isn't he? I think the only thing he said was "Linux sucks"... =)
-:- Oddity [d96-abo@spel11.nada.kth.se] has joined #tunes
<s_r> cool
-:- smkl [sami@MCCXCVI.rdyn.saunalahti.fi] has joined #tunes
<eihrul> sheesh, whatever happened to KISS these days?
<Plundis> s_r: yeah, it's really gonna kick ass.
<eihrul> have people forgotten the virtues of textual consoles? :)
<Plundis> eihrul: yes? :-)
<eihrul> gah
<eihrul> first paging, now GUIs... 
<Oddity> Re: thoughts about Tunes OTOP & JavaVM:  So, what's wrong with SmallTalk?
<Plundis> eihrul: =)
<eihrul> what's next, a builtin java vm?
<s_r> Plundis: you should distribute precompiled Chaos distributions too
10:00am
-:- SignOff eihrul: #TUNES (Leaving)
<Crimson> Plundis: I think it compiles without errors now... ;)
<s_r> the ps2 server?
<Plundis> s_r: yeah, I know.
<Plundis> s_r: a demodisk, perhaps? but we've got to have something to put on it..
<Plundis> Crimson: does the ps2 server work?
<s_r> go_background();
<s_r>    s=socket(AF_INET, SOCK_RAW, 1);
<s_r>    openlog("icmplog", 0, LOG_DAEMON);
10:10am
<s_r> new Chaos code =]
<Plundis> gha. I really need a working PC emulator..
<Plundis> s_r: we already have network support.. :)
<Plundis> s_r: the 3c509 server has a builtin IP flooder. ;)
<s_r> do you have the socket() system call?
<s_r> a built in ip flooder?????
<Plundis> s_r: it's just for testing..
<s_r> hmm i'd add ip firewalling
<Plundis> s_r: we wanted to see how much data we could broadcast from our test machine (a 386SX-16), so we wrote a floodubg origran,
<s_r> in between the OS and the raw 3c509 output
<Plundis> flooding program.
<s_r> it depends on the speed of the 3c509, and the speed of the 386 bus
<s_r> how fast was it?
<Plundis> s_r: but no, we haven't got real networking yet.
<Plundis> s_r: about 4 megabit was what we achieved.
<s_r> cool
<s_r> shows off the speed of Chaos :)
* s_r/#tunes wonders if Chaos will be faster than Linux
<s_r> probably
<Plundis> we hope so.
<Plundis> it currently boots in less than a second, including mode switching to 320x200. :)
<s_r> hahah that's great
<Plundis> (or whatever, the 640x400 doesn't seem to work very well yet)
<s_r> you'll have to include support for building custom kernels
<Crimson> s_r: why?
<Plundis> s_r: that won't be necessary, really.
<Plundis> everything customizable is implemented at userlevel.
<s_r> but what if some people have no need for, say, the ps/2 mouse server?
<s_r> oh DUH
* s_r/#tunes feels stupid
<Crimson> hehe...
<s_r> i was thinking of monolithic kernels
<Plundis> :)
<Crimson> that is an advantage of microkernels...
<Plundis> s_r: I know how different it is to get used to it, believe me..
<Plundis> in the begining, chaos wasn't going to be microkernel-based. but the Hurd brought us on the right track.
<s_r> hurd rules
<Crimson> i have never hurd of it... ;)
<s_r> heh
10:20am
* Tril/#TUNES is back from the dead. Gone 66 hrs 22 min 54 secs
<Plundis> Hurd is really great, yes.
<Crimson> Plundis: my screen seems buggy...
<Plundis> too bad it's so monolithic.
<Plundis> Crimson: your screen or your monitor? :)
<Crimson> My screen, is the chaosdev irc server running?
<Plundis> yeah..
<s_r> i'm going to install VSTa and Hurd on my 486
<Plundis> Crimson: you've been logged of for some reason..
<Plundis> s_r: ouch. Hurd is not very fast at the moment..
<Plundis> they're basing it too much stoneage technology.. :/
<Plundis> "on", even
<s_r> like what?
<s_r> i haven't run Hurd yet though i shouldn't be judging it :)
<Plundis> s_r: Unix. :-)
<Plundis> IMO it's a pretty bad idea to base an operating system on Unix today..
<s_r> unix is getting old
<s_r> i wouldn't base any new OSes on unix
<s_r> most of the currently fully implemented OSes are based on unix, sadly
<Plundis> yeah, I know.. :-(
<Plundis> argh. someone is trying to setup a warez server on our FTP server. :-/
10:30am
<Crimson> Plundis: det e non snubbe fron vesteros... ;)
<Plundis> Crimson: hur vet du det?
<Crimson> Plundis: knale hade snackat med honom...
<Plundis> Crimson: ok.. sade han åt honom att tanka upp eller?
<Plundis> (sorry guys for the swedish..)
-:- SignOff Crimson: #TUNES (Confucius say: Man with hand in pocket is having a ball.)
-:- Plundis2 [plundis@t3o73p29.telia.com] has joined #tunes
<Plundis2> whoha, X-Chat is really good. :)
-:- SignOff Plundis: #TUNES (Confucius say: Man who Farts in Church sits in own pew.)
-:- Crimson [crimson@chaosdev.org] has joined #tunes
<Plundis2> Crimson: what was that for?
-:- Plundis2 is now known as Plundis
10:40am
<s_r> if i had 0xdcba in memory
<s_r> and i mov'd it to register ax
<s_r> what would be in ah? what would be in al?
<s_r> would it be ah=0xba and al=0xdc?
<Plundis> s_r: I'm not really sure, but I think so. (I always make that mistake myself...)
<s_r> i know the x86 architecture stores it in reverse order in memory
<s_r> i just don't know how it's stored in the register
<s_r> guess i'll have to look at the mans
<s_r> manuals
<Plundis> what's wrong with trial-and-error? :)
10:50am
<s_r> can i compile Chaos under BSD?
<Plundis> yes.
<s_r> using BSD ELF right?
<Plundis> with a GNU make and GNU binutils, at least.
<Plundis> s_r: yes.
<s_r> ok cool
* Plundis/#tunes has been compiling it under FreeBSD 3.2 at least.
<s_r> i'm not going to sell my soul to linux just yet
<Plundis> but our main development platform is Debian GNU/Linux.
<Plundis> s_r: :)
<Plundis> Linux is a pretty poor alternative, but BSD has a to immature package management for my taste..
11:00am
-:- hcf [nef@me-portland-us732.javanet.com] has joined #tunes
<hcf> i recommended he talk to u about it
<Tril> i'm emailing beholder about zope
<hcf> did he mention doing the server-side prefs and style templates?
<Tril> you mean bineng's python scripts?
<Tril> or script?
<hcf> no
<hcf> layout style like we talked about a while
<hcf> ago during the meeting afaik
<hcf> s/style/styles/
11:10am
<hcf> did i say something wrong?
<Tril> you mean PHP3?
<hcf> i thot zope could do it, but php3 is fine too i /guess/
<Tril> of course zope can do it.
<hcf> whatever it takes to do it
<Tril> I think we should use zope
<hcf> thats what i was saying
<Tril> sorry, i should have just said "no, Beholder didnt say anything specifically about page layout"
<hcf> will /everything/ be in the postgresql db?
<hcf> r we staying w/ the daily-site-"make"?
<s_r> Tril: what did Fare have to say about this?
<Tril> the data for reviews, URLs, subprojects, members, will be in   db
<Tril> Fare will accept any solution if it has a way to text-back it up
<hcf> Tril: whats that leave as not being in the db?
<Tril> sql db can be dumped into ASCII SQL commands that can be made back into databases, or CVS'ed or whatever
11:20am
<Tril> not in db: DTML markup for the pages including the ones people see and the ones that members use to edit the db
<Tril> but that can be exported to text as well
<hcf> that it?
<abi> well, that it is an awesome server platform
<hcf> abi: forget that it
<abi> hcf: I forgot that it
<Tril> well , mailing list messages and newsgroup posts (if I set that up) won't go in a db...
<hcf> right
<Tril> they SHOULD, but it's really not worth it until TUNES
<s_r> Tril: do you speak French?
<Tril> one of Fare's other requirements is that the format be easily migratable to TUNES. Well, since my persistent store is a generalization of relational db's that should be easy.
<Tril> s_r: No French, but I know a little Japanese, Russian, and Spanish
<hcf> s_r: for the time being, plz stay on topic
<hcf> Tril: how goes the p-sql work?
<Tril> no work, i am working on my spec.sgml
<Tril> i put off the zope and db stuff
<hcf> u seen the sitebugs?
<Tril> never heard of it
<hcf> i found a few sitebugs
<hcf> told fare, he said tell beh
<hcf> i did, he said he wasnt able to do anything about em
<hcf> he said he'd mail u
* Crimson/#tunes is feeling tired...
11:30am
<Tril> i didn't get anything about "site bugs", can you tell me what it is?
<Tril> like bad links?
<hcf> the arrow paper isnt mentioned on the sitemap's papers section
<hcf> "Currently maintained by Ultima" on the subprojects front page should be updated
<hcf> terms like 'glossary' when appearing in the todo lists should be hyperlink'd
<s_r> Crimson: nap
<s_r> =)
<Crimson> Can u give me some urls for interresting os-projects to check out?
<hcf> Crimson: look at the tunes os review area
<Tril> that kind of stuff is useless to fix now in cvs if we are migrating to zope.  The sitemap would be auto-generated, maintainers would be read out of the db.  Th"glossary" link would be a DTML thing then.
<Tril> abi review?
<abi> tril: wish i knew
<Tril> abi review is http://www.tunes.org/Review/
<hcf> Tril: other things have been fixed/added
<Crimson> ok...
<Tril> hcf, have been? what?
<hcf> Tril: parts of review/glos
<Tril> someone is fixing stuff outside of CVS? why?
<hcf> didnt water do some work on the glos?
<Tril> yes. He made some txt files but then when work on zope started he stopped sending any until zope was up
<hcf> oh
<Tril> i mean review. i dont know if he did any glossary
<hcf> see
<Crimson> How do I get chaos added to Reviev/OSed.html?
<Crimson> OSes.html?
<Tril> crimson , post to review@tunes.org mailing list
<Tril> it will get added eventually
<Crimson> Do I get flooded with mails then?
<hcf> "until zope is up" how long must we wait?
<Tril> Crimson: no, probably not
-:- eihrul [lee@usr5-ppp111.lvdi.net] has joined #tunes
<Crimson> How subscibe?
<Tril> hcf: beholder seems eager to help
<hcf> Tril: not w/ p-sql+zope
<Tril> Crimson: you don't need to subscribe to post, but you can subscribe at http://lists.tunes.org/cgi-bin/list-control
<Crimson> ok... going to eat, but will be back soon...
<Tril> hcf: yes, he did in his e-mail, he said he wanted to learn zope and he would promise to get sometrhing up for Review if "only I got zope running"
<Tril> so i'm going to make him access to the db and see if he can make sense of my db layout
<Tril> (meanwhile him reading the zope docs)
11:40am
<hcf> he has to handle ur db layout?
<hcf> what a pain
<Tril> well, he should look at mine and comment on it, instead of just designing his own from scratch!
<Tril> and if he doesn't know anything about them, he may want to just use mine anyway.
<s_r> does anyone here know where i can buy individual Alpha CPUs?
<Tril> no
<Tril> not me, sorry :(
-:- s_r [sr@phila-dialup379.nni.com] has left #tunes []
11:50am
<Tril> gotta go eat then work on spec :)
<Tril> bye hcf
* Tril/#TUNES is away: (afk) [BX-MsgLog Off]
-:- SignOff hcf: #TUNES (Leaving)
-:- SignOff eihrul: #TUNES (Leaving)
12:00pm
-:- ElGato [Bob@20-227.nctimes.net] has joined #tunes
-:- ElGato [Bob@20-227.nctimes.net] has left #tunes []
-:- ElGato [Bob@20-227.nctimes.net] has joined #tunes
-:- SignOff ElGato: #TUNES (:D)
-:- core [core@core.suntech.fr] has joined #tunes
<core> people hi
<Crimson> I have some trouble with readonly pages on intel, anyone who have the experience?
<core> crimson: what's the problem?
<abi> i guess the problem is that it would probably be too time consuming on regular IDE systems..
<core> abi, forget the problem
<abi> core: I forgot problem
<core> thanks.
<abi> de nada core
<core> hehe
<Crimson> cool...
<core> crimson: can i help you? 
<Crimson> The problem: i set the readonly flag, but the pages are still writable.
<core> crimson: there's no readonly flag
<core> crimson: there's a write flag.
<Crimson> same shit...
<core> well, if you SET it, the pages will be writable
<core> :)
<core> if you clear it, they won't be; don't forget the x86 takes the most restrictive set of flags, between the PTE and PDE, also.
<Crimson> whatever i set it to, i can still write to the page...
<core> oh, you're running at ring 0.
<core> set WP in CR0
<Crimson> ohhh...
<core> by default you can write to readonly pages if you run as ring 0, for compatibility w/ i386
<Crimson> write protect?
<core> 486+ have the WP bit, that checks writability in ring 0 as well
<Crimson> i run in pl2
<core> crimson: hmm, maybe it's pl0-2 then i forgot.
<core> crimson: make sure you _clear_ the R/W flag tho.
<Crimson> ok...
<core> crimson: try to turn WP on, and it should generate an exception
<core> (when you write to the page after that, that is)
<core> x86 is stupid :)
<Crimson> yeah, i got the point. have to check...
<core> okay :) hope it helps.
12:40pm
* core/#tunes wake()s everyone up
<Crimson> core: read only pages work on all intels above 386 i hope?
<Crimson> including 386...
<core> crimson: no, 386 will not trigger exceptions when you write to read-only pages in supervisor mode
<core> crimson: but 486+ has that capability i believe.
<Crimson> but the pages for the programs dont have the supervisor flag set...
<core> i wish intel got the paging right in 386, and didn't wait i486 to add invlpg and wp
<core> crimson: oh, no, those will trap.
<core> crimson: regardless of WP
<Crimson> yes, and see, THAT is the problem...
<core> crimson: if you're using bochs, it emulates a plain 386, so you have to try what you're doing on a _real_ box
<core> crimson: but if you access user pages while running at ring 0-2, it won't trap, see above; only at ring 3.
<Plundis> core: okay. that is the problem.
<Plundis> the 386 simply sucks.
<Crimson> yes, i know...
<Plundis> we'll have to depend on 486 for memory protection then.. too bad. :-/
<core> plundis: yeah, i wish we had WP and invlpg (above all) on 386. (and G bits in pages while we're at it :)
<Plundis> Crimson: what do you say?
<Crimson> Plundis: no way, well use segmentation
<core> plundis: running code at pl 2.. servers or something?
<Plundis> Crimson: get a grip!
<Plundis> Crimson: it'll be much harder to debug, and everything will be more cumbersome..
<Crimson> Plundis: we HAVE to use segmentation to protect the kernel
<core> plundis: [i assume you're the chaosdev people, i skimmed over the irc logs] why don't you run them at PL 3?
<Crimson> the servers is at pl2
<Plundis> Crimson: but what is the problem? we won't test the kernel on 386 anyway.
<Plundis> core: servers must be at pl2.
<core> crimson: okay, but why not just pl3? you can do everything with just two levels
<core> plundis: why?
<Plundis> core: servers can do some more than regular programs.
<core> plundis: NT doesn't run on i386 either, but still, there are lots of 386 boxes around to test with
<core> plundis: i know, but what can't they do at pl3 that they can do at pl2?
<core> plundis: even the IOPL can be set on a task basis.
<Plundis> core: the pl2 processes can register interrupts and I/O ranges.
-:- hcf [nef@me-portland-us732.javanet.com] has joined #tunes
<Plundis> core: oh no, the IOPL is always 0. :)
<Plundis> I/O maps rule.
<core> plundis: okay, just checking :) (for IOPL that is)
<core> plundis: you don't have to rely on PL's to grant access to some code..
<Plundis> core: how would we otherwise do it?
<core> plundis: you could always run servers at PL 3 and map some extra system calls to them or something.
<Crimson> core: no, but it is easier, as the cpu dpes the work..
<Plundis> core: that would be more difficult than we would gain...
<core> crimson: but then it will not be safe on i386 for something that can be done otherwise..
<Plundis> core: the 386+ has four privilege levels. why not use as many as possible?
<core> plundis: well, you'd get running chaos on i386 and it'll be as fast.
<Plundis> core: yeah..
<Plundis> but it'll run on 386 anyway.
<core> plundis: the 386+ has segmentation, why don't you use it? :)
<Plundis> or can't you set the WP bit on a 386?
<core> plundis: it will run, but it won't be safe. servers will be able to thrash pages
<Plundis> core: segmentation really sucks.. paging rules.
<core> plundis: there is no WP bit on 386
<Plundis> core: that's okay with me. :)
<core> plundis: my point. 4 privilege levels suck :)
<Crimson> core: we can make it secure using simple segmentation...
12:50pm
<core> crimson: then it's getting hairy :)
<Plundis> core: it'll just be difficult to debug on 386:s, but that's all.
<core> i'm just giving my opinion, eh :)
<Plundis> i mean, how many real people will run it on 386:s??
* Plundis/#tunes is smelling a fork.. ;)
<core> plundis: i don't know, i don't see what's wrong with making a longer syscalls table for servers :) it'll still be enforced by H/W
<Crimson> Plundis: we must run reliable ona 386
<Plundis> core: we have a different syscall system.
<Plundis> core: all syscalls have their own interrupt.
<Plundis> Crimson: why?
<Crimson> Plundis: just because...
<Plundis> Crimson: that's no reason.
<Plundis> let's be realistic instead.
<Plundis> i mean, 100% of the users *will* use 486 or better, and those who doesn't will still have a working (although not as reliable) system. what is the problem?
<Crimson> Plundis: it is really NOT cool to require a 486!
<core> plundis: maybe a few more than you think, that won't try chaos on their latest pIII at first :) (like me, i have lots of those around)
<core> plundis: the point is that there are easy solutions to make it run on i386.. but it's up to you :) i'm just giving you an opinion :)
<core> plundis, crimson: do what you think is best :)
<core> i must remember to add chaos to the OS review too :P
<Plundis> Crimson: why not?
<Plundis> core: yeah, you do that.
<Plundis> Crimson: NT does it..
<Crimson> Plundis: I dont see the problem.
<core> plundis: i'll update the OS review in a few days
<Crimson> core: cool
<Plundis> Crimson: segmentation is very ugly and difficult to debug
<Crimson> core: what will the description be?
<Plundis> Crimson: remember that objdump -d won't work..
<core> crimson: i haven't written it yet, but so far, it looks like a nice uK system (another one :-)
<Crimson> Plundis: 0-64 kernel, the rest is free
<core> crimson: i downloaded your code this morning, skimming the irc logs of this channel :)
<Plundis> Crimson: you misunderstood. the adresses written by the exception handlers won't be correct.
<Plundis> and that is not very nice, IMO.
<core> do what you think is best, guys, i just think that not working on 386 for something you can get around in many ways, is quite unusual :)
<Plundis> of course we can, but I doubt how useful that would be..
<core> well, you can always %ifdef for 386 if you don't want to keep the kludges running on 486+, but i don't know, i don't see why you want to be complicated and use 3 PL's :) especially if you want to port it to other cpus.. all others use two
<Plundis> I know..
<Plundis> it just was an easy way to implement different syscall privileges.
01:00pm
<core> i understand :)
<core> you can always software-check that the calling task of a given syscall is a server, too :)
<Plundis> I think we'll go for the pl3-way anyway.. coz segmentation sucks bigtime.
<Plundis> core: we don't store that information anywhere right now. :)
<core> it'll take two extra ops or something
<Plundis> all checking is done by the pl-thing..
<core> plundis: you don't maintain a task state of any kind?
<Plundis> core: of course we do, but that information isn't stored because it wasn't needed until now..
<Plundis> but I think that's the solution to use.
<core> plundis: i understand.. well the task state is a good place to store it at, then you just need to test if the caller is a server, and dump a fault if not ;)
<Plundis> core: do you want to be mentioned in the THANKS file? :)
<core> plundis: well, i don't mean to influence your design. just talking :) i'd think it's also the solution to use :)
<core> plundis: hehe, not that i would mind :) sure, thanks :)
-:- AlonzoTG [Alonzo@209-122-238-235.s235.tnt2.lnh.md.dialup.rcn.com] has joined #tunes
<Plundis> you've got it.
<core> plundis: life will be much simpler for you guys once you settled all those things and can safely write servers :)
<core> plundis: thanks :)
<Plundis> yeah, of course. the kernel is just a neccessary evil.
<core> well, you need some supervisor code anyway. not necessarily a single block, but you need arbitration
<Plundis> yeah, of course.
<core> (unless you follow Fare's path of using compilers that produce safe code, but that means no C or asm or that kind of thing :)
* Plundis/#tunes is not a kernel hater really, but they're ugly to write..
<core> they can be made clean, but indeed when I have some C++ people (ie, anal-retentive object freaks :-) looking at my code, they sometime have an heart attack.
<Plundis> =)
* Plundis/#tunes <- object freak too, but open minded
<core> turning WP on is not too good for debugging anyway. you might want to write a runtime debugger than can change contents of even readonly pages
<Plundis> oh.. yes.
<core> (ie, a gdb serial debugger representative or something :)
<Plundis> would be pretty cool, yeah.
<core> it's easy to implement too. gdb has stubs you can fit in mostly any system :)
<Plundis> cool..
<core> debugging at source level while another box is running your OS across a serial cable is quite impressive :-)
<core> although i'd guess you're probably printf-debuggers ;)
<Plundis> yes.. =)
<core> hehe. same here.
<Plundis> don't really know why, but that's the best way. at least when you've got the source..
<core> true.. and i'm used to do that, so when i get to higher levels i still don't switch to gdb or similar :P 
01:10pm
<Plundis> no, me neither.
* Crimson/#tunes has no socks, but is using socks... ;)
<Plundis> socks sucks.
<core> have you considered using GGI for graphics drivers instead of writing two thousand of them? ;)
* Plundis/#tunes has socks, but not socks.
<Plundis> core: how good is GGI? is it GPL:ed?
<core> crimson: it's +38 C here, i wouldn't want the physical socks today ;)
<Plundis> core: are you living in Africa or something? :)
<core> plundis: it's a MIT license now, which means you can do anything with it. it has drivers for most boards around, even modern ones likes RIVas
<core> plundis: lol, no, south of france
<core> it's actually +30 outside but i have 5 boxes in this apartment :-)
<Tril> core is GGI working for Linux and Clementine? I thought it was early in development
<core> plundis: drivers can get on the largish side (~ 60 K) but you can easily wrap them as a PL3 server
<Plundis> core: seriously?? if so, GGI sounds really good. I'd better check it out.
<Plundis> core: yeah, that would be great!
<core> tril: yes, GGI works on Clementine, i use it for the display part. works great :)
<core> plundis: well, check it out yeah, i hope it'll be useful to you :) (www.ggi-project.org)
<core> tril: GGI has been in development for years; the libggi part is not maybe full featured, but the kernel part works great (even if linus doesn't like it :)
<core> tril: hi btw :)
<Tril> hi
<abi> what's up, Tril
<core> plundis: just not to reinvent wheels where you can't bring added value at least at first
<core> tril: i'm going to do a massive commit to clementine probably tomorrow (well, for Fare anyway :), i'll send you a new tarball
<Plundis> mmm..
<core> plundis: do you have a CVS repository ?
<Tril> any mouse support yet? :) I'm a mouse person
<Plundis> core: no, not yet.
<Plundis> all the developers live pretty much nearby each other..
<core> tril: on clem? no, but it's coming :)
<Plundis> well, all live in the same country at least. :)
<core> tril: it's not hard, i wanted bases to be settled before people wrote any more code
<core> plundis: have room on your server? otherwise i'm sure tril won't mind adding you to the tunes repositories.. if he does, i could as well :)
<Plundis> core: thanks, but we've got 17 gigs in our server.. :-)
<core> plundis: okay, just offering. set it up then :-)
<Plundis> so we've got more than enough.. :)
<Plundis> but it's just a matter of laziness.
<core> i have cron jobs to cvs update a dozen repositories at night so i can troll them in the morning ;)
01:20pm
<core> plundis: well, whenever you get time :)
<core> tril: OSS is for this fall; we're hiring steffen seeger, andrew apted and wouter scholten, on top of andreas beck
<Tril> core: I don't understand
<core> tril: clementine
<core> tril: opensourcing -> this fall
<core> tril: suntech -> hiring steffen seeger, wouter scholten, andrew apted to help out
<core> tril: me -> happy
<core> :)
<Tril> They all live in France? Or is suntech a multinational
<Tril> does suntech hire GGI developers or clem or ...
<core> tril: it's not, and they aren't, but they'll telecommute. andy has been doing so for a year
<Tril> well, that's neat then.
<core> tril: well, they're competent.. andy's been doing an awesome job
<core> tril: yeah, indeed :) i hope you'll like the upgrades, it's been a while since you toyed with it.
<core> tril: i'm getting a little less negative at my own code, but just a little less ;)
<Tril> does it still need a floppy?
<core> tril: yeah and/or bochs
<core> tril: but it does a lot more and works everywhere.
<core> tril: actual persistence will be worked on soon tho
<core> tril: have you progressed on your work? :)
<Tril> yes
<Tril> i'm working on a better spec now :)
<core> tril: great! :) nice
<Tril> btw: irc.ggi-project.net #ggi seems empty. considered moving it to OPN?
<core> tril: ha, ha
<core> tril: LILO told me 5 times he was okay to link me, then i never heard from him again
<Tril> oh, you were adding a new server. hmm. I meant just the chan
<core> oh
<core> i suppose i could too
<core> i wouldn't mind running an OPN server tho
<Tril> lilo has a habit of dropping in on new channels, ex: he's in #tpp now. If he does that to yours, you can bug him about connecting up
<core> i did bug him :) and he did say okay cool like 5 times :)
<core> he just gets sidetracked i suppose
<Tril> speaking of sidetracked.. I never get work done on irc
<Tril> see you later! :)
<core> lol.. same
* Tril/#TUNES is away: (afk) [BX-MsgLog Off]
<core> that's why i don't drop here very often
<core> later tril
01:30pm
<Crimson> core: moving servers to pl3... ;)
<Crimson> core: are you a tunes developer?
<Plundis> Crimson: but why doesn't I/O work anymore?
<core> crimson: i maintain the OS review (took it over from Fare) and i'm part of the mass of tunes people although i'm not directly developing Tunes itself
<Crimson> Plundis: iopl?
<abi> it has been said that iopl is always 0. :)
<core> abi: forget iopl
<abi> core: I forgot iopl
<core> stupid bawt
<Crimson> core: what is abi?
<abi> i am the little girl borg with a big brain or #tunes' personal little whore or gaklosmontic or flurivostuginuous or xyvarestoplik or an Artificial Idiot or braindead or a crack addict infoslut or stubborn
<core> crimson: an infobot
<Crimson> abi: fuck you!
<core> crimson: it learns expression such as "object is something"
<Crimson> no response there...
<Crimson> ;)
<core> crimson: sometimes it learns unexpected things ;)
<Crimson> core: yes, i noticed that.
<Crimson> kinda annoying sometimes... ;)
<core> crimson: no kidding ;)
<AlonzoTG> linux?
<abi> i heard linux was elementary penguin singing hare krishna or a junkie's delight
<AlonzoTG> om
<Crimson> quake?
<abi> quake is not
<Crimson> hehe...
<Plundis> abi: forget linux
<abi> Plundis: I forgot linux
<core> ack
<AlonzoTG> =\
<Plundis> abi: linux is a bloated monolithic piece of junk.
<core> ack
<core> :)
<AlonzoTG> The previous linux was better =P 
<core> alonzo: what was it?
<Crimson> abi: plundis is an idiot
<Plundis> perhaps.. =)
<core> plundis?
<abi> plundis is, like, an idiot
<core> heh
<Plundis> Crimson is a pice of junk. =)
<Crimson> haha
<AlonzoTG> (mainly because I put it there?
<Plundis> Crimson: screw you guys, I'm going home now..
<AlonzoTG> me?
<Crimson> no!
<abi> hmmm... AlonzoTG is annoying with his greetings messages or probably a bot
<Crimson> me?
<abi> you are currently working on the ide-driver for my os.
<Plundis> me?
<abi> you are an idiot
<core> me?
<Crimson> cool!
<abi> you are developing a kernel-less, component-based OS, Clementine, and the XCOM file format (http://xcom.tunes.org)
<Plundis> =DD
<Crimson> heheh!!!
<core> good girl.
<Plundis> abi, I am the ruler of the known world.
<abi> ...but plundis is an idiot...
<Plundis> abi, forget Plundis.
<abi> Plundis: I forgot plundis
<Plundis> abi, I am the ruler of the known world. I control this world at the top of my fingertips.
<Plundis> abi, Plundis?
<abi> you are the ruler of the known world. I control this world at the top of my fingertips.
<Crimson> core: abi rules!
<AlonzoTG> hehe
<Plundis> not really correct.. but what the heck.. =)
<Crimson> core: ;)
<Plundis> pretty smart for a bot..
<core> crimson: yeah, hcf maintains her i believe :)
<AlonzoTG> Core: WussupL
<Crimson> core: is he a tunes developer?
<core> hey Alonzo, not much :) working, working and working. you? did you get your project started?
01:40pm
* AlonzoTG/#tunes is spending a fortune keeping hiz kar on da road.
<core> crimson: hcf is also part of the Tunes mass :-)
<abi> okay, core.
<core> ah fuck 
<core> hcf?
<abi> i heard hcf was gone! :( or part of the Tunes mass :-)
<AlonzoTG> =\
<core> stupid thing anyway :)
<core> crimson: i view Tunes more like a communauty of people who work around OSes or languages and exchange ideas, more than an actual physical project, but it'll tend towards one eventually i guess :)
* Crimson/#tunes is cool
<AlonzoTG> Core: I just gots some moneys out of the trust fund my parents gave me and I intend to spend it on my OS. =)
<Crimson> me?
<abi> it has been said that Crimson is currently working on the ide-driver for my os.
<core> crimson is also very strange
<abi> okay, core.
<core> crimson?
<abi> crimson is currently working on the ide-driver for Crimson's os. or very strange
<Crimson> core: it is great.
<core> :-)
<AlonzoTG> IRON BUTTERFLY
<Crimson> core: were can I get a bot like abi?
<core> alonzo: still want to make a company around it or something?
<core> crimson: it's called infobot
<AlonzoTG> yes, If I can. =\
<core> i don't have the URL in mind, but i can probably find it..
<core> crimson: it's useful at answering FAQs and stuff :)
<core> http://www.cs.cmu.edu/~infobot/infobot.html 
<Crimson> core: a friend got impressed and wants one... ;)
<core> infobot?
<abi> infobot is at http://www.cs.cmu.edu/~infobot
<core> DOH!
<core> i should have known :)
<Plundis> Crimson: I'll fix it.
<core> crimson: well he can get it there ;)
<Crimson> core: heh... smart
<core> plundis: do you make nightly snapshots or just release code when you like? :)
<Plundis> core: we just release code when it works. =)
<core> plundis: ahh.. just wondering :)
<Plundis> ...which can be often or very seldom, depending on the situation.
<core> yeah, sometimes you break a big bunch of stuff :)
<core> but you still have that luxury, since noone wrote applications for the system yet :)
<Plundis> yes, that's very nice sometimes.
<core> i'm sure. i don't believe you can get a design right the first time
<core> (but soem idealists will tell you you can :)
<core> s/soem/some/
<Plundis> but they are wrong. :)
<core> i assume so :)
<Crimson> me?
<abi> you are currently working on the ide-driver for my os. or very strange
<Crimson> abi: uggla is warez
<Crimson> uggla?
<abi> somebody said uggla was warez
<core> something tells me crimson ported the 3c509 driver :
<core> :P
<Crimson> core: plundis ported it from Linux.
<core> crimson: okay, then you TWO like that word ;)
<core> crimson: i skimmed over this piece of code this morning and noticed this word ;)
01:50pm
<Plundis> core: we're very much influenced by each other.. =)
<Plundis> core: uggla is the swedish word for "owl".
<core> plundis: that, or juno reactor ;)
<core> plundis: ahh.. i see :-)
<core> hmm, your floppy image makes bochs crash :)
<Plundis> cool.. =)
<Plundis> it should work..
-lilo(lilo@varley.openprojects.net)- [GlobalNotice] Hi all.  Quick server rehubbing.  Please bear with us.
<Plundis> did you grab the one from the webpage?
<core> it loads the kernel, then exits
<core> plundis: yep
<Plundis> i think that's a feature.. ;)
!sterling.openprojects.net!! Remote CONNECT irc.linux.com 8005 from ^lilo
<core> well, i have no time to read anything :-)
<Plundis> oh..
!sterling.openprojects.net!! Remote CONNECT irc.linux.com 8005 from ^lilo
<Plundis> you should at least get the GRUB prompt?
<Crimson> ahhh!!!!!!
<core> oh, i have that
!king.openprojects.net!! Remote CONNECT irc.linux.com 8005 from ^lilo
<core> then i boot into chaos, it loads the kernel, then exits :)
<Plundis> Crimson: huh?
<core> if it's normal it's okay :)
<Plundis> weird..
<Plundis> what version of bochs do you use?
<core> hm, not normal ;)
<core> bochs: panic, iret: return selector RPL < CPL 
-:- ElGato [dave@bespin.dhs.org] has joined #tunes
<core> | EIP=08048615 (08048614)  
<core> | EAX=00000000  EBX=00000000  ECX=00000000  EDX=00000000 
<Plundis> core.. works fine here.. 
<core> | ESP=ffeffef8  EBP=00000000  ESI=00000000  EDI=00000000 
<Plundis> with the latest version.
<core> plundis: it's still 990219a on this box
-lilo(lilo@varley.openprojects.net)- [GlobalNotice] Thanks all.  That's it for now!
<Plundis> oh.. :-)
<core> plundis: i wrote some code for Bochs, so i don't trust the latest versions :)
>>> hcf [nef@me-portland-us732.javanet.com] requested PING 932418030 from TUNES
<ElGato> hi
<abi> what's up, ElGato
<core> plundis: do --enable-port-e9-hack when you compile it, then out a byte on port 0xE9, it'll show up on the bochs console, useful for very low level debugging :-)
<core> (i did that :)
<ElGato> man he arrow paper is just too much for me
<ElGato> s/he/the
<Plundis> cool.. :)
<Crimson> core: cool bochs hack
<core> elgato: it makes a lot of sense, but brian could use more definitions of words at the beginning
* Plundis/#tunes want the reset button to work..
<ElGato> i can't understand a bit of it
<core> crimson: well, i needed it when i didn't want to be bothered setting up pages to write to the screen :)
<core> crimson: and/or setup serial port ;)
<ElGato> core: aren't you good with GIMP?
<core> elgato: i suck at anything that has to do with graphics
<ElGato> oh :(
<core> elgato: i had no talent for music either, so i became a programmer :)
<core> elgato: why? :)
<ElGato> that linux girls on your page 0wnz :)
<core> elgato: oh. bellamy did it for me
<core> elgato: and the GGI logo also
<ElGato> i see
<ElGato> i was thinking that someone good with Gimp could do the logo for tunes
<core> elgato: he's a famous artist in france, but he happens to like linux, so he did those for fun and for free
<core> elgato: tunes has a logo :)
<ElGato> um
<core> ("imagine a spiffy logo here" :-)
<ElGato> isn't it like "imagine a spiffy logo here"
<core> elgato: a logo doesn't make a project. reactOS had one, look where it is :P
<ElGato> yeah
<ElGato> oh i know
<AlonzoTG> om
<core> i like the chaos pages quite, they aren't too flashy.
<ElGato> but badass logos are cool :)
<core> elgato: well if you want to make one you're probably welcome to :)
02:00pm
<ElGato> :\
<AlonzoTG> gronk?
<ElGato> yeah i could draw some stick figures and that's about it
<core> blargh, maybe if i upgrade bochs to the latest i can finally recompile its BIOS.
<ElGato> :)-<-<
<core> elgato: hehe
* AlonzoTG/#tunes contemplates acts of terrible violance against the publishers who charge soo much for the books people neeeed. =\\\
<core> i wish kevin wasn't busy w/ freemware.
<ElGato> i could make a tunes logo in ascii
<ElGato> my ascii art owns
<core> elgato: could be fun. or do it with the gimp and run it through aalib :)
<ElGato> AlonzoTG: im with you
<core> our fucking router in D.C. is dizzy again
<core> *sigh*
<ElGato> AlonzoTG: i wish companies could be more like intel
<core> intel have real developer support, whatever you think of their cpus, that's true.
* core/#tunes downloads at pathetic speeds until the dc router is fixed :/
<ElGato> DC?  Washington DC?
<core> elgato: yeah
<ElGato> aren't you in france?
<core> elgato: yeah
<core>  6  wash-dc-1.isdnet.net (195.154.1.120)  250.528 ms *  389.388 ms
<core> see, dc :)
<ElGato> ohhh...
<ElGato> 0xc001
* AlonzoTG/#tunes is in DC...
<core> isdnet are an international backbone
<AlonzoTG> who do you want me to thonk?
<core> alonzo: go to the Netsat building and kick the cisco 2629 there in the nads :)
<core> (well, the one that says "property of isdnet", not other people's :-)
<AlonzoTG> =) on my way....
<core> hehe
<AlonzoTG> oops kar won't start. =(
<core> oh well, i'll just wait till they get my email and do something then.
<core> i'm pathetically attempting to get the latest bochs, and while it usually takes 0.5 seconds, it's going at like 3 Kb/s now.
<ElGato> !!!
<Plundis> core: not very cool..
<ElGato> rig up the 300 baud acustic modem
* Plundis/#tunes and the other guys on irc.chaosdev.org is playing with our new infobot. :)
<core> plundis: no not really :) especially since this router is supposed to take us into the US at full speed :)
02:10pm
<core> plundis: hee hee
<core> oh well, it looks like half of the american backbone is fucked tonight, so i won't complain too much
<ElGato> bah
<Plundis> america sucks.. :)
<ElGato> what happened to good old telenet
<core> plundis: nah, it doesn't. but tonight the network there has hiccups :)
<core> (well, the american backbones with whom isdnet peer, anyway)
<Plundis> core: hmm.. I'm on an american server right now, I think.
<ElGato> anyone here good with pine?
<core> plundis: me too, it's okay for irc :)
<core> plundis: probably you take other routes tho
<core> plundis: we peer with level3 and alter.net and they are throwing fits right now
<Plundis> yeah..
<core> oh well, bochs is here i think, i'll stop complaining ;)
<Plundis> core: the paging still doesn't work, althought the servers run at pl3.. :/
<core> plundis: remove the W bit then ;)
<core> plundis: set = writable :)
<Plundis> i did..
<ElGato> what is bochs?
<abi> bochs is a PC emulator at http://www.bochs.com that runs on win32, un*x, beos, mac, os/2 and amiga
<core> plundis: you can still write to the page, _while_ running at PL 3?
<Plundis> core: yeah..
<core> plundis: on a real computer, not on bochs?
<Plundis> core: haven't tested..
<Plundis> should probably do?
<core> plundis: try it yeah.. but it should work
<Plundis> but I think bochs supports paging?
<core> (in bochs too, i mean)
<core> plundis: oh, yeah, it does. i use it daily
<core> plundis: and i can assure you that writing to a R/O page causes a fault in it :)
<Plundis> core: weird.. :-/
<core> plundis: probably something silly, you'll find it (try on a real box tho :)
<Plundis> shouldn't it be enough to set it nonprivileged?
<core> plundis: set what unprivileged?
<Plundis> the pages.
<Plundis> when I map them.
<core> plundis: U bit set, W bit not set, right?
<core> (and P bit set of course)
<Plundis> P?
<core> present :)
<Plundis> yes..
<core> plundis: and you flushed CR3?
<Plundis> err... something is *seriously* fucked.
<Plundis> now it
<Plundis> oops.
<core> plundis: reload CR3 before anything tho
<Plundis> I now set neither U or W, but I can still write to the pages...
<Plundis> core: how? mov eax, cr3, mov cr3, eax?
<core> plundis: but are you reloading the pagedir (on 486 you'd invlpg, but bochs doesn't support it)
<core> plundis: right
<Plundis> or do I have to do something else?
<core> plundis: the cpu will not reread the PTE otherwise
<core> plundis: no, that works
<Plundis> core: the pagedir must be reloaded, since it doens't contain anything yet..
<Crimson> Plundis: on what are you testing?
<core> plundis: ah
<Plundis> Crimson: bochs.
<core> plundis: you can't be pointing at the right page then
<Plundis> Crimson: I think we're saving the PAGE_-stuff at the wrong place.
<core> plundis: are you sure you don't get an exception too?
<Plundis> core: yeah.
<Crimson> Plundis: i will now take a look at the ide-server
<core> plundis: must be writing at the wrong place, either the wrong PTE or the wrong page
<Plundis> Crimson: wait until this works.. or is ålen there?
<Crimson> Plundis: the eel is gone.
<core> plundis: actually i think only PL 0 can write to read only pages as i wrote earlier *duck* but it's better to keep just two levels anyway, for portability ;)
<Plundis> yeah...
<Plundis> but WHY on earth does it work, regardless on what privilege bits I set?
<Crimson> core: a doc i have here says that pl0,1 and 2 are supervisor.
<Plundis> Crimson, you wrote this code.. please come here. :)
<core> plundis: you aren't writing to the right PTE ? :)
<Plundis> core: if I don't do the write, it crashes, so I must be..
<core> crimson: i know.. but i'm not sure PL 1 and PL 2 fall through like PL 0 when WP is off.. but that's not your current problem :)
<core> plundis: what value are you writing into the PTE?
<Crimson> Plundis: the mapping works perfecty
<Plundis> Crimson: I HAAAAAAAAAAAAAAAAAAAAAAAATE YOU!
<core> (the exact 32 bit value
<core> )
<Plundis> Crimson: you suck!!!!
02:20pm
<Plundis> Crimson: you don't parse the flags!!!!!!!!!!!
<Crimson> Plundis: ehh?
<Plundis> ghaaaaaaaaaaaaaaaaaaaaaaaa.
<Crimson> what?
<Plundis> you've hardcoded the PT entries...
<core> see, my point exactly :-)
<Plundis> core: everything is crimson's fault.. :)
<core> plundis: hehe
<Plundis> Crimson: I'm gonna kill you for this.
<Crimson> Plundis: did i ever say I did that? ;)
<core> plundis, crimson: well still this mistake got your servers running at PL 3 which is saner :-)
<Plundis> Crimson: look at the code and you'll see how lame you are..
<Plundis> yeah.. perhaps..
<Plundis> I think it would have worked in PL2 too.
<core> plundis: you won't have to worry about a PL 2 stack, now..
!is:*! gline?
<Plundis> core: no, that's true..
<core> plundis: i think it would too, i really think only PL 0 can write to R/O pages without WP set, as I told crimson earlier :)
-:- ElGato has changed the topic on channel #Tunes to: Boycott V8086 mode: American traitor bitch.
<Plundis> YES! I got the beloved page fault. at last.....
<core> hehe, very good :-)
<Plundis> yeah.
<Plundis> now I'll let Crimson fix the rest. :)
<core> hehe
<Crimson> Plundis: I DID parse the flags.
<Plundis> no.
<Crimson> Plundis: what was wrong?
<Plundis> you must have removed that code.
<Plundis> you hardcoded "writable" and "supervisor" stuff...
<Crimson> look farther down...
<Crimson> NO!
<Plundis> in the vm_map_linear.
<Crimson> look below
<Plundis> you hardcoded the PAGE DIRECTORY stuff.
<Plundis> at least.
<Plundis> and that was enough.
<Plundis> weird....
<Crimson> no.
<Plundis> really weird.
<Crimson> those flags should be hardcoded.
<Plundis> yeah.. but it didn't work when they were hardcoded.
<Crimson> to allow the pagetables to overide them
<Plundis> but that didn't work.
<Plundis> I think we need to reread this chapter in the Manual.. :-)
-:- SignOff core: #TUNES (Ping timeout for core[core.suntech.fr])
<Plundis> Crimson: the flags was zero, but I could still write to the pages from pl3...
<Plundis> so the page directory shit must have overridden the page tables..
<Crimson> in that case, you suck. ;)
<ElGato> indeed
<ElGato> not suck
<ElGato> damn my slow typing
-:- core [core@core.suntech.fr] has joined #tunes
<ElGato> anyone know where retro is on cvs?
<Crimson> my eyes are becoming dimmed and the room seems tilting all the time... ;)
<ElGato> i can't find the damn thing
<core> elgato: http://tunes.org/~tcn/retro.html, should be described there
<core> elgato: otherwise it's just in the tunes repository.
<core> crimson: sounds familiar.
<ElGato> Crimson: have you been smoking crack again?
<Crimson> Plundis: read the code before blaiming me!
<Plundis> Crimson: =)
<Plundis> Crimson: sorry, you were not totally guilty this time..
<Crimson> ElGato: yes? ;)
<Crimson> I waked at six oclock this morning and went fishing... ;)
<Crimson> stupid thing or what? ;)
02:30pm
<Plundis> Crimson: cool.. =)
<Plundis> did you catch anything?
<ElGato> uhhh
<ElGato> i guess so
<Crimson> Plundis: no...
<Crimson> Plundis: but i catched some kantarells... ;)
<Plundis> cool..
<ElGato> i haven't been fishing in a while
<Crimson> i wasn't out there to catch bass or something, i was after the salmons.
<Plundis> cool..
<Crimson> yeah, kinda.
<Crimson> Plundis: is the paging shit working or is the whole thing fucked?
<core> okay, enabling DT in the latest bochs is bad :P
<Plundis> core: you were right. CPL 0, 1 and 2 are supervisor mode.
<Plundis> Crimson: it doesn't work really yet, no.
-:- SignOff ElGato: #TUNES (BitchX-75+Deb1an -- just do it.)
<core> plundis: ah well, now servers run as PL 3, no more worries about PL 2 stacks or portability of this scheme to other cpus ;-)
<Crimson> Plundis: i have very little left to do on the threading functions.
<Crimson> Plundis: it is only the cloning of the stack
<Crimson> Plundis: and i have to assign the thread a unique thread number as well...
02:40pm
<Plundis> ok..
<Plundis> Crimson: why do you call the User-bit "nonprivileged"?
<Plundis> or is it the Supervisor-bit.. hmm...
<core> plundis: intel calls it the U/S bit.
<core> plundis: i guess it'd be the 'user' bit since it makes an user page when set.
<Plundis> ok..
<Plundis> then why doesn't it work? :/
<core> you said your got your pagefault earlier ;)
<Crimson> you named it!
<Plundis> I can set the page table read/write in the PD, but it doesn't work to set privileges on the pages.. :/
<Plundis> core: yeah, but I cheated. :)
<core> plundis: oh, now you're parsing ?
<Plundis> core: yeah.
<Crimson> abi: chaos?
<abi> i think chaos is yet another micro/nano-kernel based os at http://www.chaosdev.org or small, fast and secure, blah blah *snore*
<core> ah, there, chaos works with bochs 19990719
<core> s/19/18a/
<core> plundis: you aren't using 486 code like invplg by any chance ?:)
<Plundis> nopes.
<Plundis> oh, is there a new bochs out?
<core> plundis: weird. wouldn't work with the old bochs :)
* Plundis/#tunes is doing the 0708 release.
<Crimson> storm?
<Crimson> storm is the microkernel used in a chaos system
<Crimson> storm?
<abi> storm is the microkernel used in a chaos system
<core> plundis: kevin added stuff for freemware, nothing terrific
<Plundis> core: ok..
<core> hehe, 'help' -> page fault
<Plundis> core: is vmware out yet?
<core> plundis: vmware, yes; freemware (kevin's GPL clone), far from it..
<core> plundis: vmware won't run many OSes tho.
<Plundis> oh, I meant freemvware..
<Crimson> core: the system is kind of unstable  right now.
<Plundis> core: no, chaos doesn
<Plundis> doesn't work.. :(
<core> plundis: nah, it'll be a while.. although it's making progress (and that's why bochs is lagging behind)
<core> plundis: my work doesn't start either, if that makes you feel better :)
<core> plundis: vmware does weird shit.
<Plundis> core: is bochs going to be quit or what?
<core> plundis: i don't think so, but kevin seems to be concentrating on freemware lately..
<core> plundis: is it normal i get a page fault typing 'help'? :-))
-:- ArkA [user5499@bhd4-s8.mtl.colba.net] has joined #Tunes
<core> CS:EIP = 0052:0804b040
<Plundis> core: yes.. :-)
<Plundis> it doesn't work in bochs.
<core> plundis: erm, okay :-)
<Plundis> but I want my page faults otherwise.. :/
<core> plundis: why not?
-:- ArkA [user5499@bhd4-s8.mtl.colba.net] has left #Tunes []
<core> plundis: why wouldn't it work in bochs and work on a real system?
<Plundis> core: don't ask me, but that's the way it is.
<core> plundis: uninitialized memory perhaps
<Plundis> core: or something..
<core> plundis: what's at this EIP? :)
<Plundis> don't know.. I think it's out of bound.
<core> hmmm :)
<Crimson> core: is there a need for kernel-threads in a microkernel do you think?
<Crimson> core: or should i skip that?
<core> crimson: you'll need one for your idle task..
<core> crimson: unless you want to special-case it..
<core> crimson: otherwise.. if it's really an uK, and even network stacks will run as servers, i don't know where you'll need it..
02:50pm
<Plundis> YESSSSSSSSSSSSSSSSSSSSSSSSSS!
<Plundis> it woooooorks. bring me the champagne!
-:- ArkA [user5499@bhd4-s8.mtl.colba.net] has joined #Tunes
<core> i had enough champagne for this week :-)
-:- ArkA [user5499@bhd4-s8.mtl.colba.net] has left #Tunes []
<Crimson> core: the idle task already has a thread.
<Plundis> core: network stacks will run as servers, yes.
<Plundis> Crimson: the page leve protection works now.
<Crimson> Plundis: it works?
<Plundis> when I set the writable bit manually.. the FLAG stuff seems to be rotten.
<core> crimson: a kernel one, or an actual process (swapping CR3 just for the idle task would be stupid :-)
<core> plundis: ah, okay. you can thread them a lot then.
<Crimson> core: we use tss for taskswitching
<core> crimson: then i don't know why the UK should itself run threads
<core> crimson: ow.
<core> crimson: well, does the idle task still have its own page dir, etc? :)
<Plundis> core: yes.
<Crimson> core: one idea i had was to start a kernel thread to swap out unnused mem when putting tasks to sleep...
<core> plundis: yes it has its own pagedir?
<Crimson> core: the idletask has the kernels pagedir
<core> crimson: oh, okay.
<core> crimson: wait
<core> crimson: the kernel has a page dir ??
<Crimson> core: right
<core> crimson: why does the kernel have a page directory? it's always visible from everywhere.
<core> crimson: ie, why isn't kernelspace just duplicated in every process PGD?
<Crimson> core: i guess it is just a solution for now...
<core> crimson: what does it solve? :)
<Crimson> core: it gives the kernel access to the memory... ;)
<core> crimson: but every process has to map the kernel already in its pagetables, so it can do syscalls, receive interrupts, etc :-)
<core> crimson: so the kernel already has access to it ;)
<Crimson> yes, that is correct. but the kernel needs to have access to all the memory
<core> crimson: all as in what, physical?
<Crimson> yes
<core> crimson: ick! same flaw as linux :P
<Crimson> ig you have 4 gb of ram for example!
<core> crimson: i can use 4 GB of ram without mapping all physical memory all the time :)
<core> crimson: so you loose 4 MB of memory if you have 4GB, even if you don't use all of it, for page tables? :)
<Crimson> core: tell me how to make it better
<core> (well, just the kernel's.. when mapped into processes you loose another 4 MB :)
<core> crimson: don't access physical memory until it is paged in by a process? ;)
<Crimson> core: processes only has the kernel mapped.
<core> crimson: yeah but my question is - why do you need to touch physical pages before you allocate and map them into a process space?
<Crimson> core: i dont want the processes to access their tss etc...
<core> crimson: they don't have to? that's what supervisor pages are for :)
<core> crimson: i don't think we understand eachother, tho :)
<Crimson> core: i think i do.
<core> crimson: do you actually write to physical pages until they are allocated?
<Crimson> huh?
<core> crimson: why do you map all physical memory in kernel space, instead of mapping every page when allocated :)
<core> crimson: does the kernel write stuff in pages that haven't been allocated yet?
03:00pm
<Crimson> i see what you mean, and it is a very smart way of doing it.
<core> thanks :-)
<Crimson> core: have to rethink some minor functions and implement this system some day.
<core> crimson: yeah.. as long as it works, you can change stuff and retest it still works :)
<Crimson> it shouldnt be to hard to change this...
<core> probably not :)
<Crimson> i was working on my ide server today, but it sucks! ;)
<Crimson> no, but it wont work the way i want it to.
<core> that's what i do; i believe *bsd does that as well. linux maps all physical ram, and uses the processes pgdirs, but then is limited to the size of kernel space. ie, with default config, you can't have more than 960 megs of memory. it's stupid :)
<core> crimson: think about it for a while. i don't want to influence you. i just like to talk about OS stuff ;-)
<Crimson> I want you to infulence me. ;)
<core> hehe, why's that :)
<Crimson> core: you seem having some great ideas.
<core> hmm.. thanks :)
<Crimson> core: and getting inspired by others is not in any way negative.
<core> crimson: true. i absorb a lot of influence from other people as well :)
<core> crimson: and i love to find about new OSes (like i did this morning with chaOS :)
<Crimson> core: cool.
<core> okay, i should leave you to your coding :) i work tomorrow morning
<Crimson> core: the development is really intense for the moment.
<core> crimson: good.. i hope you people don't loose motivation
<Crimson> core: can you add us to your pages
-:- SignOff hcf: #TUNES (Leaving)
-:- hcf [nef@me-portland-us315.javanet.com] has joined #tunes
<core> crimson: i will, i've been supposed to update the OS review for ~ 2 months :-)
<Crimson> core: hehe...
<Crimson> core: much to do?
<core> crimson: i have a bunch of links to check and then a cvs commit to do, i just didn't get any time for myself lately :)
<core> okay, good night, happy coding :) i can be reached at core@suntech.fr if i don't drop by here often enough :)
<Crimson> ok, see you around...
<core> see ya, and good luck! :)
-:- SignOff core: #TUNES (ircII EPIC4pre2.003 -- Accept no limitations)
<Plundis> Crimson: I can't get the paging to work correctly.. I think we're overwriting the pagetable somewhere.
<Crimson> which pagetable?
<Plundis> Crimson: 1-7 megs.
<Crimson> no, they are allocated by page_allocate_range
<Plundis> I mean, it has it flags byte set to *zero* (no write, supervisor) but it still works..
<Crimson> hmmm...
<Crimson> Plundis: core had some great ideas about memorystuff.
<Plundis> but if i set all flags to zero (in vm_map_linear, it gives a pagefault.. really strange)
03:10pm
<Plundis> Crimson: yeah, I know.
<Plundis> but let's wait with that until this works..
<Crimson> Plundis: of course...
<Crimson> hmmm... are you running under bochs?
<Crimson> what cpu does it emulate?
<Plundis> 386.
<Plundis> the pagefault abiously works.. but not as it should.
<Crimson> Plundis: we had to flee our own irc-server... ;)
<Crimson> Plundis: kill that bot.
<Plundis> strange.
<Plundis> no.
<Plundis> all the people have left anyway. :)
-:- SignOff smoke: #TUNES (zzz)
<Plundis> Crimson: get back there..
<Crimson> Plundis: but it is too much craptalk
<Plundis> Crimson: no, not anymore
<Crimson> Plundis: ok...
-:- _ruiner_ [nate@t-7-180-141.dialup.wisc.edu] has joined #tunes
<_ruiner_> microsoft?
<abi> rumour has it microsoft is definitely on the dark side of the source
<_ruiner_> windows?
<abi> i heard windows was a 32bit patch for a 16bit shell on an 8bit OS originally made for 4bit machines by a 2bit company that can't stand 1bit of competition
<hcf> hoy _ruiner_
<Crimson> abi is cool
03:20pm
<_ruiner_> hi
<abi> hi, _ruiner_
<_ruiner_> how is be a 64 bit os on 32 bit architecture?
<Crimson> _ruiner_: annoying underscores in you nick... ;)
<Plundis> Crimson: no.
<Crimson> Plundis: what?
<Plundis> it's not annyoing.
03:30pm
-:- NetSplit: varley.openprojects.net split from sterling.openprojects.net [03:32pm]
-:- BitchX+Deb1an: Press Ctrl-F to see who left Ctrl-E to change to [varley.openprojects.net]
-:- ArkA [user3674@bhd3-s29.mtl.colba.net] has joined #Tunes
-:- ArkA [user3674@bhd3-s29.mtl.colba.net] has left #Tunes []
-:- Netjoined: varley.openprojects.net sterling.openprojects.net
-:- _ruiner_ [nate@t-7-180-141.dialup.wisc.edu] has joined #tunes
<_ruiner_> my roommate is telling me that computers didn't have cache pre-pentiumII
<_ruiner_> anybody want to refute that?
<smkl> your friend doesn't know what cache means
<_ruiner_> uhhuh, tell him why
<smkl> because he is stupid
-:- SignOff Plundis: #TUNES (X-Chat!)
<_ruiner_> that doesn't fly....
<_ruiner_> explain what cache is
<_ruiner_> I've told him it contains the instruction set for the processor...
<_ruiner_> he isn't buying it
-:- hcf [nef@me-portland-us315.javanet.com] has joined #tunes
<smkl> cache usually means a container for values for faster access
03:50pm
<smkl> there are many different kinds of caches in hardware and software
<smkl> what kind of cache he means?
-:- SignOff Oddity: #TUNES (Leaving)
<_ruiner_> the kind containing the instruction sets for the processor
-:- SignOff _ruiner_: #TUNES (Leaving)
04:00pm
-:- SignOff smkl: #TUNES (sleeping ...)
-:- SignOff hcf: #TUNES (Signed off)
<AlonzoTG> om
04:20pm
-:- iepos [magister@ns4.tecinfo.net] has joined #TUNES
-:- SignOff iepos: #TUNES (iepos)
-:- ChanServ has changed the topic on channel #Tunes to: www.tunes.org
-:- binEng [Anders@dialup44-1-29.swipnet.se] has joined #tunes
-:- SignOff binEng: #TUNES (Ping timeout for binEng[dialup44-1-29.swipnet.se])
-:- ElGato [dave@bespin.dhs.org] has joined #tunes
<ElGato> hello
<abi> what's up, ElGato
* ElGato/#tunes is said to think that the only oe to greet him is an info bot
<AlonzoTG> om
<AlonzoTG> =\
<ElGato> hi al
<AlonzoTG> elgato?
<AlonzoTG> ElGato?
<AlonzoTG> o
<AlonzoTG> =P
<AlonzoTG> hick
<AlonzoTG> that mehecan...
<ElGato> siii
05:20pm
<ElGato> watcha been up to lately al?
<AlonzoTG> whining about what a piece of crap my car is...
<AlonzoTG> I just became the moderator of a list for the car I own. =P
<ElGato> :D
<ElGato> what kinda car you got?
<AlonzoTG> 1989 MX-6 GT Turbo...
<AlonzoTG> =P
<AlonzoTG> I got the thing up to 110 reciently...
<ElGato> heh
<ElGato> 0xc001
<AlonzoTG> I put custom wheels on it and now the thing shakes more than it did when it had the bent factory wheels. =((((
<ElGato> put a turbo charged on it
<AlonzoTG> I am going to take it to a specialist...
<AlonzoTG> Factory turbo...
<ElGato> oh
<AlonzoTG> 110 MPH not KPH
<AlonzoTG> =P
<ElGato> ohhh.
<AlonzoTG> That equates to 180 kph...
<ElGato> i thought you meant horse power
<ElGato> hehehe
<AlonzoTG> no, more like 240 kph...
<AlonzoTG> =P
<ElGato> btw al i live in the US
<ElGato> don't bother telling me the KPH
<AlonzoTG> eeek, I were afraid you would come north. =\
<ElGato> uh oh
<ElGato> :D
<ElGato> a los estados unidos?
<ElGato> nm
<ElGato> how's Al
<ElGato> er
<ElGato> how's Al's Great OS coming?
<ElGato> (that enter key always gets in the way)
05:30pm
<AlonzoTG> I got 50,000 pennies out of the bank today to fiks my kar, what's left will be spent developing my OS. =) 
<ElGato> pleeeaaase al
<ElGato> you don't need money to dev an os
<AlonzoTG> sure you do. :((((
<ElGato> you probably have enough books as it is
<AlonzoTG> I am missing one critical book: How to link object files.
<AlonzoTG> I also need to learn more about multithreaded programming, LiNuKs programming, and Information Theory.
<AlonzoTG> =\
<ElGato> #tunes is the greatest assembly of minds since Thomas Jefferson dined alone
<ElGato> i've never heard you ask for help
-:- s_r [sr@phila-dialup187.nni.com] has joined #tunes
<AlonzoTG> I have asked for help on linkers untill my fingers were numb and tril was kicking me out whenever I raised the subject...
<ElGato> email fare
<ElGato> hi s_r
<s_r> alonzo what's your question
<ElGato> <AlonzoTG> how do i link object files?
<s_r> use gcc
<s_r> gcc uses ld
<ElGato> he needs to write a linker
<ElGato> not use one
<s_r> read ld source code
<s_r> read source code first and ask specific questions later
<s_r> =)
<ElGato> wait a second
<ElGato> i looked on gnu.org
<ElGato> but i couldn't find the gcc source
<ElGato> mabye gcc.org
05:40pm
<s_r> the gcc source comes with the BSD source tree
<s_r> and ftp.gnu.org has gcc
<ElGato> okey dokey
<ElGato> ld will help me out but gcc won't
<ElGato> cause c isn't SEXP
<s_r> what is SEXP?
-:- _QZ [brand@p0wer.qzx.com] has joined #tunes
<AlonzoTG> I can't figure out the phreaking source because it is a paininthebutt unix tarball that I can't access in DOS and barely under my piece of shit LiNuKs installation. =(
<s_r> haha Linuks
<s_r> alonzo are you writing a compiler too?
<ElGato> oh sorry
<ElGato> SEXP == symbolic expressions
<ElGato> like lisp
<s_r> QZ, is the roger wilco server up?
<ElGato> every function has a return value
<ElGato> i had access to a mic yesterday but QZ left
<_QZ> s_r: yes
<s_r> symbolic expressions? elaborate or tell me where i can learn more
<ElGato> it's an easy concept a function along with it's arguments represents an expression which returns a value which can inturn be another arguments
<ElGato> eg (+ (- 4 5) (+ 3 7))
<s_r> why isn't C like that?
<s_r> a C function returns things and can be used as an argument
<ElGato> C's TEX or MEX or something
<ElGato> fare knows all about that crap
<AlonzoTG> eye hafto, don't eye?
* AlonzoTG/#tunes just found the perfeckt book.... In his own library. =P
<AlonzoTG> hehe
-:- SignOff s_r: #TUNES (Read error to s_r[phila-dialup187.nni.com]: Connection reset by peer)
05:50pm
-:- s_r [sr@phila-dialup095.nni.com] has joined #tunes
<ElGato> sr: does ld come with gcc?
<s_r> hmm i don't know
<s_r> i have a question myself
<ElGato> i guess i'll just have to see
<s_r> is there any way to suspend a process as a result of a system call
<s_r> like let's say i had system call wait()
<s_r> and a process called wait()
<s_r> wait() called int 80h
<s_r> and int 80h handler changed process status to _P_SLEEP_
<s_r> of course with a normal system call
<s_r> it would iret
<s_r> and then the library function wait() would just ret
<s_r> but then the program would continue running
<s_r> i want it to stop right after the system call
<s_r> how would i do that?
<ElGato> beats me
<ElGato> lemme think
<ElGato> don't use wait() the function
<ElGato> just make an inline that calls a system call
<s_r> the interrupt would still iret
<s_r> and go back to where it was in the process
<ElGato> well i don't know much about linux system calls so i cant;t tell you
<ElGato> there could be one
<s_r> ok
<s_r> i'll ask fare :)
<ElGato> well i will be going now
<s_r> elgato: get on rogver wilco
<ElGato> fare seems to know everything :D
<ElGato> ok
<ElGato> on p0wer?
-:- SignOff ElGato: #TUNES (ElGato has no reason)
06:00pm
<AlonzoTG> how do I get help on LiNuKs LiBrArIeS?
<AlonzoTG> like i.am.obfuscated.so?
<AlonzoTG> and headerfiles?
-:- SignOff AlonzoTG: #TUNES (Have Nice Day :))
<s_r> what libraries do you need help on?
06:10pm
<s_r> QZ: i'm thinking of buying an Alpha
<s_r> QZ: have you ever used an alpha?
-:- AlonzoTG [Alonzo@216-164-244-219.s473.tnt11.lnh.md.dialup.rcn.com] has joined #tunes
<s_r> alonzo, what do you need help with?
06:20pm
-:- SignOff AlonzoTG: #TUNES (Ping timeout for AlonzoTG[216-164-244-219.s473.tnt11.lnh.md.dialup.rcn.com])
-:- hcf [nef@me-portland-us202.javanet.com] has joined #tunes
-:- hcf has changed the topic on channel #Tunes to: Free Reflective Computing System <http://www.tunes.org> || slim <http://birch.eecs.lehigh.edu/slim/> || chaos <http://www.chaosdev.org> || rkt <http://www.mibsoftware.com/reuse/index.html>
<s_r> hi hcf
-:- AlonzoTG [Alonzo@216-164-244-19.s273.tnt11.lnh.md.dialup.rcn.com] has joined #tunes
<hcf> hoy s_r
<s_r> hi alonzo
<s_r> hcf: promoting vapour i see :)
<hcf> s_r: how so?
<s_r> <hcf> hoy s_r
<s_r> in some language ( i forget which) "hoy" means "is great"
<s_r> nm
<s_r> i was rambling
<hcf> hoy, an exclamation to attract attention, drive hogs, etc. --webster dict
06:50pm
<AlonzoTG> phucking winblowze 84 is multithreaded and this POS 9.11 aint. =*((((((((((((((((((((((((((((((((((((((
07:00pm
-:- SignOff _QZ: #TUNES (Ping timeout for _QZ[p0wer.qzx.com])
-:- _QZ [brand@p0wer.qzx.com] has joined #tunes
<_QZ> does anyone know of a program that pings about 10 different sites every couple seconds and sets a light green/yellow/red?
<s_r> a perl script that keeps the connection alive?
07:40pm
-:- kermit [kermit@rrosd8.impsat.net.ar] has joined #tunes
<kermit> hello all
07:50pm
<s_r> kermit dA fROG
<kermit> oh yeah
<AlonzoTG> Kermy!
<AlonzoTG> =P
08:00pm
<kermit> hey
<abi> hi, kermit
<AlonzoTG> what's your dream os, kermy?
<s_r> Windows 98
<s_r> Alonzo: what's wrong with LiNuKs?
* AlonzoTG/#tunes smashes s_r with an Iron Butterfly
* s_r/#tunes shoots Alonzo with a bullet with butterfly wings
<AlonzoTG> IT FUCKING PISSES ME OFF
<s_r> alonzo: then run Windows 98
<s_r> you can do your work from it
* AlonzoTG/#tunes runs windoze 3.11
<kermit> my dream OS ...???
<AlonzoTG> 'doze 78 is on my father's computer.
<kermit> hmmm ... dos 3.3
<kermit> (;
<AlonzoTG> hehe
* AlonzoTG/#tunes has IBM PC-DOS 7.0
<AlonzoTG> =)
<s_r> alozno are you going to switch to TUNES?
<s_r> when it is fully developed?
<AlonzoTG> Depends on what its like
<s_r> alonzo: just learn to like LiNuKs
<s_r> you know you need to
<AlonzoTG> linux is evil
<s_r> no not linux
<s_r> it's "LiNuKs"
<AlonzoTG> hehehhehehehhe
<s_r> hEheHehEheHeHehE
* s_r/#tunes lAuGhS liKe AlOnZotg
<AlonzoTG> Kermit: this is basically an OS development chan, If you create an OS from schatch, what features would it have, what would it be like?
<s_r> an object oriented system
* AlonzoTG/#tunes ignites a flamethrower and eyes s_r
<kermit> it would be totally customizable ...
<kermit> even the FS and the sched, etc ...
<kermit> how about that ?? (:
<s_r> kermit: Alonzo is having trouble developing his OS
* AlonzoTG/#tunes thinks my design could do that. =)
<s_r> Alonzo: are you having trouble?
<AlonzoTG> Sorta...
* kermit/#tunes is having problems developing his OS too
<AlonzoTG> trying to figure out how to put the prototype togeather.
<AlonzoTG> kermy: do you have web and or docs on your OS?
08:10pm
<kermit> nope ...
<AlonzoTG> seen my stuff?
<kermit> because i havent even started, thats the main problem w/my OS ... (:
<kermit> nope
<AlonzoTG> Hey, lets either work on the same project or help each other out, 
<AlonzoTG> I need business partners.
<AlonzoTG> =)
<s_r> Alonzo: why are you having trouble putting together a prototype?
<kermit> me too !
<kermit> i like the idea ...
<s_r> my modem has received 15 meg of data so far
<s_r> in 2 hours
<s_r> alonzO: eh? answer me
<AlonzoTG> you are on /IgNoRe
<s_r> why?
-:- SignOff Mr_Wrong: #TUNES (Ping timeout for Mr_Wrong[adsl-63-192-213-146.dsl.snfc21.pacbell.net])
08:20pm
-:- SignOff kermit: #TUNES ([BX] Reserve your copy of BitchX-75p3 for the Sony Playstation today!)
-:- SignOff abi: #TUNES (dying by hcf's request)
-:- ElGato [dave@bespin.dhs.org] has joined #tunes
<ElGato> hey tril
<ElGato> you awake?
-:- eihrul [lee@usr5-ppp73.lvdi.net] has joined #tunes
08:50pm
<ElGato> :\
<AlonzoTG> me kat is rippin up me blanket
<ElGato> nevermind he's away
-:- SignOff AlonzoTG: #TUNES (rebooting into linux)
<ElGato> uhhh
<ElGato> heh
<ElGato> hey qz 
<ElGato> msg _QZ y0
<ElGato> er
<ElGato> damnit 
09:00pm
* ElGato/#tunes is away: (Auto-Away after 10 mins) [BX-MsgLog On]
* ElGato/#tunes is back from the dead. Gone 0 hrs 9 min 31 secs
<ElGato> i hate auto away
09:20pm
<_QZ> what
* ElGato/#tunes is away: (Auto-Away after 10 mins) [BX-MsgLog On]
* ElGato/#tunes is back from the dead. Gone 0 hrs 1 min 8 secs
<ElGato> ok
<ElGato> what is ECC when it comes to ram?
09:40pm
<hcf> Error Checking and Correction
<ElGato> oic
<ElGato> i wonder if i could sell off all this dell software on ebay
<ElGato> i bet they get windows for like $10
09:50pm
-:- SignOff eihrul: #TUNES (Leaving)
<ElGato> what would anyone here rather have, a celeron 433 or a PII 400 (no overclocking)
<ElGato> ?
10:00pm
-:- core [core@core.suntech.fr] has joined #tunes
<core> people hi
<ElGato> core
<ElGato> !!!
<ElGato> how are you today core?
<core> hey elgato :)
<core> mmm.. sleepy, it's 7:33 AM here ;) my girlfriend woke me up and left for work, so i'm bored :-)
<core> what about you?
10:20pm
<ElGato> my brother is kicking me off :(((
<core> off where?
<ElGato> his computer
<ElGato> i must go now
<core> oh.. get your own :)
<core> okay.. ttys then
<ElGato> sorry i can't talk to you
<core> it's okay :)
<ElGato> core: i am :)
<core> cool :)
<ElGato> im gonna order it tonight
-:- SignOff ElGato: #TUNES (bye)
<_QZ> hey core
<core> hey QZ, how goes :)
<_QZ> its good
<core> what are you working on? :)
<_QZ> got any good clementine binaries?
<_QZ> i was working on boosting my starcraft score :)
<core> yeah.. i did neat improvements yesterday.. probably can give you a binary sometime this week :)
<_QZ> kewl
<core> hehe.. that's not work :)
<_QZ> :)
<_QZ> hopefully i will have the kernel debugged by the end of this week
<_QZ> then i can make a binary
<core> can release a binary then?
<core> ah cool :)
<core> we can trade l33t binaries :-)
<_QZ> heh
<core> hehe
<_QZ> so what does clementine have right now?
<core> quite a bit.. :)
<_QZ> tcp/ip and web browser? :)
<core> hehe.. no. tcp/ip and a libc for legacy software is the main things it misses
<core> (i know you won't want a libc in brix, but it's handy to put a large load on the system running stupid software, to test how well it works :)
<_QZ> i dont need libc for brix
<core> hehe.. you're predictable ;)
<_QZ> well legacy code wont work on brix
<core> well, software written for clementine doesn't need a libc either, this is just so you can recompile most useful stuff before it is really ported
<core> i know :)
<_QZ> well maybe it would work
<core> the libc will just be a wrapper for the right components
10:30pm
<core> someone else than me will do it, anyway
<_QZ> u could run the entire program in a thread and just have a dummy object to run the thread on
-:- SignOff hcf: #TUNES (Leaving)
-:- hcf [nef@me-portland-us803.javanet.com] has joined #tunes
<core> well, when ld produces output for clementine on an ANSI C program it will generate an interface for it still, as part of the stub that ld allows for every OS i suppose.
<core> i mean, you can hide your generic object definition for legacy programs in ld's target crt0.o or something
<_QZ> hmm i am somehow gaining the ability to hear monitors and tv's
<core> hear what? their radiations? :)
<_QZ> it wasnt exactly an ability i had wished for :)
<core> i have the ability to predict when my cellphone is going to ring, like most people with a 17" :)
<_QZ> huh?
<core> my screen goes berzerk when my cellphone is about to ring
<core> like loses hsync
<core> and it _is_ shielded
<_QZ> hmm, mine dont
<core> (sgi granite screen)
<core> mm, maybe it's the 1.8 Ghz we use in france
<_QZ> must be
<core> i think the US uses 900 mhz or something
<core> and they plan on switching to 2.44 Ghz here :P
<core> same frequency as microwaves :)
<core> i can't wait till i can fry chicken with my cellphone
<_QZ> hahaha
<core> hehe
<_QZ> maybe my cellfone is responsible for my changed hearing
<core> is it? mutated your genes, or you just have a better hearing?
<_QZ> or maybe my jawjicular pressure is damaging my ear drum
<core> or just sending a 900 mhz wave at 2 watts in your brain did it )
<core> :)
<core> have you checked out chaOS (www.chaosdev.org) ? it's a classical microkernel-based system but it might become nice
<_QZ> how can classical uk become nice?
<core> hehe
<core> well, by being cleanly programmed i guess. i'm not a big fan of uK's
<_QZ> u think i should remove input validation from my kernel and use trusted languages?
<core> but it's still interesting :) more projects == good
10:40pm
<core> well, if you use a safe language, there's no need to validate input by hand indeed :)
<core> you can say bye to arbitrary pointers then tho
<_QZ> ya
<core> although if you remove those, garbage collection becomes way easier
<core> without inferent pointers you can design a GC that's as efficient as someone doing proper free()s i think.
<_QZ> i already garbage collection in the kernel
<core> but it must be very conservative?
<core> how do you know a piece of memory isn't used anymore, otherwise
<_QZ> when a thread dies it frees any objects it used
<_QZ> and i want threads to die fast in brix
<_QZ> 1 second threads
<core> mmm if you're compiling stuff under brix, the compiler will be removed entirely from memory every second then reloaded from the store? :)
<_QZ> uhh, no
<core> so that thread would have to last much longer :)
<_QZ> no
<_QZ> brix uses a cache
<_QZ> being freed doesnt mean being removed
<_QZ> and the only thing the compiler compiles is single functions
<core> okay, so back to my question, how do you know an object (or a piece of memory, anything) isn't in use anymore
<_QZ> 1 compiler thread can compile a function in a second
<_QZ> the thread dies :)
<core> what if it's doing, i don't know, constant propagation on 150.000 lines of code :)
<_QZ> if anyone writes a function for brix that is 150,000 lines i will shoot them myself
<core> (or okay, global common subexpression elimination if you tell me that'd take < 1 second on pII/400 :)
<core> well, you might have anything that takes more than a second tho, so freeing everything it uses then letting it reallocate what it still uses (even if not physically removed) could be a hit? 
<_QZ> no
10:50pm
<core> no? :)
<_QZ> the object and its page table remain, it just gets put on a list of dead objects when the last thread dies
<_QZ> when the system needs more memory it will remove the first object in the list
<core> hm, so it's not GC, just freeing resources when the thread is removed :)
<_QZ> if the object is reused before it gets removed then it gets removed from the list
<core> yeah, it's a cache?
<_QZ> ya
<core> okay, just making sure :)
<_QZ> well the language helps out the kernel by telling it when objects should be freed
<core> what i mean by GC would be like being able to know at any time if a particular object is in use or can be safely discarded; and this, without a close() or decreasing a reference count by hand; that's a vast problem with no ideal solution though
<core> how does the language know?
<_QZ> it finds the last sport where it gets used
<_QZ> spot
<core> what if you use in a seldom called spot? like a routine called if the user manipulates the UI, that might never be.. will it still lock the object in memory?
<core> s/in/it in/
<_QZ> eh?
<core> how does the language figure out when the object is used last, when you have conditional branches :)
<_QZ> its intelligent :)
<core> hehe.. #define intelligent
<_QZ> the language begins compiling, testing and optimizing the code while u write it.
<_QZ> no typos, no bugs
<core> how will that help with conditional branches?
<_QZ> it will predict how often that branch branches
<core> what does it have to do with analyzing how you write the code?
<_QZ> eh?
<core> "the language begins compiling, testing and optimizing the code while u
<core>    write it."
<core> how does that help predicting branches
11:00pm
<_QZ> uhh chaos bashes linux cuz its based on stone-age OSes, but hurd is MUCH MUCH better and its free???
<_QZ> 'testing'
<_QZ> profiling
<core> so, what if i never call a function during profiling phase, that's called a lot during production :)
<hcf> _QZ: so ur bashing the chaos ppl? ;)
<_QZ> it tags the hell outta the code
<core> _qz: what if code sits and waits for external events from the user, how do you profile that.
<_QZ> core: i think yer missing something here
<core> _qz: obviously
<_QZ> if a function is not called then it is not loaded or needed
<_QZ> a function does one thing really good
<core> _qz: but what if that function not loaded or needed, is loaded by some external event occuring and needing it - that never happened during profiling phase - and your language had said it was safe to free an object, that this seldom called function uses.
<_QZ> that wont happen
<core> why not?
<_QZ> and if it does it wont hurt much
<core> it'll just crash :P
<_QZ> no it wont
-:- Plundis [plundis@chaosdev.org] has joined #tunes
<core> well, it will, unless you systematically write everything back to the store, even things that needn't be
<core> hey plundis, awake already?
<_QZ> 1) most functions will only use an average of 1 object
<Plundis> core: yeah. =)
<core> plundis: same, girlfriend woke me up, left for work, and left me bored :)
<Plundis> core: we'll have to be up at 8 here..
<core> (well, i'm leaving for work sometime too, but no way in hell i'd go at 7:30 AM :)
<_QZ> 2) that 1 object will last until the thread dies cuz the parent opened it
* Plundis/#tunes is doing his service right now.
<core> plundis: you have net access from military service? cool.
<core> _qz: so you're keeping objects in memory that could be freed, and yoru language doesn't help :)
<core> s/yoru/your/
<_QZ> they are objects that shouldnt be freed
<core> _qz: but how do you know that :)
<_QZ> and the average lifespan of an object is 10ms
<core> oh well, when brix and its sourcecode come out i'll look at them
<_QZ> ya :)
<hcf> core: 9yrs from now
11:10pm
<core> hcf: it's already been 3, we're progressing :)
<_QZ> hcf: yer so funny i dont know if i should laugh or bash yer face in :)
* hcf/#tunes hands _QZ a bat
-:- Mr_Wrong [seanl@adsl-63-192-213-146.dsl.snfc21.pacbell.net] has joined #tunes
<_QZ> Plundis: did u write the web page?
<core> plundis: did you get your code to work last night?
<Plundis> core: yeah, it's pretty nice.
<Plundis> core: yes, finally. :)
<core> plundis: cool :) new version soon?
<Plundis> _QZ: mostly, heah.
<Plundis> core: perhaps.. but we'll have to fix some other bugs first.
<_QZ> Plundis: or is based on
<_QZ>        old stone-age operating systems (Linux). Some of them are much better, and free (The Hurd
<core> plundis: like the pagefault w/ 'help' *snicker* :)
<_QZ> damnit E just crashed
<Plundis> _QZ: do you have any objections?
<_QZ> Plundis: did u know linux is free? and how is hurd better, cuz its uk? it runs the same shit on top
<_QZ> Plundis: linux just runs that same shit faster
<Plundis> _QZ: yes, I know Linux is free.
<Plundis> _QZ: Hurd is better, since it is less monolithic
<_QZ> Plundis: i understand how u can bash linux cuz it does have many many flaws but hurd is no better
<Plundis> _QZ: I think it is.
<Plundis> it's filesystems run in userspace, for example.
<_QZ> Plundis: u need to group hurd,linux,freebsd,all un*x oses together and call them all bad
<Plundis> and the networking too. that's truly a step in the right direction.
<_QZ> but slower
<Plundis> that'
<Plundis> oops.
<Plundis> that's just a matter of implementation.
<_QZ> if yer going for speed linux is the right choice
<Plundis> remember that Hurd isn't really written from scracth. it runs on an old poor kernel.
<Plundis> _QZ: FreeBSD is a better choice, IMO.
<core> mach is the worst piece of crap ever known to man
<Plundis> core: yes, absolutely.
<_QZ> core: agreed :)
<Plundis> can't really understand how anyone can call that one a "microkernel"..
<core> well, i can find worse, but i am impressed at all the bad design choices Mach took :)
<core> plundis: they mean it doesn't do much :-)
<_QZ> Plundis: u should just say all un*x systems are bad and that chaos is better :)
<core> i mean, when you look at something like L4, you know there's some progress to be made in uk's
<core> a message will never be as fast as a direct call, but there are ways to make them faster :)
<Plundis> _QZ: I think that's what we're already trying to say.. :)
11:20pm
<Plundis> core: how does L4 do it?
<_QZ> Plundis: but in doing so u say that hurd is better than linux but u dont say why, and if u do then u need to say that linux is faster than hurd to be fair
<core> plundis: first, it never copies the contents; it directly drops the message in the destination process space (normally you have two copies, one from caller->kernel then kernel->destination); and it also passes short messages in registers
<core> (as you know, most messages are short)
<Plundis> core: that's how we do it to.. at least the non-copying part.
<_QZ> Plundis: when i goto os project pages and they make false/incorrect comments about other oses then it tells me they dont know much about the competition. and they really should know how the competition does stuff so they dont repeat mistakes.
<Plundis> _QZ: speed is not everything. :-)
<Plundis> _QZ: okay, I'll fix it.
<core> plundis: coolness. so your message send() operation is atomic?
<_QZ> Plundis: what about server applications? speed is everything
<core> _qz: easy, the chaosdev page is one of the less mistaken ones i saw ;) i read things like "windows NT is based on Mach" on some os project page :)
<core> (did you notice that VMS + 1 = WNT? heh)
<Plundis> core: whddaya mean atomic?
<_QZ> core: that is what im talking about :)
<core> plundis: it can't be interrupted until complete
<Plundis> core: the send() operation is currently non-blocking.. but I guess it should be blocking.
<core> _qz: that comment absolutely killed me (on jenios page or something)
* Plundis/#tunes will be away for a while..
<core> plundis: i'm just wondering what happens if the destination thread happens to be checking if it has messages, while you're sending one to it :)
* core/#tunes shudders at amiga, linux, java and "leveraging" used in the same sentence
<Plundis> core: the syscall_send_packet() is so fast that will never happen.
<Plundis> core: at least not in non-SMP systems.
<Plundis> oh, I should be away.. sorry.. :)
<core> plundis: if you use preemptive multitasking, it will happen :)
<core> plundis: ok, ttyl :)
<core> "never" isn't a word that goes with "computers" :)
<_QZ> core: what dont u think its funny that amigaOS 4 will use linux :)
<core> _qz: not really :)
<_QZ> brix eliminates the need to have threads check for messages
<core> _qz: i was reading that 'technical' paper which is just marketing mumbo jumbo destined to get funds, really (i'd know, i just got some)
<_QZ> cuz the message was sent to the thread when it started :)
11:30pm
<core> _qz: amigaOS was a really smart piece of code, i'm sad it's being discontinued
<core> especially to be replaced by *linux* of all OSes. good grief.
<_QZ> heheh
<core> the suits strike again i guess :)
<_QZ> do any ggi drivers exist?
<core> what do you mean? there are lots of ggi drivers :)
<core> i booted clementine on my riva TNT2 *big grin*
<core> (somehow the graphics test were fast on it :-)
<core> s/test/tests/
<_QZ> do enuf ggi drivers exist?
<core> there are ggi drivers for most boards on the market
<_QZ> then why dont u try and get amiga to use clementine
<core> some of them aren't "commercial quality", but it's just a matter of someone having such a card polishing the code
<core> _qz: clementine isn't nearly as functional as linux yet..
<_QZ> they dont need funtional they need graphics :)
<core> _qz: and my guess is that amiga are more interested by the momentum around linux (they mention that like 5 times) than actual technology
<core> _qz: well, next year maybe i'll tell them :)
<_QZ> gotta tell em now before they get hooked on linux
<core> well, they mention using another RTOs for their "appliances"
<core> s/s/S/
<core> so they must have some kind of stupid abstraction to the kernel i guess
<core> so it'll still be time :)
<core> they're doing the complete opposite of what made amiga a success, oh well :)
<_QZ> qnx?
<_QZ> i could never understand how amiga got anywhere with that butt ugly ui
11:40pm
<core> well i'm glad they dropped qnx, but no i mean, amigaOS was tightly integrated and coupled with h/w
<core> true the UI wasn't too pretty, but it was extremely functional, and fast, even on a 7.09 mhz cpu :)
<_QZ> but it was amiga hardware
<core> and remember, it came out in 1984.. at the time the mac had 2 colors, the pc was doing text mode only :) and neither was preemptively multitasking
<core> well, the hardware was good, but the software was even better.
<_QZ> its easy to couple an os to yer own hardware
<core> well, they had done it intelligently :)
<core> and amigaOS wasn't a pig on top of it
<core> i saw one in 1989 or so at a friend and had an A2000 two days later :-)
<_QZ> when u have to support a million video cards and other devices then u need to uncouple everything 
<core> true
<core> but that's the pc for you :)
<core> the amiga hw was probably too expensive tho, it even did color clock computations and such, for videomodes, on its own.
<core> had what PCI rediscovered (mmio and device IDs) in 1984 tho :)
<_QZ> the amiga hardware was the best back then but that was back in the hacker days when it was easy to think up ways to be 100 times better then competition
<core> that, and you didn't have to shell out too much to be competitive :)
<core> okay, i must dash to work :) see ya :)
<_QZ> cya
-:- SignOff core: #TUNES (vroom!)
11:50pm
[msg(TUNES)] newlog 1999.0720
IRC log ended Tue Jul 20 00:00:00 1999