00:00:00 --- log: started forth/04.06.01
02:13:12 --- log: started forth/04.06.01
02:13:12 --- join: clog (nef@bespin.org) joined #forth
02:13:12 --- topic: 'A channel dedicated to the Forth programming language, its implications, its applications, and its philosophy.  :NOTICE:  Kestrel's microprocessor has changed.  For more information, check the weblog and project page for the Kestrel on http://www.falvotech.com .'
02:13:12 --- topic: set by kc5tja on [Mon May 10 07:46:47 2004]
02:13:12 --- names: list (clog warpzero Serg Fractal skylan onetom arke fridge lalalim_ @ChanServ Herkamire madwork Robert @futhin gl mur ianp cmeme)
02:19:09 --- quit: warpzero (Read error: 60 (Operation timed out))
02:32:03 --- quit: madwork (Read error: 60 (Operation timed out))
02:59:26 --- join: warpzero (~warpzero@dsl.142.mt.onewest.net) joined #forth
03:05:22 --- quit: warpzero ("Tried to warn you about Chino and Daddy Gee, but I can't seem to get to you through the U.S. Mail.")
03:30:12 --- join: qFox (C00K13S@cp12172-a.roose1.nb.home.nl) joined #forth
03:30:15 --- join: kuvos (C00K13S@cp12172-a.roose1.nb.home.nl) joined #forth
03:30:25 <qFox> .p
03:30:33 --- part: kuvos left #forth
03:38:50 --- join: ttest (~ttest@212.34.52.140) joined #forth
03:39:09 <ttest> Serg test from telnet
03:39:53 <ttest> i se myself in Mirc ;)
03:40:14 <ttest> backspace
03:40:37 <ttest> ovarstrike
03:42:14 <ttest> bye
03:42:16 --- quit: ttest (Client Quit)
03:49:17 --- join: crc (crc@0-1pool88-55.nas48.philadelphia1.pa.us.da.qwest.net) joined #forth
04:12:13 <qFox> lol :)
04:12:35 <crc> qFox: what's funny?
04:15:16 <qFox> [12:39:06] <ttest> Serg test from telnet
04:15:17 <qFox> [12:39:50] <ttest> i se myself in Mirc ;)
04:15:17 <qFox> [12:40:11] <ttest> backspace
04:15:17 <qFox> [12:40:34] <ttest> ovarstrike
04:15:17 <qFox> [12:42:09] <ttest> bye
04:15:48 <qFox> (like 5 minutes before you joined)
04:18:09 <crc> :-)
04:18:19 * crc always misses the fun things
04:18:25 <qFox> hehe
04:22:51 * crc heads over to meme.b9.com to check the logs :-)
04:23:51 <qFox> i actually never read the logs
04:24:10 <qFox> if there's something i should know, i'm sure someone will tell me, eventually :)
04:25:15 <crc> I always read the logs
04:25:53 <crc> At least, I've read all of the logs since I started hanging out here
04:26:24 <crc> Sometimes I find interesting things in them
04:37:07 <Serg> i was chatting via telnet from home, asking for help on creating Perl irc client ;))
04:37:16 <Serg> actually - filtered telnet 
04:40:00 <crc> cool
04:40:07 <crc> though telnet is insecure
04:40:34 --- join: warpzero (~warpzero@dsl.142.mt.onewest.net) joined #forth
05:00:04 --- join: harm^kuvo (C00K13S@cp12172-a.roose1.nb.home.nl) joined #forth
05:01:35 <Serg> ever wrote something telnet-like, i.e. async comm ?
05:01:35 --- quit: qFox (Read error: 54 (Connection reset by peer))
05:05:03 --- nick: harm^kuvo -> kuvos
05:23:10 --- join: qFox (C00K13S@cp12172-a.roose1.nb.home.nl) joined #forth
05:24:16 <qFox> .nick kuvos
05:24:17 <qFox> .p
05:24:18 --- part: kuvos left #forth
05:24:40 <qFox> hm i should change that connect script
06:25:41 <mur> http://www.costofwar.com/
06:25:43 --- quit: crc (Read error: 104 (Connection reset by peer))
06:31:01 --- join: madwork (~madgarden@derby.metrics.com) joined #forth
06:32:22 --- join: I440r (~FooBlah@168-215-246-243.gen.twtelecom.net) joined #forth
06:32:22 --- mode: ChanServ set +o I440r
06:32:55 --- mode: I440r set +l 83
06:32:57 <I440r> hehe
06:36:06 <mur> why 83?
06:36:35 <I440r> the 83 standard lol
06:36:47 <mur> Serg, i wrote telnet server project manager when in high school
06:36:48 <I440r> limit to the 83 forth standard lol
06:36:52 <I440r> its an anti ans joke :P
06:37:16 <mur> Serg, java one though, but still from scratch
06:37:24 <I440r> it was one of the default channel modes when i created #forth
06:37:39 <mur> :)
06:41:25 <Serg> mur: and how did u solve async problem ?
06:41:46 <Serg> btw, server is not client, i'm unsure does it have the trubble at all
06:42:18 <mur> Serg, i can't remember that now, but i guess just fine
06:47:08 <Serg> i lack the 'non-blocking' mode: if nothing to read, just return 0
08:44:11 --- join: ASau (~asau@158.250.48.204) joined #forth
08:44:18 <ASau> Dobryjj vecher!
08:51:23 <Robert> Privet. :)
09:03:10 <Serg> hi
09:03:24 * Serg is messing arownd w/ M$ crap
09:03:43 <futhin> of course you are
09:03:57 <futhin> *cough*lamer*cough*
09:07:12 <qFox> at least he admits it... ;)
09:07:28 * qFox does too btw.
09:21:45 <Serg> we are tied to some book-keeping soft ;(( windoze only ^((
09:22:29 <ASau> 1C?
09:40:29 <Serg> yeah, 1-ASS !!!
10:18:27 <Serg> dhamn, it failed - had to roll back ;(
10:19:09 --- quit: Serg ()
11:02:17 --- quit: onetom (Read error: 104 (Connection reset by peer))
11:08:16 --- join: MrReach (~spam@209.181.43.190) joined #forth
11:08:16 --- mode: ChanServ set +o MrReach
11:08:49 <MrReach> heya, I440r, got questions about overlays if you want to talk about it
11:41:31 <I440r> sure
11:41:36 <MrReach> heh, not much here
11:41:46 <I440r> ive not implemented them yet but ive been thinking about it
11:42:02 <MrReach> same here, obviously
11:42:08 <I440r> :)
11:42:25 <MrReach> i'm thinking of a webserver written in forth
11:42:42 <MrReach> how would one dynamically change behavior w/o restarting it?
11:43:27 <MrReach> the only thing i've been able to come up with so far is a vector table to dispatch published functions
11:43:46 <I440r> hrm
11:44:19 <I440r> what do you mean by change its behaviour ?
11:44:47 <MrReach> ok ...
11:45:35 <MrReach> let's pretend that http://www.mrreach.net/test/page1.html calls a word that generates a response
11:46:09 * qFox pretends
11:46:28 <MrReach> but I want to change that response ... so I edit sources ... now, how does the server pick up the new behavior without restarting?
11:46:47 <MrReach> it's very similar to your MUD problem
11:47:05 <I440r> well. web content shouldnt extend the server so i would say just make it re-parse the soruces every time a page is loaded
11:47:27 <I440r> you would have multiple instances of the server running
11:47:38 <MrReach> multiple threads, rather?
11:47:44 <I440r> the master would fork and the child would parse the page
11:47:53 <I440r> and extend itself with the specified file
11:48:09 <I440r> the next time the server forked on that page it would re-read the data
11:48:11 <MrReach> hmmm ...
11:48:49 <MrReach> well, constantly reparsing would be a no-no ... they could get rather sizable, but it might reparse if filedate changes
11:50:05 <I440r> is this to run in linux ?
11:50:11 <qFox> flush them on change
11:50:12 <MrReach> i spose I could allcate some memory, aim H at it, compile the file, store the entry point into a variable
11:50:17 <MrReach> yes
11:50:44 <MrReach> qFox: you're saying to check for changes every X seconds?
11:50:59 <MrReach> linux or windows, it really doesn't matter
11:51:14 <I440r> theres a mechanism for a process to be notified that a particular file has been modified 
11:51:20 <I440r> i cant remember where i saw it tho
11:51:26 <I440r> might be a kernel compile option 
11:51:29 <MrReach> ah! yes, seen that
11:51:30 <I440r> i cant remember lol
11:51:45 <MrReach> so does windows
11:51:52 <MrReach> that's ok, i've seen the specs
11:52:16 <I440r> well theres your answer
11:52:18 <MrReach> my concern isn't really with when to refresh object code
11:52:25 <MrReach> but rather how to store/use it
11:52:40 <I440r> well you would memory map it
11:52:55 * MrReach raises his eyebrows
11:52:56 <qFox> MrReach> no, rather have an event that triggers anytime when yo uchange a document thats available by the webserver, or make a check for change every time the document is called (this is a decision that should be made depending on load, and overhead)
11:53:13 <fridge> I'm not sure why I sit in forth, I rarely write any =P
11:53:47 <I440r> lol
11:53:54 <MrReach> qFox: i'd check at request time ... it should be too bad an issue if a user gets a slow respnse once as it's called for the first time
11:53:55 <I440r> fridge because were cool :)
11:54:33 <MrReach> heh, maybe fridge needs a life?
11:54:44 * MrReach ducks
11:55:27 <MrReach> the thing is, I can't just use the fence/empty, forget, or marker interfaces for memory management
11:55:43 <qFox> well it depends on your load (hits), the timeframe in which you want the update to be occur, and speed...
11:55:52 <fridge> well, when I do night shift
11:56:03 <fridge> all the channels with aussie in them are dead
11:56:07 <qFox> if there are many hits, checking for updates on each hit creates a lot of overhead, so a x second/minute timer would be better
11:56:15 <fridge> because they are sane, and asleep
11:56:27 <MrReach> heh, probably true, fridge
11:56:31 <I440r> mrreach empty/forget is complex
11:56:42 <fridge> luckily this channel has attendants from a number of countries
11:56:56 <fridge> so there are interesting conversaations happening at most times of the day 
11:57:01 <I440r> because you have to ddiscard all words in all vocabs (in context or not) that are above the forgotten word
11:57:17 <I440r> so you have to search all threads of all vocabs
11:57:29 <MrReach> yep, they're not the proper tool for the task at hand
11:57:56 <I440r> you definatly dont want to be forgetting things constantly, forget is kinda dangerous
11:58:23 <MrReach> so a given module needs to go into allocated mem, somehow
11:58:26 <I440r> err isnt there a law requiring banks to cache payroll checks ?
11:58:40 <I440r> mrreach a web server HAS to fork for each request!
11:58:42 <MrReach> I don't know
11:58:50 <MrReach> no it doesn't
11:58:54 <I440r> so just have the fork compile the extensions.
11:59:02 <I440r> you wont be able to handle more than one connection then
11:59:18 <MrReach> sure you can, in fact in more than one way
11:59:33 <MrReach> method 1: event driven sockets
11:59:50 <MrReach> method 2: hand off request to a thread
12:01:46 <Herkamire> A web server need not have multiple threads/instances
12:02:05 <Herkamire> it just takes some kind of multitasking to handle multiple connections simultaniously
12:02:19 <MrReach> yep, that's event-driven 
12:02:26 <Herkamire> cooperative multitasking would work great
12:02:34 <MrReach> well, yes and no
12:03:11 <Herkamire> Here's how I imagine it:
12:03:12 <MrReach> it actually turns out to be quite efficient ... but if a request has a lock-up for some reason, the whole server dies brutally
12:03:40 <MrReach> TCL uses that method, actually, so I'm quite familiar with it
12:04:42 <Herkamire> the core, manages connections, and reads and writes to them. when it recieves a connection, it openes it and reads the request into memory. when it has the full request (or maybe a significant portion) it calls a word to handle it. This word can read the request, and writes a response (in memory). (this is fast.) then returns to the core
12:05:13 <Herkamire> the core continues looping through the open connections, reading available data, and sending responses when they are available
12:05:31 * MrReach nods
12:05:43 <MrReach> what happens if one of the handler words locks up?
12:05:49 <Herkamire> bad things
12:06:31 <I440r> lol
12:06:53 <MrReach> on the other hand, if one uses threads or fork, a crashed word poses a rather large memory leak
12:07:21 <MrReach> at least it doesn't quit altgether, though
12:08:22 <Herkamire> it does make things more complicated to protect against crappy code
12:08:55 <Herkamire> one solution boing to give it it's own protected memory space.
12:09:07 <MrReach> well, even if the code is excellent, problems can still happen
12:09:30 <MrReach> a db request might take 30 secs to fulfil, for example
12:09:34 <Herkamire> another being use an interpreted/emulated language that cannot crash, or exceed it's memory bounds, and can stopped after a given execution time
12:10:25 <MrReach> trying to think of something else that might go wrong and be noone's fault
12:10:56 <MrReach> filesystem might get full on a message posting page
12:11:02 <Herkamire> if a db request takes 30 secconds (under GNU/Linux anyway) the whole machine will grind to an almost halt.
12:11:04 <MrReach> or during a file upload
12:11:16 <MrReach> yep, Herkamire
12:11:43 <MrReach> even excellent code can't foresee every possible thing that might go wrong
12:11:56 <Herkamire> sure it can
12:12:03 <Herkamire> asside from hardware failure
12:12:16 <MrReach> add to that a hostile environment where people "out there" are deliberately trying to crash the software
12:12:25 <Herkamire> I don't see any problem with writing software that knows it's limits.
12:12:45 <Herkamire> it knows how much ram it has, and how much HD space, and execution time. It then does what it can with those.
12:13:43 <MrReach> i usually work under the assumption "this code will crash eventually"
12:14:06 <MrReach> and try to plan for how to clean up the mess and go on
12:14:11 <Herkamire> UNIX programming encourages people to pretend that there is as much RAM/disk/whatever available as you want, and you should just ask for how much you want.
12:14:23 <MrReach> yep
12:15:05 <Herkamire> I think that's stupid
12:15:46 <MrReach> it's not generally the way I think
12:16:11 <MrReach> but I no longer try to conserve every byte like when I was writing for a C64
12:16:26 <Herkamire> (to make sure I'm clear, when I said 'stupid' I was refering to what I was talking about)
12:16:36 <MrReach> right, I understood
12:17:06 <Herkamire> I don't conserve every byte either
12:17:20 <Herkamire> I have not made a dynamic memory allocator for herkforth, and I'm very reluctant to do so
12:17:22 <MrReach> i've got lots of faults, but criminal stupidity isn't one of them (unlike Pres. Bush)
12:17:48 <Herkamire> <nod>
12:17:57 <MrReach> in any case ...
12:18:13 <MrReach> the source text eventually ends up as a memory image
12:18:28 <MrReach> a binary memory image
12:18:50 <MrReach> which will get called at the appropriate time
12:19:07 <MrReach> but how does one change that image at runtime?
12:19:18 <Herkamire> huh? what source text?
12:19:47 <MrReach> Herkamire: you must have missed the axioms for this convo
12:20:07 <MrReach> I was talking about a web server that was under development
12:20:08 <Herkamire> I did not read much backlog before I joined.
12:20:28 <MrReach> and how to change behavior at runtime without rebooting the entire server
12:20:44 <MrReach> (incremental development)
12:21:39 <Herkamire> I see. a particular web server?
12:21:47 <MrReach> no
12:21:51 <Herkamire> ok cool
12:21:59 <MrReach> one i'm considering writing in forth
12:22:06 <Herkamire> very cool
12:22:20 <Herkamire> changing it while it runs is an interesting problem
12:22:41 <MrReach> i've been using one written in TCL ... which is an interpretive language, and I *REALLY* like the live editing
12:23:17 <MrReach> but it's not all that effient, or stable
12:23:28 <Herkamire> you want to be able to change the core? or just the request handlers?
12:23:50 <MrReach> handlers, mostly, i'd expect a reboot when changing the core
12:24:21 <MrReach> although everything past reading config files could be considered "handlers"
12:24:49 <Herkamire> in my mind the core includes managing connections and multitasking
12:24:54 <MrReach> "We got a new connection ... Who ya' gonna call?!"
12:25:07 <Herkamire> Ghost Busters!
12:26:03 --- join: tathi (~josh@pcp02123722pcs.milfrd01.pa.comcast.net) joined #forth
12:27:39 <MrReach> I really like the idea of allocated image blocks, although it would be a nightmare to implement
12:27:59 <MrReach> it would fit *SO* elegantly with threading
12:31:02 <MrReach> not so elegantly with fork(), but fork() has large overhead anyway
12:32:39 <Herkamire> with herkforth I could just change the source, and call the bootstrapping process
12:32:51 <Herkamire> it would recompile it's self, and call the new core
12:33:11 <MrReach> would that be in a new process??
12:33:12 <Herkamire> this should work fine so long as the connections are held as persistant data
12:33:22 <Herkamire> everything that's still in the sources
12:34:33 --- join: slava (~slava@CPE00096ba44261-CM000e5cdfda14.cpe.net.cable.rogers.com) joined #forth
12:34:35 <slava> howdy
12:34:45 <MrReach> hiya, slava
12:35:27 <MrReach> I440r: how hard would the binary image blocks be to do in IsForth??
12:35:47 <Herkamire> MrReach: what do you mean "binary image blocks"?
12:35:55 --- join: madwork_ (~madgarden@derby.metrics.com) joined #forth
12:36:07 <MrReach> something like ...
12:36:40 --- part: madwork_ left #forth
12:36:48 <I440r> hmmm well. if you allocate a memory map and specify its address not that difficult
12:36:54 <MrReach> allocating some memory ... pointing H at it, loading sources, saving entry points into a variable, reseting H back to the main dictionary
12:37:05 <I440r> compiled isforth code isnt exactly relocatable
12:37:21 <I440r> yes. but the buffer would have to be at a specific address.
12:37:24 <MrReach> right, it has to be compiled where it's going to live
12:37:30 <I440r> yes
12:37:48 <MrReach> heh, reloading it pretty easy ...
12:37:52 <I440r> thats how i was planning on doing overlays
12:38:07 <I440r> just memory map an overlay buffer at a specific address, divide it into overlay list and overlay head
12:38:39 <MrReach> allocate a new block, do the same process, reset the variable that holds the entry point ... FREE the previous block (after waiting for stuff that might be using it)
12:38:52 <I440r> you would need an overlay vocabulary and IT would need to be saved with the given overlay
12:39:04 <I440r> it would also restrict you to ONE active overlay at a time
12:39:33 <I440r> dont need to free - you would do an anonymous mapping at the required address and read the file into that buffer
12:39:42 <I440r> you can unload the file without unmapping the buffer
12:39:49 <MrReach> right, but what about when you need to recompile?
12:40:08 <MrReach> (source change)
12:40:24 <I440r> well you would unload the buffer, reset the head and list pointers and erase all threads in the overlay vocabulary
12:40:42 <I440r> when the overlay is compiled you would save it out with some sort of file format
12:40:50 <I440r> thers more than just list and head space to save
12:41:12 <I440r> like i said, you would have to also save the vocabulary itself (i.e. all the threads in the vocab array)
12:41:36 <I440r> the vocabulary would be in main program space - 
12:41:45 <MrReach> erm ... i'm wondering if you envisioning the same thing I am
12:41:55 <I440r> you wouldnt be allowed to do 'overlay definitions"
12:42:12 <I440r> because that wouldnt adjust head and list pointers for the overlay
12:42:23 <MrReach> actually, how it's done isn't important ... only that the server picks up source changes automatically
12:42:37 <MrReach> and that a lockup doesn't crash the whole contraption
12:42:39 <I440r> not sure how that would be accomplished
12:43:09 <MrReach> ok, what segments does isForth have by default?
12:43:22 <MrReach> there's 3 I can think of
12:43:27 --- quit: madwork (Read error: 60 (Operation timed out))
12:43:33 <MrReach> code, list, and dictionary
12:43:47 <MrReach> are there others? oh! user variables
12:43:52 <MrReach> that's 4
12:43:59 <I440r> isforth does not have user variables
12:44:26 <MrReach> no, but it needs a place to store info, and that prob seperate from the other three mem segs
12:44:35 <I440r> define info
12:44:49 <MrReach> VARIABLE blah  blah OFF
12:45:15 <I440r> well the variable body holds its value
12:45:26 <I440r> the variable would be in overlay list space
12:45:29 <MrReach> so it's in list segment?
12:45:33 <I440r> yes
12:45:37 <MrReach> ok
12:45:49 <I440r> the x86 is NOT a freekin harvard architecture
12:45:51 <I440r> dammit :P
12:46:02 <I440r> pmode tries to turn it into one tho
12:46:13 <I440r> isforth unmakes that restriction
12:46:13 <MrReach> well, yes and no
12:46:27 <I440r> i sysmprotect all memory to +rwx
12:46:30 <MrReach> it *CAN* be harvard arch, but nobody writes OS's like that
12:46:37 <I440r> linux
12:46:53 <I440r> you cannot execute out of a memory map defined as DATA
12:47:05 <I440r> i.e. you cannot normally execute writeable memory
12:47:10 <MrReach> sure you can
12:47:17 <I440r> not unless you mprotect it
12:47:23 <I440r> for instance
12:47:25 <MrReach> ok
12:47:26 <I440r> foo:
12:47:31 <I440r>  mov [foo], 0
12:47:41 <I440r> assemble that to an elf and gdb it
12:47:44 <I440r> it will segfault
12:47:49 <I440r> in linux
12:48:10 <MrReach> yep
12:48:14 <I440r> because the code is in a code segment (.text) and thats not normally writeable
12:48:43 <I440r> when isforth runs it breaks its memory out to 1 meg and then mprotects it to +rwx
12:48:47 <MrReach> but you can write code that "MOV EAX , [ some number from variable ] ; JMP EAX
12:49:09 <I440r> only because on an x86 executable code HAS to be readable
12:49:14 <I440r> not so on other processors
12:49:29 <MrReach> heh, not writing for other processor
12:49:29 <I440r> anything thats +x on an x86 is automatically +r
12:49:33 <I440r> and vice versa
12:49:46 <slava> so how does openbsd implement r^x then?
12:49:48 <I440r> me either :)
12:50:03 <I440r> no idea
12:50:13 <I440r> chances are it emulats it on an x86
12:50:13 <MrReach> how hard would it be ...
12:50:40 <MrReach> to compile into a seperate memory segment(s) that can be discarded/replaced later?
12:50:48 <I440r> you physically cannot prevent reading of a block thats defined as executable on an x86
12:50:56 <I440r> easy
12:50:58 <MrReach> into allocated memory, rather
12:51:12 <I440r> look at how i do headerless words
12:51:21 <I440r> <headers switches into headerless mode
12:51:30 <I440r> headers> switches back into headerless mode
12:51:45 <I440r> i think of them as arrows pointing to where the headerfull words are
12:51:56 <MrReach> the only issue I can see is where to store variables in truely threaded applications
12:52:12 <I440r> you could have ovelay> and <overlay
12:52:30 <I440r> THAT would need user variables
12:52:42 <I440r> would be almost impossible, would be better to use fork :)
12:52:47 <I440r> i.e. copy on write
12:52:55 <MrReach> naw, don't say impossible
12:53:09 <I440r> i said "almost" heh
12:53:20 <MrReach> it would need some fundamental changes in isForth, though
12:53:32 <I440r> such as ??? :)
12:54:02 <MrReach> a pointer to the user variable area, and words like @ and ! would need to use it
12:54:22 <I440r> how about a "user" extension and u@ u!
12:54:32 <I440r> uw@ uw! ud@ ud!
12:54:47 <I440r> erm - would probably need ucmove> etc too
12:54:55 <MrReach> and never touch a variable from within anything but the main thread?
12:55:00 <I440r> i.e. a complete copy of every word in memory.1 but for user space
12:55:14 <I440r> yes
12:55:22 <I440r> but thats anti forth
12:55:34 <I440r> thats making a distinction between system and user
12:55:42 * MrReach nods
12:55:48 <I440r> there is no user. there is no system, there is NO SPOON! ther is only forth!
12:55:55 <I440r> heh
12:56:07 <I440r> but i can see the need for the spoon in this case :/
12:56:08 <slava> haha
12:56:15 <MrReach> each thread should assume it has all the variables, with deliberate action needed to talk among threads
12:56:38 <MrReach> rather than the other way around
12:56:48 <I440r> why threads ?
12:56:51 <I440r> why not fork ?
12:57:09 <I440r> that scenario lends itself to fork 
12:57:14 <MrReach> 1. processes can't talk to each other as easily as threads
12:57:29 <I440r> correct. they would need pipes and those are slow
12:57:30 <MrReach> 2. threads generally don't share allocated memory
12:57:30 <I440r> OR!!!!
12:57:32 <I440r> shared memory
12:57:39 <MrReach> oops, typo ...
12:57:46 <MrReach> 2. processes generally don't share allocated memory
12:57:47 <I440r> they can in linux i believe
12:57:51 <I440r> err might be wrong about that
12:57:56 <slava> sysv shm!
12:57:59 <slava> its really simple
12:58:04 <I440r> yup
12:58:14 <I440r> cept thats deprecated now isnt it ?
12:58:18 <slava> no
12:58:22 <slava> a lot of programs use it
12:58:22 <I440r> lol so many new fangled ways of doing things
12:58:25 <MrReach> 3. task switch among processes has 10x the overhead of thread switching
12:58:26 <I440r> i ferget :)
12:58:28 <slava> its been around for a long time
12:58:37 <slava> since SVR4 at least
12:58:45 <MrReach> yeah, SysV dates to the 70s
12:58:45 <I440r> mrreach not if isforth is the one doing the switching :)
12:58:52 <I440r> no need to jump into kernel space at all
12:59:14 <MrReach> if there are seperate processes, the kernel needs to switch among them
12:59:39 <MrReach> but you're back to cooperative MT in forth again
12:59:44 <I440r> maybe not
13:00:02 <I440r> just need to implement timers and signals in isforth 
13:00:06 <I440r> which i havent done yet
13:00:15 <I440r> could be preemptive
13:00:31 <I440r> hrm maybe not
13:00:40 <I440r> that would still require some kernel interaction
13:00:51 <MrReach> hey, would you have a conniption is someone asked you to move variables into a single variable block that could be easily cloned?
13:01:00 <I440r> and would mean your still doing user/kernel space switching
13:01:11 <I440r> lol
13:01:37 <I440r> tell you what. you implement it and ill look it over and decide then :)
13:01:45 <MrReach> ok
13:01:52 <MrReach> there will be SOME overhead
13:01:53 <I440r> how about having a variable base register
13:02:18 <MrReach> that's what I'm thinking, but it will reduce the number of available registers by one
13:02:34 <I440r> err no i dont mean a cpu register
13:02:38 <I440r> there arent any free
13:02:43 <MrReach> but allow for EXTREMELY fast thread switching
13:02:45 <I440r> erm well edi is
13:03:00 <I440r> but some words use edi for scratch
13:03:03 <I440r> i.e. its not saved
13:03:21 <MrReach> right, I understand I'd have to hunt down all words that used a particular register
13:03:22 <I440r> it would make the kernel a little more complex
13:03:52 <I440r> plus you would have to document the whole adventure :)
13:04:07 <MrReach> you know, this would allow the difference between kernel thread changes and forth thread changes to become almost invisible
13:04:19 <I440r> i was going to do some more documenting last night, got drunk instead lol
13:04:35 <MrReach> ah! wish I'd know ... I'd have joined you
13:04:46 <I440r> which is why im not having the aforementioned conniption over the added complexity 
13:04:51 <I440r> lol
13:05:17 <MrReach> heh, I have probs predicting which ideas you're gonna like and which you're not
13:05:29 <I440r> well never be too scared to make suggestions
13:05:36 <I440r> all i can do is bite your head off :)
13:05:50 <MrReach> i'm not generally, but always prepared to get my ass chewed
13:05:56 <I440r> lol
13:06:04 <I440r> wear padding
13:06:07 --- join: wossname (wossname@HSE-MTL-ppp66560.qc.sympatico.ca) joined #forth
13:06:28 <I440r> this shouldnt add TOO much complexity if done right
13:06:30 <MrReach> now i'm wondering ...
13:06:45 <MrReach> is this change really needed for what I want to accomplish?
13:06:46 <I440r> and there are some places in the kernel that are quite complex
13:07:48 <MrReach> yes, i'd get to deal with I440r-isms
13:07:57 <I440r> such as ???
13:07:58 <I440r> lol
13:08:02 <I440r> my ism's are GOOD ones :P
13:08:12 <MrReach> the mix of variables and constants
13:08:25 <I440r> dont forget var and const too
13:08:27 <MrReach> they're ok, but often unexpected
13:08:35 <I440r> erm i should tell you my plan for const however
13:08:39 <I440r> and THIS is very close to it
13:08:47 <MrReach> (reading over sources ... "WTF?")
13:09:04 <I440r> i want all vocabularies compiled into their own spaces... i.e. one here/hhere per vocaulary
13:09:05 * fridge ponders
13:09:20 <I440r> and i want to be able to discard an entire vocaulary on demand
13:09:33 <MrReach> oh, good, that would make my dynamic dictionary easy
13:09:45 <I440r> for instance, when you turnkey there isnt a single definition in the compiler vocab that has any further use
13:09:49 <I440r> i.e. its ALL deadwood
13:09:57 <MrReach> yep
13:10:04 <I440r> i just havent figured out how to implement this
13:10:05 <MrReach> zimmer fought with that, too
13:10:11 <I440r> he did ?
13:10:33 <MrReach> yep, Win32Forth has system/application
13:10:34 <I440r> i think i need to be metacompiling before i can do this
13:10:50 <MrReach> TURNKEY pitches everything in system
13:10:59 <I440r> oh... i wasnt finished heh - the compiler vocab would be marked as transient (some flag in the vocab array)
13:11:14 <I440r> when you fsave it remains but turnkey discards transient vocs
13:11:24 * MrReach nods
13:11:31 <I440r> ALL consts would be compiled into the compiler vocab no matter what voc is currently current
13:11:31 <MrReach> all vocabs need that flag
13:11:41 <MrReach> some get kept, others don't
13:11:53 <I440r> no const would need to be kept accross a turnkey
13:12:01 <I440r> a const is a state smart immediate constant
13:12:11 <MrReach> huh??
13:12:15 <I440r> it compiles a literal if in compile mode
13:12:17 <I440r> 0 const foo
13:12:20 <I440r> foo . 0 ok
13:12:21 <MrReach> oh! ok
13:12:27 <I440r> : blah ... foo .... ;
13:12:34 <I440r> foo there compiels as a literal 0
13:12:47 <I440r> the definition for foo is deadwood on turnkey!!!
13:13:19 <I440r> you would be able to define 28467529378465927384659 things and the ones you dont use dont take up any space in your turnkey
13:13:30 <I440r> i.e. take all those c .h files and create forth versions of them
13:13:52 <I440r> the .h file would define 2985462978467849 consts and you would use ONE of them
13:14:20 <futhin> i love you mrreach, everytime you're here, there's a lot of forth talk :D
13:14:22 <MrReach> you've redicovered that a constant is not a variable is not a value
13:14:47 <I440r> theyi have variable, var, const and constant
13:14:53 <futhin> #forth usually has too little forth talk
13:14:57 <MrReach> erm ... thanks, futhin ... I think
13:15:02 <I440r> i discourage variable and constant
13:15:21 <I440r> var and const are better. i think they work the way variable and constant should have from day 1
13:16:08 <fridge> do any linux forth's have words for putting graphics on a framebuffer?
13:16:08 <MrReach> ok, ESI is the return stack pointer?
13:16:16 <I440r> the name "var" is MUCH better than "value" <-- someone had a gay moment when he named that
13:16:26 <I440r> isforth will... eventually
13:16:31 <I440r> no
13:16:33 <I440r> esi is IP
13:16:35 <MrReach> I440r: they were trying to avoid breaking existing code
13:16:37 <I440r> EBP is RP
13:16:42 * slava is about to have a gay moment
13:16:44 * slava grabs I440r's leg
13:16:54 <I440r> i dont object to the method
13:16:56 <I440r> just the name
13:17:03 <slava> fridge, that would be cool
13:17:09 <I440r> (camp voice and limp wrist) "value"
13:17:16 <slava> fridge, none that i know of but it wouldn't be too hard to implement i guess
13:18:08 <I440r> mrreach do var and const would need to be in variable space too
13:18:11 <MrReach> heh
13:18:13 <I440r> and thats a problem
13:18:23 <I440r> because this means that all vocabs will need their own variable space
13:18:43 <I440r> actually const wont need to be
13:18:55 <I440r> my head hurts :P
13:19:00 <MrReach> right, was just about to say that
13:19:08 <MrReach> no reason that lits need to change
13:19:40 <MrReach> anything that writes to or points at list space needs to be changed
13:19:50 <MrReach> except the compiler, of course
13:19:58 <I440r> right
13:20:11 <I440r> ooh this will break the decompiler too :)
13:20:25 <MrReach> shouldn't massively break it
13:20:28 <I440r> no
13:20:34 <MrReach> if it uses @ and !
13:20:43 <I440r> i dont think it will massivly break anything
13:20:54 <I440r> but the decompiler would need to be made user space aware
13:21:26 <I440r> well actually, it wont need to be if i cansel one LITLE plan :)
13:21:38 <I440r> i was planning on listing values when i "see" a variable or constant
13:21:47 <I440r> or a deferred word
13:21:56 <MrReach> no reason that shouldn't work
13:22:00 <I440r> oh! user deferred words ?
13:22:07 <I440r> will deferred words need to be userified ?
13:22:13 <I440r> i think so
13:22:16 <MrReach> not really
13:22:27 <MrReach> oh! yes
13:22:28 <I440r> y not ?
13:22:35 <I440r> eek
13:22:38 <MrReach> they are like regular variables, right?
13:22:45 <I440r> they are like constants
13:22:47 <I440r> that execute
13:22:58 <I440r> a_constant:
13:23:02 <I440r>  call doconstant
13:23:04 <I440r>  dd value
13:23:10 <I440r> a_deferred:
13:23:11 <MrReach> well, that depends if the compiler is the only thing allowed to change them
13:23:12 <I440r>  call dodefer
13:23:18 <I440r>  dd whereto
13:23:30 <I440r> ' blah is foo
13:23:36 <I440r> ANYONE can change them
13:23:48 <I440r> turnkey relies on this
13:23:57 <I440r> ' my-main is quit (or default)
13:24:00 <I440r> turnkey my-app
13:24:16 <I440r> i would object to a drastic modification of isforth into user/system
13:24:21 <MrReach> better move the dd into user space ... then dd on offest into the list space
13:24:47 <I440r> doubly indiret it
13:25:14 <I440r> ok, im confuzed again
13:25:34 <I440r> were talking about a modification that allows a forth tasker
13:25:44 <I440r> i.e. two instances of the same variable for two different tasks
13:25:46 <MrReach> dodefer might look like "mov eax, [esi]; add esi # 4;jmp eax"
13:26:09 <MrReach> and it might change to ...
13:26:24 <I440r> you need to look at the code for dodefer
13:26:27 <I440r> its in exec.1
13:26:29 <MrReach> dodefer might look like "mov eax, [esi]; mov eax, [eax];add esi # 4;jmp eax"
13:26:43 <I440r> it would be pop eax
13:26:47 <I440r> mov eax, [eax]
13:26:48 <I440r> jmp eax
13:26:53 <MrReach> yeah, a detail thanks for mentioning it ...
13:27:41 <MrReach> different threads might defer to different targets ... ergo, there needs to be a place in user space to store the adr
13:27:43 <I440r> cant do this without fork
13:28:12 <I440r> how can you store multiple addresses in the same variable/constant/var/const unless your forking
13:28:15 <MrReach> right, the kernel API uses fork to start threads
13:28:23 <I440r> no it doesnt
13:28:28 <I440r> it doesnt use fork
13:28:30 <I440r> it uses clone
13:28:38 <MrReach> no, a new thread copies (or copy on writes) the user variables
13:28:40 <I440r> and theres no copy on write for a clone
13:28:48 <I440r> threads dont
13:28:53 <I440r> only a fork does copy on write
13:29:01 <MrReach> ok
13:29:22 <I440r> which - if we need to fork totally negates the need for user variables :/
13:29:32 <MrReach> part of the overhead of starting a new thread would be allocating a new user variable space, and copying the old one into the new one
13:30:05 <MrReach> which can still be fairly effient, 'cause it can be moved as a single block
13:30:14 <I440r> the variable will contain a different value for each task. it will either be a different indirect target address for the value of the variable or simply the value of the variable for each task
13:30:24 <I440r> ok
13:30:32 <I440r> that makes selse
13:30:52 <I440r> erm. no. that would be one variable space per vocabulary
13:30:55 <I440r> cant get away from taht
13:31:02 <I440r> because of my plans for transient vocabs
13:31:09 <MrReach> no, one user variable block per thread
13:31:15 <I440r> nope
13:31:22 <MrReach> only once thread can compile
13:31:26 <I440r> one user variable block per thread per vocab
13:31:29 <MrReach> otherwise it's total chaos
13:31:37 <I440r> ok
13:31:46 <I440r> how do you block compiles in a different thread
13:32:07 <I440r> and how do you make variables which are compiled into a transient vocabulary disappear
13:32:09 <MrReach> you don't
13:32:17 <MrReach> forth has never provided safety nets
13:32:23 <I440r> correct
13:32:41 <MrReach> compile code, spawn thread that uses it
13:32:42 <I440r> i can accept users shooting themselves in the foot/head :)
13:33:08 <MrReach> do NOT spawn thread that compiles
13:33:24 <MrReach> or ALLOTs or commas
13:33:35 <I440r> well... you can spawn a thread that asks the master thread to compile something :)
13:33:49 <MrReach> if it needs memory, it has to ALLOCATE it and store the pointer into a previously existing VARIABLE
13:34:15 <I440r> and what if some other thread is already using that variable ?
13:34:19 <I440r> oh it wont matter
13:34:26 <MrReach> right, private copy
13:34:27 <I440r> this thread will have its own variable space
13:34:39 <I440r> erm variable/constant/const/deferred space
13:34:48 <MrReach> no, can't have master compile for you, either
13:34:49 <I440r> ALL of those need to be user-ified
13:35:00 <MrReach> because dict variables won't be updated
13:35:04 <I440r> which brings us back to the "one per thread per vocabulary"
13:35:25 <I440r> define a deferred word in the foo vocab
13:35:39 <I440r> it MUST be in the foo vocab
13:35:51 <I440r> not in some funky magic overlay vocab
13:36:09 <MrReach> right, but the address is stored in variable space
13:36:12 <I440r> you have to be able to HIDE that item by removing its parent vocab from contex
13:36:20 <I440r> you dont understand
13:36:24 <MrReach> ok
13:36:31 <I440r> every vocabulary will HAVE to have its own user space
13:36:52 <MrReach> ok, the part I don't understand is "why"
13:36:55 <I440r> if i define a variable in the foo vocabulary it will have to go in the foo vocabs user space
13:37:04 <I440r> foo previous
13:37:08 <I440r> foo is no longer in context
13:37:13 <I440r> so neither should any of its variables
13:37:37 <I440r> otherwise you are making it impossible to make multiple words of the same name
13:37:42 <I440r> simply because they are of type variable
13:38:00 <MrReach> ok
13:38:10 <I440r> and hers another problem
13:38:20 <I440r> this extension will need to be part of the kernel 
13:38:32 <I440r> and it will need to have the memory manager available to it
13:38:35 <MrReach> deleting the vocab might leave holes in variable space ... is this unacceptable?
13:38:38 <I440r> i.e. isforth memory manager
13:38:59 <I440r> it wont because deleting the vocab will also delete that vocabs variable space
13:39:18 <I440r> the kernel will be relying on an extension 
13:39:25 <I440r> i cant see a way out of that one either
13:39:47 <MrReach> we COULD have multiple variable areas, but it would greatly complicate matters, because the value in the cpu register is no longer a constant
13:40:36 <I440r> you would need one pointer per voc
13:40:46 <I440r> err
13:40:49 <I440r> foo @ 
13:41:03 <I440r> search context for foo
13:41:06 <I440r> find foo
13:41:11 <I440r> see if its a variable
13:41:27 <I440r> if so read variabe base register for this vocab
13:41:27 <MrReach> that could turn into insane overhead
13:41:32 <I440r> then execute variable
13:41:36 <I440r> way more complex
13:41:44 <I440r> ya
13:41:56 <I440r> !!!!!!!!
13:41:58 <I440r> lol
13:42:35 <I440r> thers a solution but it will restrict you to a specific number of defineable vocabs
13:42:45 <I440r> just create an array of "max-vocabs" dwords
13:42:49 <MrReach> ok, a variable allocates a word in user space
13:43:14 <MrReach> all variables in all vocabs allocate in same region 
13:43:22 <I440r> when find searches a given vocab it loads that vocabs user space into edi
13:43:31 <MrReach> if a vocab is deleted, those vars are no longer accessible
13:43:37 <I440r> weather were searching for a variable or not
13:43:48 <I440r> the overhead isnt too big but its a restriction to N vocabs
13:43:51 <MrReach> that would be the simplest method
13:44:14 <I440r> wastefull of space
13:44:22 <I440r> but acceptable
13:44:28 <I440r> maybe
13:44:38 * MrReach scratches his head
13:44:44 <I440r> yea lol
13:44:49 <I440r> which is better
13:44:57 <I440r> restricting you to N vocabularies
13:44:59 <I440r> or wasting space
13:45:14 <MrReach> runtime overhead is to be avoided, no matter the effort needed
13:45:31 <I440r> the overhead for one user space per vocab is small
13:45:43 <I440r> i solved that already. but it adds the above restriction
13:45:51 <I440r> add an entry to each vocab
13:45:52 <MrReach> yeah
13:46:08 <I440r> when find searches a given vocab simply have IT load that vocs user space base pointer into edi
13:46:25 <I440r> edi will always be correct when the variable/const/var/constant/deferred executes
13:46:47 <I440r> but the array has to be pre-allocated
13:46:57 <MrReach> not really
13:47:01 <I440r> and a forget of a given vocabulary will need to bubble the array
13:47:06 <I440r> no ?
13:47:08 <MrReach> each vocab could have a base pointer
13:47:18 <MrReach> as part of the vocab def
13:47:24 <MrReach> i think
13:47:35 <MrReach> arg! complicated piece of shit!!!
13:48:41 <I440r> lol
13:49:57 <I440r> the vocabs will eventually need this when they are in their own space
13:50:07 <MrReach> C progs never have this problem because they don't suddenly drop large chunks of their object code
13:50:08 <I440r> but that will definatly not be implemented till im metacompiling
13:50:26 <I440r> unless you can show me how to get nasm to assemble into multiple .text and .data regeons
13:50:31 <I440r> i.e. NOT concat them
13:50:38 <I440r> lol
13:50:45 <I440r> how about forget forget :P
13:50:51 <MrReach> have you started on the metacomplier?
13:50:55 <I440r> no
13:50:59 <I440r> that will need the assembler
13:51:11 <I440r> when i have the assembler i will be able to do the metacompiler
13:51:14 <MrReach> what level??
13:51:20 <I440r> ?
13:51:29 <MrReach> i386? pentium?
13:51:45 <I440r> 386 to p6 including all fpu, mmx, see and sse2 is what i want
13:51:53 <I440r> but thers another absolute requirement :(
13:52:02 <MrReach> just grab the i486 assembler out of Win32Forth
13:52:18 <I440r> i will not accept any assembler that requires anything other than SLIGHT modifications to the existing sources
13:52:20 <I440r> err no.
13:52:21 <I440r> i cant
13:52:26 <I440r> it will taint my lisence
13:52:44 <Robert> :D
13:52:47 <I440r> its PURE GPL
13:53:00 <I440r> if i extend then i will have to be PURE GPL
13:53:07 <I440r> that is absolutely NOT acceptable to me
13:53:12 <MrReach> right
13:53:24 <I440r> so the only solution is for me to write the assembler myself
13:53:27 <I440r> oh. 
13:53:36 <I440r> theres anotehr thing i would like
13:53:45 <MrReach> ???
13:53:47 <I440r> all opcodes generated in OCTAL
13:53:57 <MrReach> huh?
13:54:01 <I440r> hex opcodes on an x86 is utter stupidity
13:54:03 <MrReach> in the disassembler?
13:54:06 <I440r> no
13:54:07 <I440r> in the assembler
13:54:25 <I440r> intel opcodes are encoded in octal NOT hex
13:54:28 <MrReach> whatever ... BASE works excellently for that
13:54:42 <I440r> intel howeve ronly publshes them in hex and totally fucking obfuscates the encodings
13:54:46 <MrReach> it's all bit-stuffing, in any case
13:54:47 <I440r> you dont understand
13:54:56 <I440r> hang on
13:59:55 <I440r> http://www.sunbeem.net/quickforward/forward/qf/src/docs/opcode.html
13:59:58 <I440r> read that
14:00:05 <I440r> i want to base my assembler on THAT info
14:00:20 <I440r> i kinda already started but aborted work till i did the docs
14:01:11 <MrReach> ok
14:01:51 <I440r> did i ever show you my 8051 assembler ?
14:03:44 <MrReach> i've seen it, don't remember it
14:03:57 <I440r> i dont think you will have, its unreleased
14:04:11 <I440r> the disassembler is on the web page and in the bz2 but this isnt released yet
14:04:37 <I440r> i did it as a proof of concept, i want my x86 assembler to work in a similar fassion
14:04:42 <I440r> got an email ?
14:04:43 <I440r> message it
14:04:53 <I440r> so spam bots dont mine your addy :)
14:06:20 <MrReach> my email is totally messed up right now while setting up domain
14:06:27 <I440r> doh
14:06:35 <I440r> not sure if i can dcc or if you can accept lol
14:08:04 <I440r> one or other is failing
14:08:26 <MrReach> try again, please
14:09:30 <I440r> see it ?
14:09:45 <MrReach> yep, and accepted
14:09:52 <I440r> not moving
14:09:58 <MrReach> firewalls are stoping it
14:09:59 <I440r> doh
14:10:18 <MrReach> ???
14:10:29 <I440r> see if can send in icq
14:11:49 <MrReach> ok, nevermind then
14:12:02 <MrReach> later, put it on your webserver
14:12:27 <I440r> cant from here and have no connection in motel :/
14:12:37 <MrReach> heh
14:12:59 <MrReach> ok, i've gotta go for now
14:13:08 <MrReach> will try to get back this evening
14:13:14 <I440r> k
14:13:19 <I440r> try implement that stuff:)
14:14:03 * MrReach norts through his nose
14:14:07 --- part: MrReach left #forth
14:19:47 --- quit: warpzero ("Tried to warn you about Chino and Daddy Gee, but I can't seem to get to you through the U.S. Mail.")
14:24:12 --- join: lalalim (~lalalim@pD95EA31E.dip.t-dialin.net) joined #forth
14:28:19 --- quit: lalalim_ (Read error: 60 (Operation timed out))
14:57:26 <slava> i'm coding complex trigonometric and hyperbolic functions and going nuts w/ managing the stack
14:59:33 <I440r> lol
14:59:35 <I440r> y ?
14:59:45 <I440r> you using floats ?
15:00:27 <slava> just for the hell of it
15:00:33 <I440r> ick 
15:00:35 <I440r> lol
15:00:35 <slava> an excercise, if you will
15:01:15 <slava> also to insipire me to do a benchmark and a) see that its incredibly slow b) optimize my compiler for numeric code
15:01:26 <I440r> hjeh
15:02:50 --- quit: wossname ("woah")
15:03:52 <slava> its real ugly, look at this factor for instance
15:03:53 <slava> : real-cosh-sinh ( x -- cosh sinh )
15:03:53 <slava>     dup exp swap neg exp 2dup + 2 / -rot - 2 / ;
15:04:55 <slava> and here is the encore:
15:04:56 <slava> : cosh ( z -- cosh )
15:04:56 <slava>     #! Complex hyperbolic cosine.
15:04:56 <slava>     re-im dup real-sin swap real-cos rot real-cosh-sinh dup 2* ;
15:05:35 <slava> yes that's all i've achieved so far
15:05:40 <I440r> heh
15:05:44 <I440r> well its a start
15:05:52 <I440r> erm why are you not using table driven sin/cos ?
15:06:08 <slava> real-cos/sin is the C library
15:06:17 <I440r> ick :)
15:06:17 <slava> real = real numbers
15:06:23 <I440r> right
15:18:29 --- join: warpzero (~warpzero@dsl.142.mt.onewest.net) joined #forth
15:49:08 --- quit: tathi ("volleyball")
15:49:56 --- join: madwork_ (~madgarden@derby.metrics.com) joined #forth
15:50:04 --- part: madwork_ left #forth
15:51:21 <Herkamire> mmmm... volleyball
16:00:18 <ASau> slava, 1/x is faster than exp.
16:01:20 <ASau> And do not write universal functions like "sincos."
16:01:49 <ASau> That's overcomplexification.
16:02:52 <ASau> : ch exp dup 1/x + 2/ ;
16:03:03 <ASau> : sh exp dup 1/x - 2/ ;
16:03:32 <ASau> Feel difference.
16:06:36 <slava> thanks
16:07:26 <ASau> And for complex functions it's better to write one zexp and use it:
16:08:02 <ASau> : zch zexp 2dup z1/x z+ z2/ ;
16:08:17 <ASau> Everything's the same.
16:09:35 <arke> mark is opped?
16:09:39 * arke checks buffer
16:11:59 * arke checks logs
16:15:11 <arke> oh, autoop
16:15:12 <arke> heh
16:21:57 --- join: SDO (~SDO@co-trinidad1a-156.clspco.adelphia.net) joined #forth
16:22:31 <SDO> looksi, relisted... http://www.jdate.com/Applications/HotList/View.aspx?CategoryID=-1
16:47:25 * arke is away: brb
16:59:03 <futhin> arke: that's right, all the cool guys get auto-opped
17:11:17 <arke> ya know what?
17:11:19 <arke> fuck you./
17:11:19 <arke> :)
17:18:51 <slava> http://eugeneciurana.com/pastebin/pastebin.php?show=1925
17:20:05 <arke> slava: for factor?
17:20:08 <slava> yup
17:20:20 <arke> cool
17:20:26 <arke> looks just like forth
17:20:38 <slava> except its typed
17:20:46 <arke> bleh.
17:20:58 <arke> as i was just pointing out in #hprog..
17:21:00 <arke> typing=bad
17:21:26 <slava> depends on what you're coding etc
17:21:38 <ASau> Slava, what does "recip" mean?
17:21:46 <slava> ASau, 1/x
17:21:58 <ASau> Is it abbrev.?
17:22:14 <slava> reciprocal
17:22:18 <slava> bad name for a word i guess
17:22:25 <slava> there's some stuff i want to rename, and eliminate...
17:22:52 <ASau> Maybe it should be named "inverse?"
17:23:13 <ASau> Like "inversion" in geometry.
17:23:15 <slava> maybe 1/x -- i never considered that name until you posted code
17:24:15 <ASau> Though it seems "recip" is good too.
17:24:25 <ASau> Thanks for an idea.
17:29:12 <ASau> cos z = ch (iz)
17:29:35 --- quit: Robert (Read error: 54 (Connection reset by peer))
17:29:36 <slava> a few of the arc formulas are wrong. i haven't tested those yet.
17:29:37 <ASau> sin z = -i sh (iz)
17:30:15 <slava> you think i should define sin/cos like that?
17:30:33 <ASau> Just consider it.
17:30:38 <slava> yup.
17:34:32 --- join: Robert (~snofs@c-bf5a71d5.17-1-64736c10.cust.bredbandsbolaget.se) joined #forth
17:35:30 <ASau> BTW. Why not double and calculate some functions from ch (2z)?
17:35:39 <ASau> That'd be more effective.
17:36:26 <arke> what would you call a Forth word which does nothing but rearrange the stack and then call a few other words?
17:36:56 <ASau> Stack juggler.
17:37:11 <ASau> Why not?
17:37:15 <slava> i call it a 'shuffle' :)
17:37:27 <slava> i like ASau's term too
18:04:14 <Herkamire> "bad"
18:13:54 <Herkamire> I just sent off a proposal for 44% more than I usually charge.
18:14:23 <Herkamire> to someone who is not good to work for
18:24:51 --- join: Sonarman (~matt@adsl-64-169-92-7.dsl.snfc21.pacbell.net) joined #forth
18:26:31 * Herkamire goes off in search of his anaglyph glasses
18:28:52 <Sonarman> What for ?
18:30:37 --- join: doublec (~doublec@coretech.co.nz) joined #forth
18:30:55 <Herkamire> I wanna play with 3d graphics some more
18:32:03 --- quit: qFox ("this is mirc's last attempt of communication...")
19:18:12 <Herkamire> dang
19:31:58 <Herkamire> I can't find them
19:32:00 <Herkamire> poooey
19:35:51 <arke> hniii'
19:36:15 <warpzero> sup guys
19:36:21 <warpzero> you should listen to the radio :)
19:36:27 <warpzero> wza.us/radio.html
19:36:32 <arke> ack
19:36:32 <arke> ack
19:36:32 <arke> ack
19:36:33 <arke> ack
19:36:33 <arke> ack
19:36:34 <arke> ack
19:36:36 <arke> ack
19:41:58 --- join: kc5tja (~kc5tja@66-74-218-202.san.rr.com) joined #forth
19:42:05 --- mode: ChanServ set +o kc5tja
19:42:24 <fridge> arke: nak
19:42:40 <arke> hi kc5tja
19:42:43 <arke> hi fridge
19:45:50 <warpzero> hey kc5tja you should listen to my radio internet station
19:46:36 <kc5tja> warpzero: URL again?
19:46:47 <warpzero> wza.us/radio.html
19:47:41 <arke> warpzero: anything of interest playing? Rush? Genesis? Red Hot Chili Peppers? AC/DC?
19:48:48 <kc5tja> arke: So far, seems to be techno.
19:49:01 <arke> thats fine too. :)
19:49:22 <warpzero> arke: no :( i kicked it off with AC/DC
19:49:26 <warpzero> kc5tja: i do a wide range of stuff
19:49:33 --- join: hefner (~hefner@pool-141-157-60-191.balt.east.verizon.net) joined #forth
19:49:38 <kc5tja> I'm just reporting what I'm hearing so far.
19:49:59 <arke> yeah, techno right now.
19:50:04 <arke> kc5tja: how do you lke lisps?
19:50:25 <kc5tja> arke: I like them, but not as much as Forth.  But they are very powerful languages.
19:50:59 <arke> Scheme is nice, although I haven't done too much with it.
19:51:07 <arke> its in my top 3 favorite HLL :)
19:51:23 <arke> which is, in order, Forth, Objective C, Scheme
19:52:15 <warpzero> i need to try out Objective C
19:52:22 <arke> warpzero: it's AWESOME
19:52:34 <arke> warpzero: albeit still C, it does OOP just great.
19:52:41 <arke> warpzero: MUCH better than C++, imho
19:52:47 <warpzero> yeah i bet
19:52:52 <warpzero> C++ kinda... uh... sucks
19:52:58 <arke> uh huh 
19:53:14 <slava> C++ has stronger compile time type safety if you make use of templates.
19:53:18 <slava> (compared to C).
19:53:30 <slava> that is its advantage, IMO.
19:53:42 <arke> templates are shit.
19:53:49 <arke> well, i shouldn't say that.
19:53:50 <warpzero> its kindof an advantage
19:53:52 <arke> they're very nice
19:53:56 <arke> but they're horrible
19:54:48 <kc5tja> Darn, for a bit that sounded like Santana.
19:54:57 <warpzero> it is
19:55:09 <kc5tja> Hmmm..never heard this song from them before.
19:55:13 <kc5tja> Not at all like their usual style.
19:55:17 <warpzero> really?
19:55:26 <warpzero> huh
19:56:03 <kc5tja> When Santana breaks into a rip-roaring guitar solo, then I'll be re-affirmed.  :)
19:56:55 <warpzero> yeah here it is
19:56:58 <kc5tja> :)
19:57:01 <arke> i like this song :)
19:57:05 <kc5tja> It's good.
19:57:10 <kc5tja> Santana is an awesome band.
19:57:18 <arke> :)
19:57:47 <kc5tja> Yup, it's Santana.  Obligatory rhythm section solo.  :)
19:57:52 <kc5tja> Although short.  :/
19:58:06 <warpzero> :/
19:58:41 <kc5tja> This song has an interesting introduction.  Almost introspective.  But I know that won't last long.. :D
19:58:51 <arke> I'm here today to pump up the uplift mofo party plan
19:58:51 <arke> a plan based on a band, a band based on a plan
19:58:51 <arke> there should be no slaves in the land of lands
19:58:51 <arke> it's a hollywood jam
19:59:40 <warpzero> nah it pretty much sticks to its guns kc5tja
20:01:41 <arke> this is a cool song...
20:01:45 <warpzero> yeah
20:01:53 <warpzero> Mu-Ziq is pretty fucking good.
20:02:05 <kc5tja> It is, amazingly.  I don't normally like electronic music like this, but this is pretty good.
20:02:35 <kc5tja> Except for the captain insano drumming, the music behind it reminds me much of seascapes (above and below the water).
20:02:53 <arke> warpzero: send me mp3s of that!
20:03:00 <kc5tja> OGG!
20:03:06 <kc5tja> Support your local open source software!
20:03:16 <warpzero> i have them in what i have them in
20:03:29 <kc5tja> POLICE!!  I haven't heard this song in YEARS!!!
20:03:45 <kc5tja> God, I feel so old now... :)
20:03:49 <warpzero> its a good song
20:03:53 <arke> :-P~
20:04:06 <warpzero> do you folks want some more Mu-Ziq
20:04:08 <arke> warpzero: well, send me that song and all the other ones like it
20:04:14 <arke> warpzero: :)
20:06:40 <warpzero> well see how you like this next one its pretty mellow for the artist that does it
20:08:56 <arke> i like
20:09:15 <kc5tja> I don't like this song nearly as much as the previous techo song.
20:09:25 <warpzero> oh
20:09:32 <arke> same ehre
20:09:35 <arke> but I still like it
20:09:43 <arke> except that it sounds like a little acid on crack :P
20:10:28 <kc5tja> For all the raver music fans out there, y'all should listen to Be Careful With That Axe, Eugene.  :D
20:10:35 <kc5tja> That is easily the single most evil song I've ever heard.
20:10:40 <kc5tja> (by Pink Floyd, of course!)
20:10:59 <arke> warpzero: play that song if you have it
20:11:23 <kc5tja> for x in Acro*; do
20:11:24 <kc5tja> oops
20:11:33 <arke> :)
20:11:46 <warpzero> i dont have any pink floyd sorry
20:14:20 <arke> skynyrd
20:14:21 <arke> :)
20:14:25 <kc5tja> HAH!  Lynrd Skynnyrd (yr hyw evyr Ys yre ysed.)
20:15:06 <arke> :)
20:15:25 <arke> I wonder how bad the DrScheme GUI stuff is.
20:17:05 <slava> its *usable* but nothing stellar.
20:17:26 <warpzero> after the next song i think ill sign off
20:17:32 <arke> :(
20:17:35 <warpzero> im getting tired
20:17:43 <arke> warpzero: send me that muziq song though
20:19:04 <warpzero> k
20:19:08 <kc5tja> warpzero: Is this bjork?
20:19:26 <warpzero> Bjork and Plaid
20:20:06 <arke> kc5tja: btw, do you know about FICL?
20:22:18 <kc5tja> arke: I know OF it, but I've never used it.
20:22:59 <arke> kc5tja: well, it's VM is pretty horrible
20:23:08 <arke> even the new one thats supposed to be so much better.
20:23:26 <Sonarman> what's   hoirrible about it?
20:23:38 <kc5tja> arke: FICL is intended to be a scripting language; not a self-hosted Forth environment.
20:23:45 <arke> Sonarman: No direct literals (except -2, -1, 0, 1, 2)
20:23:52 <arke> Sonarman: and also, its switch-threaded :)
20:23:59 <arke> kc5tja: I know. even for that, it's bad. 
20:24:02 <arke> kc5tja: well, not bad
20:24:05 <arke> kc5tja: just kinda sucky :)
20:24:31 <arke> kc5tja: I was thinking of a simple stack-addressing bytecode
20:24:37 <Sonarman> you'd rather have DTC/ITC than switch-threaded bytecode?
20:24:51 <Sonarman> what do you mean, no direct literals?
20:24:55 <arke> Sonarman: definetely
20:25:03 <Sonarman> yeah, me too i think
20:25:28 <doublec> I've used ficl on a cellphone to develop an nntp newsreader. worked well and ran well.
20:25:35 <warpzero> kc5tja: what do you think of this song?
20:25:41 <arke> Sonarman: when you want to push, say, 5, you have to compile 2 2 1 + +
20:26:06 <arke> doublec: :)
20:26:06 <Sonarman> jeezus
20:26:14 <arke> doublec: could be faster though :)
20:26:18 <arke> Sonarman: :)
20:26:21 <doublec> we are talking about ficl.sourceforge.net right?
20:26:45 <doublec> If I want to push 5 I just push 5. No need to do 2 2 1 + + or am I misunderstanding?
20:26:57 <kc5tja> warpzero: Better than the other one.  But not as good as Enigma.  :)
20:27:06 <warpzero> oh pooh
20:27:08 <Sonarman> Enigma?
20:27:08 <warpzero> well goodnight
20:27:09 <doublec> The phone was the Nokia 9210 - it has a keyboard so developing on it wasn't too bad.
20:27:20 <kc5tja> Sonarman: A very cool band.
20:27:21 <Sonarman> kc5tja: oh, i thought you were talking about a virtual machine
20:27:21 <arke> doublec: well, _you_ do 5, but FICL's internal VM executes 2 2 1 + +
20:27:25 <Sonarman> :)
20:27:30 <arke> doublec: unless I'm missing something.
20:27:32 <doublec> oh, ok
20:27:40 <arke> doublec: but from the code here, that's how it seems like.
20:27:46 <kc5tja> warpzero: 'night
20:28:06 <arke> night warpzero
20:30:52 <ASau> Hm. And what is good in ObjC, I wonder?
20:33:11 <ASau> As for me, I like Scheme more than all Cs.
20:33:30 <ASau> At least it's very good calculator.
20:45:44 <Sonarman> arke: ok, i have found evidence to suggest that ficl's system of handlign literals isn't as stupid as we'd thought
20:46:25 <Sonarman> apparently, you were looking in vm.c
20:46:35 <Sonarman> in that file PRIMITIVES are handled
20:46:54 <Sonarman> but in ficl.h there is:
20:47:08 <Sonarman> typedef enum { ... LITERAL, ... } WORDKIND;
20:51:06 <Sonarman> it''s just that they made, -2 .. 2 primitives because those numbners are often pushed, so eliminating the overhead of fetching the literal value from IP for those numebrs seemded lieke a good idea
20:53:58 <arke> then why isn't that handled in the innerLoop?
20:54:51 <arke> Hrm.
20:54:52 <arke> Ooh.
20:54:53 <arke> I get it.
20:55:03 <arke> the innerLoop is separate :P
21:02:32 <slava> switch threading is silly
21:03:12 <arke> indeed.
21:03:16 <ASau> Ban byte code!
21:03:20 <ASau> Ban C!
21:03:31 <arke> ASau: sometimes you just gotta go with the flow though :)
21:03:38 <arke> slava: a jump table would be better, ey?
21:04:15 <ASau> arke. Take eForth.
21:04:28 <ASau> Hm.
21:04:47 <ASau> Make eforth.
21:06:19 <arke> I would make a simple, stack-addressing bytecode with few instrucitons.
21:06:25 <arke> I guess 16 might even be enough.
21:06:33 <arke> or 32
21:06:39 <arke> defininetely so that I can mask though.
21:12:58 <Herkamire> arke: what are you programming?
21:16:25 <ASau> He's advancing Robert's F, AFAIK.
21:17:31 <futhin> anyone know a free shell account provider off the top of their head?
21:17:38 <futhin> telnet
21:19:09 <doublec> http://sdf.lonestar.org/
21:20:46 <Herkamire> heh, we should have a forthswap week
21:21:06 <Herkamire> where we each work on somebody else's forth instead of our own for a week
21:23:23 <ASau> It's interesting idea.
21:23:46 <ASau> Another week, though.
21:25:49 <Sonarman> switch-threading does use a jump-table
21:26:07 <Sonarman> unless yoru C compiler really sucks
21:26:58 <Herkamire> Sonarman: I put out herkforth-0.6
21:27:14 <Herkamire> s/put out/released/
21:34:43 <Sonarman> cool!
21:35:28 <Sonarman> i'll check it out soon (can't do it now; homework, which IRC is keeping me from doing efficiently :) )
21:51:26 --- quit: doublec ("Leaving")
22:13:23 --- join: Serg (~z@212.34.52.140) joined #forth
22:13:34 <arke> hi Serg
22:20:18 <Serg> hi !
22:20:37 * Serg gonna need new pants - main database was nearly lost
22:21:10 <Sonarman> lol :)
22:23:43 <kc5tja> Serg is my clock -- when he joins, I know I need to get to bed.  :)
22:24:11 <arke> teehee :)
22:24:21 <arke> kc5tja: can I ask you something tho?
22:24:31 <kc5tja> What's up?
22:24:35 <arke> kc5tja: what do you think about a stack-addressing bytecode?
22:25:07 <kc5tja> You are aware that Java Virtual Machine is a stack-addressing bytecode, right?
22:25:28 <kc5tja> (it's not a 100% 0-operand stack engine, but it nonetheless IS a stack machine)
22:25:42 <futhin> all forthers of the world, move to las vegas!
22:25:43 <arke> er
22:25:46 <arke> mabe I got this wrong :)
22:26:00 <Sonarman> what does stack-adressing mean?
22:26:03 <arke> I meant that instructions reference items that are X distance from TOS
22:26:05 <arke> like that
22:26:11 <futhin> kc5tja: you planning on moving to las vegas anytime soon? ;)
22:26:17 <arke> xchg 0,1 <-- swap
22:26:18 <kc5tja> futhin: No.
22:26:29 <kc5tja> arke: Whatever.  I have no opinions.
22:26:30 <futhin> awww
22:26:40 <kc5tja> futhin: Why, are you?
22:26:42 <arke> kc5tja: ...
22:26:42 <arke> ok
22:26:43 <arke> :)
22:26:45 <futhin> yup
22:26:57 <kc5tja> futhin: Wow.  Big change.  What's the occasion?
22:27:00 <arke> AND THEN HE MIGHT GET REAL ISP
22:27:05 <futhin> haha arke
22:27:37 <kc5tja> LV is a lot of fun sometimes, but boring a lot of times too.
22:27:43 <kc5tja> Lots of shows to see there.
22:27:46 <futhin> kc5tja: well i'd like to move to las vegas.  i'll probably go there towards the end of the summer, and maybe stay there depending on how things develop
22:27:49 <kc5tja> But lots of casinos too.
22:28:11 <futhin> got two IRC friends with plans to move to las vegas
22:28:20 <futhin> so if i move there that'd be interesting :D
22:28:32 <futhin> two separate IRC friends
22:28:34 <kc5tja> I have no problems with visiting LV now and again.
22:28:38 <kc5tja> Maybe next year, I dunno.
22:28:44 <kc5tja> But I sure as hell wouldn't want to live there.
22:28:49 <kc5tja> OUTSIDE of LV, yes.
22:28:54 <kc5tja> But not in LV itself.
22:29:00 <futhin> how come?
22:29:06 <Serg> i call our Moscow "Las Figas" - lotsa casinos and "figa" (fuck gesture) in a pocket
22:29:06 <kc5tja> Because I like my solitude.
22:29:07 <futhin> too flashy? noisy?
22:29:09 <futhin> ah
22:29:24 <arke> :)
22:29:25 <kc5tja> When the sun sets, I like to see at least a few stars in the sky, you know?
22:29:27 * arke had this cool idea
22:29:30 <kc5tja> Being an amateur astronomer and all.
22:29:40 <arke> a city comprised of nerds 
22:29:45 <arke> :P
22:29:58 <futhin> heh
22:30:11 <futhin> heh to kc5tja i mean
22:30:23 <futhin> arke: yeah i want a geek commune
22:30:24 <arke> I really think this would work.
22:30:37 <arke> (1) We need a cool company.
22:30:42 <kc5tja> futhin: I'm shocked.
22:30:52 <kc5tja> futhin: Communes imply collectivism!!  HEATHEN!!  :D
22:30:59 <futhin> haha
22:31:07 <arke> like, for example, IBM. They aren't prefect, but they're cool. For various reasons.
22:31:25 <arke> And also, they have teh money.
22:31:35 <arke> (2) It needs to be a large dome structure.
22:31:41 <arke> no rain, no anything.
22:31:41 <futhin> kc5tja: is there anyway to avoid collectivism? and what about spouses? 
22:31:43 <arke> everything is open.
22:32:13 <arke> and you move around not using cars, but rather using magneto trains that move up and down.
22:32:23 <arke> as needed.
22:32:48 <arke> It would look a little like that one star wars city in that one movie
22:32:52 <kc5tja> futhin: Listen to the introduction to collectivism seminars that Dr. Peikoff gives (I'd have to find you the URL, but I'm sure Google can find it just as well).  He discusses it.
22:33:20 <futhin> arke: as you grow up you'll realize that your dream is unachievable. then you'll realize the solution is to colonize the asteroid belt!
22:33:29 <kc5tja> arke: In case you haven't noticed, that "one" city comprised the whole planet!!  Corrascant.
22:33:39 <arke> bleh.
22:33:46 <arke> kc5tja: well, it wouldn't.
22:33:52 <futhin> jeez i forget why i'm using firefox and not IE
22:33:54 <arke> kc5tja: it'd be about as big as camarillo.
22:34:17 <arke> futhin: because IE doesn't allow tabbed browsing, doesn't follow standards, and is in general a big hunk of steamin' shit
22:34:25 <arke> kc5tja: with a big dome.
22:34:38 <futhin> the standards are IE's standards :P
22:34:39 <arke> kc5tja: you can walk anywhere, or you can take the magneto train.
22:34:47 <arke> kc5tja: its completely socialistic.
22:34:52 * kc5tja refuses to trust magnetic vehicles that hover more than about 4 inches off the ground.
22:35:35 <futhin> arke: time for you to read "The Fountainhead" and "Atlas Shrugged" and "microeconomics 101" :P
22:35:38 <arke> kc5tja: you work for IBM, and your reward is living there and an allowance depending on what you do and how good you do it and all that
22:35:44 <kc5tja> arke: I think I'd like to see socialism through objectivism, rather than socialism by fiat.
22:36:02 <arke> every computer is shared.
22:36:05 <futhin> kc5tja: agreed
22:36:12 <arke> nobody owns their own computer.
22:36:26 <kc5tja> arke: Sorry, that I *patently* disagree with.
22:36:30 <arke> you just log into a terminal, and choose what OS>
22:36:40 <kc5tja> If I shared the philosophy, I'd be using Windows, what with all their Digital Rights Management bullshit on it.
22:36:43 <futhin> arke: nobody owns the roads.  same set of problems, it requires a government with all the flaws.
22:37:21 <arke> thats, of course, just a deram.
22:37:36 <kc5tja> Anyway, I'm off to bed.
22:37:41 <kc5tja> I really do need to get some sleep.
22:37:43 <arke> kc5tja: 'nigh5t
22:37:50 --- quit: kc5tja ("THX QSO ES 73 DE KC5TJA/6 CL ES QRT AR SK")
22:38:15 <arke> futhin: so, whaddaya think mi amigo!?
22:38:33 <futhin> its a good start, its good to be an idealist and to dream :)
22:38:40 <futhin> i had similar thoughts too
22:39:13 <futhin> i recommend reading a book on economics particularly microeconomics
22:39:37 <futhin> ignore any macroeconomics book as its all bullshit
22:47:53 <arke> :)
22:48:03 <arke> I'm not too much into economics etc.
22:48:32 <arke> but I just put all my messaging programs away and installed gaim
22:48:34 <arke> much better
22:53:26 <futhin> i had strong communist/socialist leanings until i learned about economics
22:53:47 <futhin> actually before i took the class i thought economics was quite boring & weird
22:53:59 <futhin> but i took a class and lucked out with a very interesting textbook
22:54:06 <futhin> and read the whole thing
22:54:22 <futhin> read it twice in fact, again before the final exam
22:55:00 <futhin> anyways it was quite exciting because now i could actually understand why poverty existed and how to solve it
22:55:12 <arke> :)
22:55:41 <futhin> main reason for poverty is actually the governments and their interference with the local economy
22:55:55 --- quit: Sonarman ("leaving")
22:56:00 <futhin> as well as the bureaucratic red tape etc
23:07:19 * arke is away: sleep
23:10:10 --- join: arke_ (~Chris@wbar8.lax1-4-11-100-108.dsl-verizon.net) joined #forth
23:10:31 --- quit: arke (Read error: 104 (Connection reset by peer))
23:13:17 <Serg> futhin: root of poverty is in minds of poors
23:13:52 <Serg> i was poor to misery all relatives, changed mind -> above average
23:14:04 <Serg> AS all relatives, typo
23:14:21 <Serg> now need to change to another step - business ;))
23:15:08 * Serg reboots, BRB
23:15:10 --- quit: Serg ()
23:58:03 --- join: Serg (~z@212.34.52.140) joined #forth
23:59:59 --- log: ended forth/04.06.01