00:00:00 --- log: started forth/04.05.18 00:00:01 kc5: I apologize, either way. 00:00:10 synova: I agree with your principles mostly 00:00:26 but your method just got kc defensive. 00:00:33 and I started doing it with you for a little while 00:01:15 defensive means I'm touching nerves 00:01:21 I know I'm not heeding my own advice real well right now, but I don't know how else to say this. 00:01:25 that's what happened to me 00:01:27 it had a negative efect 00:01:44 we should find a better way to share our knowledge 00:01:46 I can accept that 00:02:01 --- quit: rO| ("maintenance bbl") 00:02:13 and it's not just with the personal growth stuff 00:02:37 I've wasted lots of hours trying to convince people of things or change their attitudes about programming and lots of stuff 00:03:00 and in the end all I did was piss them off and waste lots of time 00:03:15 I don't feel like I wasted my time here. 00:03:20 granted, I did piss people off. 00:03:22 I know this doesn't work and I'm trying to learn a better way 00:03:25 but, definitely I did not waste time. 00:03:42 synova: what have you accomplished or gotten out of this conversation? 00:04:16 well, I can't honestly state that. 00:04:18 what kc5tja heard was the same old @#$% that he's gotten from a list of other people. 00:04:29 but frankly, I feel like it was a lot. 00:04:30 and he got pissed and left 00:04:34 even if I cannot put it into words. 00:04:51 are you thinking you helped others? or that you learned something? 00:04:59 I learned something for myself. 00:05:03 I don't know if I helped others. 00:05:04 bonus! 00:05:07 I'm content with that. 00:05:14 it is all I can do. 00:05:34 what did you learn? 00:05:41 at least, with my framework as I know how to use it at this point in time. 00:05:46 maybe I could benifit from this. I've been a jackass here today too :) 00:06:02 herk: you have not been a jackass. 00:06:09 that was my job =) 00:06:16 :) 00:06:32 I'm afraid it is too early to express what I got out of this 00:07:18 sure I have. first I contributed to alienating kc5tja, then I was being a hypocrite by tell you that you were doing it wrong, and that you should go around telling people that their way is bad. 00:07:29 at the very least, it saw what I used to bathe in every day with others 00:07:36 the phenomenal negativity 00:07:51 the "world constantly getting worse" with "no place for me in it" 00:07:54 day in day out 00:07:57 justifying yourself all the time 00:08:09 finding a way to "substantiate" yourself, just living on the earth 00:08:14 its absurd! 00:08:21 :) 00:08:46 yeah, I have some friends that live in very dramatic worlds 00:09:15 Bedtime for me. 00:09:22 kc5: likewise 00:09:25 --- quit: kc5tja ("THX QSO ES 73 DE KC5TJA/6 CL ES QRT AR SK") 00:09:42 he's got the chat logs 00:09:44 I hope he mails me 00:09:50 even if he doesn't like the book. 00:09:53 he won't 00:10:02 well, I won't tell anyone if he does. =) 00:10:14 I guess that is one positive thing about channel loggers. 00:10:29 what? 00:10:48 this message/chat hopefully nailed for posterity 00:11:09 the whole point of channel loggers is so you can go back later and find something you only partially remember. 00:11:18 or so you can tell others to read it 00:11:29 for posterity 00:11:30 herk: I guess surveillance does have its small benefits. 00:11:31 as you say 00:11:34 yeah :D 00:11:40 it's not intended as surveillance 00:11:47 this is a channel for discussing programming 00:12:01 the logs contain a wealth of programming knowledge 00:12:33 anyhow, I'm thankful for both you and kc5's time. 00:12:35 I mean that 00:13:02 I've stayed up way past my bedtime. 00:13:03 (and a lot of political, psychological, mechanical, medical, etc rantings/arguments/discussions like ours) 00:13:14 herk: always a good thing. 00:13:43 have a pleasant night 00:13:58 (or day, depending on your location.) 00:13:59 I don't like logs for personal channels (except maybe just for the last day or something in case you got disconnected and wanted to say the end of a conversation) 00:14:07 but for technical info they are very very useful 00:14:20 synova: you too. goodnight 00:14:32 thank you. take care. 00:14:37 --- quit: synova ("...") 00:35:12 --- join: crc (~Charles_C@0-1pool176-7.nas6.philadelphia1.pa.us.da.qwest.net) joined #forth 01:00:16 --- join: rO| (rO|@pD9EE1D96.dip.t-dialin.net) joined #forth 01:00:44 how is it goin warpzero? 01:07:58 just 10 seconds ago got a kernel to boot 01:09:03 is there any good initrd-making tools 01:09:23 congrats ;-) good timing :-) 01:09:33 yeah now i just need to make an initrd 01:13:28 --- join: fridge (~Jim@CommSecureAustPtyLtd.sb1.optus.net.au) joined #forth 01:13:55 --- quit: fridge (Read error: 54 (Connection reset by peer)) 01:18:59 warpzero: why do you want netboot? 01:22:35 --- quit: crc ("ChatZilla 0.9.61 [Mozilla rv:1.7b/20040316]") 01:24:37 Herkamire: netboot is cool for via epic mobos eg. 01:26:24 via epic mobo is 15x15cm 1ghz incl graphic, silent no fan and even relatively cheap 01:26:59 a sweetie 01:29:14 Herkamire: bcuz i am moving to a single-filesystem server->diskless setup 01:32:37 --- join: fridge (~Jim@CommSecureAustPtyLtd.sb1.optus.net.au) joined #forth 01:50:25 rO|: very cool 01:54:51 Herkamire: :-) you mean the mobo? 01:56:50 yeah. how much do they cost? 01:57:06 what kind of processor? 02:05:30 via's intel clones. prize 100/100+ depends 02:05:38 --- join: Serg (~z@212.34.52.140) joined #forth 02:05:57 Herkamire: sorry, no ppc ;-) 02:06:06 :-P 02:06:48 --- quit: Herkamire ("ok, bedtime") 02:26:41 --- quit: SDO (Read error: 54 (Connection reset by peer)) 02:35:12 --- quit: madwork_ (Read error: 104 (Connection reset by peer)) 02:36:47 --- join: madwork_ (~madgarden@204.138.110.15) joined #forth 02:42:15 --- quit: fridge (Remote closed the connection) 02:43:18 --- join: fridge (~Jim@CommSecureAustPtyLtd.sb1.optus.net.au) joined #forth 02:51:16 hmm 02:51:20 init won't start 02:59:29 ah fuck 03:01:34 sometimes channel looks like radio in a battle ;) 03:02:07 pieces of curses and screams, "A-a-ah! I'm hit ! Guns jammed..." 03:06:55 --- quit: madwork_ (Read error: 54 (Connection reset by peer)) 03:14:55 Serg, it's a mental battle 03:15:09 sometimes you get defeated by the problem =( 03:15:35 problem is victor! we lie in the corner, swearing to ourselves 03:37:00 nice said ! 04:32:49 --- join: qFox (~C00K13S@cp12172-a.roose1.nb.home.nl) joined #forth 04:47:17 --- quit: warpzero (Read error: 113 (No route to host)) 04:48:22 --- join: warpzero (~warpzero@dsl.142.mt.onewest.net) joined #forth 06:23:13 --- join: madwork_ (~madgarden@derby.metrics.com) joined #forth 06:44:13 --- join: madwork__ (~madgarden@derby.metrics.com) joined #forth 06:45:12 --- quit: madwork_ (Read error: 104 (Connection reset by peer)) 06:45:20 --- nick: madwork__ -> madwork 06:54:57 --- quit: fridge ("Leaving") 07:01:12 --- quit: I440r_ (Read error: 54 (Connection reset by peer)) 07:09:50 --- quit: Serg () 07:10:03 --- quit: warpzero (Read error: 110 (Connection timed out)) 07:10:43 --- join: warpzero (~warpzero@dsl.142.mt.onewest.net) joined #forth 07:42:07 --- join: I440r (~mark4@168-215-246-243.gen.twtelecom.net) joined #forth 07:42:07 --- quit: I440r (Client Quit) 07:42:11 --- join: I440r (~mark4@168-215-246-243.gen.twtelecom.net) joined #forth 08:11:18 --- join: proteusguy (~proteusgu@69.79.24.31) joined #forth 08:14:44 --- quit: I440r (Read error: 110 (Connection timed out)) 08:46:53 --- join: Herkamire (stjohns@h000094d30ba2.ne.client2.attbi.com) joined #forth 09:19:30 --- join: SDO (~SDO@co-trinidad1a-156.clspco.adelphia.net) joined #forth 09:20:24 hello 09:20:26 some discussion here of Think and Grow Rich, and personal development I see. 09:20:28 I can only recommend what worked for me, if it doesn't work for someone else, so be it. 09:21:04 There is always one way to find out if some text is worth reading, and that is to read it, decide if you want to take the steps it recommends, and find out if it works. If the results come, then it works, if not, look somewhere else, or within the same to find what is missing and incorporate. 09:27:03 think and grow rich? 09:28:15 sounds like the header of spams i get 09:28:16 :p 09:37:47 lol 09:38:25 fine observation, though 09:41:57 a bit too obvious... "is dead permanent, lets try!" 10:30:00 --- quit: SDO (Read error: 104 (Connection reset by peer)) 11:21:05 --- join: crc (~Charles_C@0-1pool88-136.nas48.philadelphia1.pa.us.da.qwest.net) joined #forth 11:40:21 --- quit: crc ("ChatZilla 0.9.61 [Mozilla rv:1.7b/20040316]") 12:04:45 rO|: well the kernel boots and the ramdisk works 12:04:58 but when it runs exec /sbin/init it just reboots 12:07:59 you went through the init script already, yes? 12:09:14 is it an error reboot or a faulty devised reboot 12:12:34 no dude 12:12:36 it just reboots 12:12:48 i can start a bash shell 12:12:51 but i can't start init 12:16:02 ld=ld-linux.so.2 12:16:22 it's i86, yes? 12:16:35 ahA! 12:16:46 i guess if init can't make /dev/initctl it does that 12:19:50 better a friendly dude than an expert asshole ;-) lol 12:25:57 --- quit: proteusguy (zelazny.freenode.net irc.freenode.net) 12:26:30 --- join: proteusguy (~proteusgu@69.79.24.31) joined #forth 12:54:33 okay 12:54:37 its mostly working now 12:54:41 /etc/mtab is still all fucked up 13:05:55 who said recently gentoo is a nobrainer install? 13:13:58 --- join: wossname (wossname@HSE-Hamilton-ppp190701.sympatico.ca) joined #forth 13:16:20 well its not if you install it thisway 13:19:20 do we have any lisp experts/knowers here? 13:45:55 --- join: I440r (~mark4@168-215-246-243.gen.twtelecom.net) joined #forth 13:46:12 --- join: rO|_ (rO|@pD9EE1A10.dip.t-dialin.net) joined #forth 13:47:03 Hi 13:48:43 --- join: kc5tja (~kc5tja@66-74-218-202.san.rr.com) joined #forth 13:49:02 --- mode: ChanServ set +o kc5tja 13:54:22 --- quit: rO| (Read error: 60 (Operation timed out)) 14:16:23 --- join: crc (~Charles_C@0-1pool176-19.nas6.philadelphia1.pa.us.da.qwest.net) joined #forth 14:21:45 --- join: lalalim_ (~lalalim@pD95EA0C0.dip.t-dialin.net) joined #forth 14:25:53 --- quit: lalalim (Read error: 60 (Operation timed out)) 14:26:29 hi 14:26:39 re ianp 14:26:39 Hi ianp 14:38:23 Good lord. 14:38:27 I don't believe what just happened. 14:38:33 I *really* don't believe what just happened. 14:38:43 --- nick: rO|_ -> rO| 14:39:03 I've actually come full circle. Totally, utterly, 100% full circle. 14:39:14 What? 14:39:40 When I first wrote Dolphin, it was intended to be an OS within which I ran primarily Forth software, heavily inspired by the user interface and ease of programmability of AmigaOS (obviously not restricting myself to Forth, therefore). 14:40:05 over time, Dolphin moved away from that goal, slowly becoming more similar to other operating systems. 14:40:31 Dolphin 0.5 Spec I (the last spec I made) was almost very much like a cross between EROS and Unix. 14:40:39 As if that doesn't happen all the time... 14:40:47 (e.g., it treated handles as capabilities, but still relied on ACLs to derive its INITIAL security state from) 14:41:38 And now I just realized, if I make the Dolphin kernel (a) orthogonally persistant, (b) 100% object capability protected, and (c) memory protected, I will have an environment that is *perfect* for FTS/Forth to run in! 14:41:56 Completely by accident, Dolphin and FTS/Forth, which split their ways back in the mid 80s, have come back together! 14:42:02 ot, but capabilities are just the "integer key security" type thingy? 14:42:23 nice 14:43:03 wossname: No. The handles are just small integers. A true capability can be referenced by other means. But the core is the intent: a capability refers to an object *AND* some functionality on that object (e.g., an object is a state plus its interface, while a capability is some state plus *ONE* method of its total interface). 14:44:12 wossname: There is a document on erights.org that explains why capabilities are markedly different than just "another view of the access control matrix." 14:44:39 i read a doc by this shapiro bloke on them, seemed neat but too simple to work 14:45:01 It's the difference between algebra and calculus -- with discrete math, you can ultimately derive anything that calculus can, given enough terms in the series. But calculus can do things *instantaneously* and represent the concept of an infinitesmally small time-frame. 14:45:01 i'll put that url away for later reading 14:45:18 wossname: Capabilities MUST work. 14:45:20 There can be no other alternative. 14:45:29 Consider, if you lack a pointer to an object, how can you muck with it? 14:45:55 overwrite every part of memory with your modification 14:46:01 wossname: You can't do that. 14:46:04 you'll hit the object eventually 14:46:05 You don't have access to its memory space. 14:46:16 well, then you take a hammer to the ram chips 14:46:19 You *physically* cannot express a pointer to the object in question. Period. 14:46:27 wossname: Well, gee, ACLs can't handle that either. 14:46:29 Get real. 14:46:58 call rand() and cross your fingers? 14:47:03 wossname: Nope. 14:47:14 Doesn't work. 14:47:44 is there something besides odds preventing it from working? 14:48:02 You can call rand() in Unix and hope to hit upon some kind of file handle, but unless the kernel has an entry in YOUR process' handle table, which you CAN'T get unless (a) either the parent process put it there, or (b) you invoked open()/socket()/etc., you simply arne't going to hit anything. 14:48:02 combination of needing the right handle and needing the right user id, or something? 14:48:10 wossname: Yes. Mathematical certainty. 14:48:27 I already said: if you just plain don't ahve a pointer to an object, you just plain can't muck with an object. PERIOD. 14:48:43 There is no concept of a user ID in a capability secure OS. 14:48:52 It's just plain not needed. 14:48:55 Not, at least, at the kernel level. 14:49:07 Maybe at the "init-process" level, but definitely not inside the kernel. 14:49:26 hi all. 14:49:38 just processes with capabilities 'rented out' to them. ok, i can buy that 14:49:53 More accurately, capabilities given to them. 14:49:56 kc5tja, it might be hard to fully secure this arrangement -- consider objects 'leaking' out 14:50:06 But the act of giving a capability ITSELF requires a capability too. 14:50:12 So not any ol' program can hand out capabilities. 14:50:39 slava: It's not possible. See both EROS and KeyKOS for two prior art OSes that are 100% secure against what you describe. 14:51:27 For example, in Linux, if I open a file, there is no possible way for my file handle to leak out of the process, and even if it DID, the number is meaningless, because it has meaning ONLY within the context of MY process. 14:51:50 of course 14:52:08 but in java, you can have problems where an object gets a reference to another object it shouldn't have access to 14:52:15 because its all in 1 address space 14:52:33 Well . . . um . . . don't do that. 14:52:53 Capabilities, and ACLs, are useless in a single-address space environment without single-address-space MMU hardware. 14:53:54 Do not proscribe the faults of a specific language to that of the general concept. 14:54:10 E demonstrates that it IS possible to use capability security on the JVM too, since E runs on the JVM. 14:54:34 But also consider, too, that E is a "safe" language too. 14:54:54 By the sounds of things, safer even than Java. 14:57:21 E is interesting, i forgot about that. 15:01:10 hey ianp 15:01:50 ianp, you use a mac right? 15:03:00 Damn, the thought of having a complete OS for the PC that treats Forth on par with other languages is damn appealing.... 15:03:03 yes 15:03:08 sup slava 15:03:15 ianp, ever used UserLand Frontier? 15:03:20 kc5tja: no joke. 15:03:30 slava nope 15:03:37 is that a blog deal? 15:03:47 well that's how they market it 15:03:52 webserver too i see 15:03:54 but its actually a scripting language and a database 15:03:57 slava: That's the first XML-RPC application -- XML-RPC was actually invented precisely to support that application. :D 15:03:58 why would i use it? 15:04:00 the nice thing is they're integrated 15:04:05 you write workspace.foo = 5 15:04:08 kc5tja: heh, no way 15:04:10 and it saves worksapce.foo to the database 15:04:18 slava: heh, cool :) 15:04:22 ianp: Yep, it's true. 15:04:28 ianp, and code is in the db as well. 15:04:38 ianp: That's why Frontier's author won't SHUT UP about XML-RPC, and why it's so much better than SOAP. :D 15:04:40 ianp, frontier is one of the more innovative mac os programs (though they ported it to win32 too now) 15:04:47 (which is something I'd do too if I invented a new technology) 15:04:56 kc5tja: well, anything lighter weight than SOAP sounds like a good thing 15:05:12 3eheh 15:05:18 ianp: Yep. I do wish XML-RPC was a bit more rigorous though. But, it's all good. It seems to be working quite well. 15:05:20 slava: cool... 15:05:37 kc5tja: I intend to check out ASN.1 if i ever have a need for that kind of stuff, too 15:05:50 ASN.1 is WAY complicated, from what I understand. 15:06:02 ianp, i'm adding a similar capability to factor -- object database -- but i'm taking it further, with support for BeOS-style 'live queries'. 15:06:07 * kc5tja prefers CORBA myself. IIOP is very, very simple. 15:06:29 ianp, at the moment, it stores a hierarchical object database; i'm working on adding the capability to persist word definitions in the database 15:08:34 My ideal object model system would use a free version of Component Object Model for in-process objects, IIOP for RPC. 15:08:45 Too bad IIOP is incompatible with COM's concepts. 15:10:06 Although, the use of capabilities for invoking methods across machine boundaries would likely invalidate the use of IIOP anyway. 15:10:23 (since IIOP isn't "secure") 15:11:48 i'm stuck on a design issue right now 15:11:56 what if the user stores an object in the object db that cannot be serialized 15:12:00 eg, a network socket 15:12:58 that's interesting. 15:13:12 at the moment it just ignores the object when the table is loaded from disk again 15:13:22 some sort of orthogonal persistence is definitely something that's desirable 15:13:29 yes 15:13:58 --- quit: proteusguy (Read error: 104 (Connection reset by peer)) 15:14:04 what is the meaning of the term orthogonal in the name though? 15:14:28 it's saying it's a separate "process"(bad word i guess?) that intersects 15:14:46 --- quit: crc ("ChatZilla 0.9.61 [Mozilla rv:1.7b/20040316]") 15:14:50 i guess that does make sense 15:15:00 ianp, once the db is a bit more complete i will extend my httpd to serve records from the db 15:17:09 ianp: Orthogonal means that the processes that are running do not have to manually manage their program storage. That is to say, there is no concept of files, no concept of directories, etc. 15:17:14 No filesystems. 15:17:42 is the issue of 'exporting/importing data to an orthogonal store in a portable manner' been solved? 15:17:55 Think of a Palm-compatible device, and you'll immediately know what I mean by Orthogonally Persistant. 15:18:26 slava: It was never a problem. You access external storage volumes and filesystems through normal, already familiar means. 15:18:37 kc5tja: yeah I definitely grasp the term 15:18:39 What I'm saying is that the run-time environment of the system is orthogonally persistent. 15:18:42 there was a good anecdote on that 15:18:51 i think in the TUNES or some related project's wiki 15:19:08 about how someones grandmother just turned off the power switch and wondered why their document wasn't saved ....:) 15:19:15 heheh 15:19:19 ianp: Huh? 15:19:25 that doesn't make any sense. 15:20:39 Oh, you mean, as a problem that orthogonal persistence _solves_? 15:20:39 tunes creates alot of interesting papers 15:20:43 and almost NO code 15:21:27 haha 15:22:03 slava: Actually, you CAN persist network socket information. I've seen plenty of cases where I'd pull an Ethernet cable from the computer (100% equivalent to just shutting off the computer as far as the remote side is concerned), and after plugging it in again, the same sockets work just fine. 15:22:10 slava im using jedit now for some things, i still go runnin home to mamma when im in a hurry tho (joe :) 15:22:28 kc5tja, yes 15:22:36 kc5tja: yes, that was a case where orthogonal persistence would have made the computer work as the person expected 15:22:36 I440r, cool ;) 15:22:45 I440r, i admit i use vi sometimes too :) 15:22:56 eek! 15:22:57 I440r: yeah, it is kind of funny, but I have respect for anyone even trying anything which is more than what I do 15:22:58 ianp: OK, I just wasn't sure from which direction you were coming from. :) 15:23:05 *VIM!* 15:23:08 kc5tja: yeah I explained a bit poorly 15:23:28 ianp respect me for trying jedit ? hehe 15:23:42 or tunes for trying to get their project going 15:23:51 haha 15:23:59 nah tunes of course 15:24:03 lol 15:24:13 the world needs dreamers 15:24:18 <-- dreamer 15:24:24 Tunes lost most of my respect. Their goals are so mutually contradictory that the only way to achieve them is by breaking the laws of thermodynamics. 15:24:34 I could be negatiev, but I don't know 1/10th of what anyone in TUNES does, so I wont make myself look dumb by even attempting to comment. 15:24:43 kc5tja: hehe yeah 15:24:53 does seem a bit of a tail-chase 15:24:55 I value TUNES *purely* as a scientific paper repository, which has mostly reasonable reviews of them too. 15:25:22 And I hope it stays that way. 15:25:24 Personally. 15:25:27 They excel at that. 15:25:29 kc5tja: indeed i see value there too, - figuring out what is the struggle, what are the problems, that is helpful for a novice like me 15:25:40 * kc5tja nods 15:25:44 'how COULD things be better' 15:25:49 They should leave the OS writing to the "experts." :D 15:26:01 and let Darwinism take its course. 15:26:02 are there any os writing experts ? 15:26:03 lol 15:26:17 I440r: Yes. Carl Sassenrath is one. 15:26:47 kc5 i KNOW there are experts out there lol - i was being funny lol 15:26:58 Heh 15:27:01 i would say jef raskin knew a bit too :P 15:27:05 but he is weird 15:27:10 Some might actually claim I am one, having written a few OS kernels myself. But I don't claim that I am. 15:27:16 I have yet to successfully master memory protection. 15:28:19 what will the general design of boxOS be like? 15:28:58 slava: BoxOS is no more; there is already a project of that name. It's now just BoxForth. 15:29:14 And, for the Kestrel, it'll just be a single-address-space Forth implementation that supports multitasking. 15:29:17 BF :) 15:29:24 BrainF.... 15:29:53 orth? ;))) 15:30:10 motherforther 15:30:49 slava: In a later generation of the processor, I'm planning on experimenting with an MMU; in that case, I may replace BoxForth with Dolphin, where BoxForth would run as a libOS (in the exokernel sense) 15:31:53 i see 15:32:59 You sound somehow disappointed. 15:33:50 why? 15:37:08 --- join: blockhead (default@dialin-920-tnt.nyc.bestweb.net) joined #forth 15:41:44 grr 15:41:51 i cannot tell if linux supports this card or not :( 15:42:12 [16:30:51] slava: In a later generation of the processor, I'm planning on experimenting with an MMU; in that case, I may replace BoxForth with Dolphin, where BoxForth would run as a libOS (in the exokernel sense) <-- what 15:45:51 some things kc5tja says just confuse the hell out of me 15:45:57 :D 15:47:57 its more like a learning wall around here than a learning curve 15:48:08 one of those learning walls that you slam into at 60mph 15:48:35 well, if you ask for calrification, kc5tja is very good about it 15:48:47 clarification, even :) 15:48:56 that doesn't prevent that dizzy feeling though 15:49:02 agrred 15:50:57 I need an UATA133 card 15:51:23 I need to get laid :D 15:51:37 Sex is over-rated 15:52:56 Unlike Forth? 15:52:59 An exokernel is a kernel that provides *no* policies of its own, save one: Don't fsck with the exokernel, and don't fsck with other applications. 15:53:14 Robert gets us back on topic :D 15:53:40 Hehe. At your service 15:53:45 If Forth provided protection services to prevent trashing Forth kernel memory and other application memory, AND securely multiplexed physical devices across all running tasks, literally, it'd be an exokernel. 15:54:20 --- join: Robert__ (~snofs@213.113.90.191) joined #forth 15:54:22 A cache kernel works similarly, but subtly differently. 15:54:59 wait 15:54:59 --- quit: Robert (Nick collision from services.) 15:55:04 --- nick: Robert__ -> Robert 15:55:16 so an exokernel doensn't do the more advanced kernel things like drivers and such? 15:55:22 Nope. 15:55:58 And "operating systems" as we'd know it are implemented as user-space *libraries* (hence the term "libOS") 15:56:12 Applications are linked (usually dynamically) against their libOS. 15:56:46 so where do drivers fit in? 15:57:31 So, for example, to save a file to disk, even though the exokernel doesn't know how to access a harddrive, it IS aware that a harddrive has an array of disk blocks. The goal of the application is to prove to the libOS that it's use of blocks doesn't interfere with another program's. Once this is proven, the program has free reign over those disk blocks. 15:59:20 So a language environment that prevents messing with other application's memory [and its own memory] is an exokernel? 15:59:25 hey all, did i miss anything? :) 15:59:44 In a loose sense of the term, yes. 15:59:47 qFox: exokernals :) 15:59:49 The reverse is not (of course) true. 16:00:12 exokernal? that sounds like space exploration lingo :p 16:00:48 qFox: An exokernel is an OS kernel that is so minimal that it doesn't even provide *policy* for accessing hardware resources. *ONLY* protection. 16:00:55 kc5tja: what if that language environment didn't control devices at all? 16:01:07 oh i can do that 16:01:09 i did that 16:01:11 :p 16:01:30 but that proved to be quite easy to do 16:01:38 (for me, with the p24) 16:01:47 hm ok nobody cares sorry 16:01:55 warpzero: Then it doesn't. Typically what happens in this case is the exokernel is aware of a series of I/O ports, which are doled out to applications on a first-come-first-served basis, with full permissions. 16:02:06 One and only one application would be bound to a set of I/O ports. 16:02:12 (with full permissions) 16:02:43 And here is where capabilities come in -- if other apps want access (which is sometimes handy), your driver tasks would have the ability to dole out restricted rights to I/O ports too. 16:03:18 qFox: What? 16:03:27 no nvm 16:03:27 :) 16:03:31 qFox: Your P24 emulator never had an OS on it; it wasn't an exokernel. 16:03:48 the bootcode i wrote for it would've worked 16:04:02 and you were able to define new words 16:04:07 what counts as an OS? 16:04:27 kc5tja: lets say the language environment has a running application in it that controls all the devices and their accessibility to the other applications, what then? 16:05:39 --- quit: I440r ("Leaving") 16:05:44 qFox: Any program that lets you control the machine and its actions, especially with respect to other applications. 16:06:13 qFox: However, your system just wasn't an exokernel -- it provided absolutely no protection to resources what-so-ever. It couldn't have; the P24 emulator lacked a memory management unit. :) 16:06:13 --- join: quantis (~darthspud@quantis.demon.co.uk) joined #forth 16:06:28 my god mur is everywhere.... 16:06:37 the code i wrote .... ah hell i dont know, it was a forth environment and you were able to define new words in CODE or threaded 16:06:45 quantis: And I'm not? 16:07:01 warpzero: Well, what then? The exokernel still only provides protection of resources. The application running *under* the exokernel is what provides the policy. 16:07:03 your not im my head.... 16:07:23 wohoo exokernals 16:07:24 qFox: Again, that's not an exokernel. 16:07:32 That's just a language run-time. 16:07:35 kc5tja: well i mean what is it, its not a libOS because its a running application 16:08:13 The criteria for an exokernel are: (1) no policy, and (2) secure multiplexing of hardware resources to deserving applications (defined as applications which can PROVE that their actions won't interfere with another program's resources or behavior, at least directly). 16:08:53 why is the burden of proof on the application? 16:09:13 warpzero: It'd have the behavioral characteristics of a microkernel in that case. But there would still be a libOS attached to it, which manages the interprocess communications with other applications. 16:09:32 warpzero: Because the kernel provides NO policy, except for securely multiplexing resources. 16:09:44 It's also WAY more secure that way. 16:10:09 This is why scientists have to PROVE their statements via repeatable experiments and backing facts when they publish their works. 16:10:28 Otherwise, the Regents of some college can just "vouch" for them, and for all we know, the scientist in question is really a quack. 16:10:31 kc5tja: so If I read you right, if process A and process B both want the parallel prt, the on that asks first gets it and can hold it indefinately? 16:10:50 "prt --> "port" "the on" --> "the one" 16:10:56 Consider Tesla -- a GREAT example of a quack who just *happened* to make a lot of great advancements. 16:11:05 kc5tja: yeah but i mean what if there's nothing to prove -- if an application doesn't have the language to interfere with another applciation? 16:11:25 blockhead: Since it's a single, non-sharable resource by definition, yes. 16:11:48 'k thanks 16:12:03 warpzero: There's always something to prove. 16:12:17 If you have access to a printer port, you've OBVIOUSLY been given some kind of permission to use it from SOMEONE. 16:12:34 Ultimately, that permission comes from the exokernel, in some form or another. 16:12:42 well im talking more about MMU-type stuff 16:12:50 That too. 16:13:22 ANY hardware resource, including but not limited to an application's own page tables, require permission to modify. 16:13:25 well with an MMU how does an application PROVE that a load isn't from another applications space? it cant even ADRESS that space? 16:14:18 Very good. Unless it has been GIVEN a physical page frame ID from another application (through some means not discussed here, because it's unimportant), there is NO WAY for the application to access it. 16:14:34 Otherwise, it can install the page frame ID into its page tables and access it. And the exokernel will HAPPILY let you do this. 16:14:46 okay i see 16:15:05 This is, in fact, precisely how IPC is performed in an exokernel. 16:15:18 If two libOSes aren't aware of each other's IPC mechanisms, too bad -- you can't IPC to them. :D 16:15:43 It's also a great illustration of how capability-based security works. PERFECT example, in fact. 16:15:57