00:00:02 --- log: started forth/00.12.08 00:00:02 ok, then I won't either 00:00:25 sorry hehe 00:00:34 not to worry, I solved part of it 00:00:41 :) 00:00:42 and relearned some bash 00:01:06 bash is cool, sometimes i think i shud realy learn it 00:01:15 i was going to do a forth interpreter in bash 00:01:19 you know, I would very much like to see a forth as versatile as TCL 00:01:26 and then write a bash interpreter in forth 00:01:30 and have them interpret each other heh 00:01:48 but I think forth has a very long way to go until it gets there 00:02:00 and I'm not sure it ever will 00:02:19 it may remain forever the domain of embedded controllers 00:02:30 where it belongs 00:03:05 you think so? 00:03:05 well i know it does 00:03:05 but i dont think its restricted to there 00:03:05 i just think that that is where it excells 00:03:05 you're spending a lot of effert to make a clean, fast one for a full-blown op-sys 00:03:05 it is 00:03:21 but I think it *could* excell in general programming, too 00:03:33 if someone would develop it properly 00:03:46 and I think the job is too big for just me 00:04:12 it needs a true following 00:04:17 but it will never have that 00:04:28 not a very big one, anyway 00:05:27 no 00:05:33 just a few dedicated individuals 00:05:37 I think I'm going to document it similar to [IF] 00:05:44 want to see a clip? 00:05:48 sure 00:06:14 Compilation: Perform the execution semantics given below. 00:06:17 Execution: ( flag | flag "name" ... -- ) 00:06:27 If flag is true, do nothing. Otherwise, skipping leading spaces, parse and discard space-delimited words from 00:06:27 the parse area, including nested occurrences of [IF] ... [THEN] and [IF] ... [ELSE] ... [THEN], until either the 00:06:41 word [ELSE] or the word [THEN] has been parsed and discarded. If the parse area becomes exhausted, it is 00:06:42 refilled as with REFILL. [IF] is an immediate word. 00:06:55 [done] 00:07:25 I think that is quite clear clear, and fits my need pretty closely, thank you for reminding of [IF] 00:08:21 i didnt... you rememberd it urself hehe 00:08:25 i just nudged u that way hehe 00:08:38 yes, something you said reminded me of it 00:08:47 yup 00:09:18 rendered HTML just doesn't paste into IRC worth a shit 00:09:58 hehehe 00:12:33 hey, how should methods be specified? 00:12:50 can I give you an two examples for you to compare? 00:14:06 sure 00:14:13 ok 00:14:16 give examples, dont ask if u can :P 00:14:18 ---- start 1 00:14:28 s" 279AFA83-4981-11CE-A5210020AF0BE560" Interface: IDSound 00:14:28 IDSound Methods: 00:14:28 IDS.CreateSoundBuffer 00:14:28 IDS.GetCaps 00:14:28 ;methods 00:14:35 --- end 1 00:14:37 ACK! 00:14:39 erm 00:14:40 wtf is that? 00:15:14 lemme try again ... 00:15:18 it's forth 00:15:26 ---- start 1 00:15:28 s" 279AFA83-4981-11CE-A5210020AF0BE560" Interface: IDSound 00:15:28 IDSound Methods: 00:15:28 IDS.CreateSoundBuffer 00:15:28 IDS.GetCaps 00:15:28 ;methods 00:15:33 ---- end 1 00:16:10 ---- start 2 00:16:11 wow 00:16:13 s" 279AFA83-4981-11CE-A5210020AF0BE560" Interface: IDSound 00:16:13 IDS.CreateSoundBuffer 00:16:13 IDS.GetCaps 00:16:13 ;methods 00:16:16 weird forth 00:16:20 ---- end 2 00:16:25 wow what? 00:16:54 actually, it's pretty mundane forth, your overlooking the s" at the beginning of the ID string 00:17:27 how is that s" associated with what interface: creates ? 00:17:40 I'm going to chang ";methods" to ";interface" 00:17:56 i would suggest that actualy hehe 00:18:14 I4: it is the OSF format representation of the GUID that identifies that particular interface 00:18:16 but how is what s" compiles associated with what interface: compiles 00:18:23 yes i know 00:18:26 i fogured that part 00:18:36 but you have s" and interface: 00:18:44 they both compile stuff 00:18:49 this is meant to be used in interpretive mode ... to build structures 00:18:53 but the stuff they compile are not associated that i can see 00:19:17 s" normally wouldn't place anything in the dictionary 00:19:27 well, it is actually 00:19:32 hehe 00:19:42 the string specifies an interfaces 00:19:45 the string specifies an interface 00:19:45 who compiles the string data then ? 00:19:58 hrm 00:20:19 Interface: parses the string and places the binary array in the dataspace for the DSound word 00:20:32 the string specifies an interface 00:20:49 the interface consists of a group of function calls 00:21:00 the are not named, but are numbered 00:21:19 :) 00:21:22 the number is incremented and stored with each name that is parsed 00:22:20 this is all AFTER the object is created 00:22:45 an object can have more than one interface 00:23:23 when an interface for an object is created, you get a pointer to a table of function vectors 00:23:24 :) 00:23:28 this is all beyond me hehe 00:23:48 the names are mnemonics for offsets into the function table 00:24:18 ok, do you want me to start over, or just drop it? 00:24:41 my forte is LOW level code hehe 00:25:25 believe me .... when you execute IDS.CreateSoundBuffer , it does @ @ EXECUTE 00:25:35 that's about as low-level as you get 00:25:58 and it does the same thing on Linux, BSD, Windoze, and MacOS 00:26:34 hehe 00:26:54 I'm wondering if the second example doesn't make Interface: too complex 00:27:12 because it parses the string AND all the method names 00:27:30 the first example had two different words to do it 00:27:35 ok, do you want me to start over, or just drop it? 00:27:43 hehe 00:27:46 erm 00:27:54 all compiled code is low level 00:28:00 the source is what is high level 00:28:05 right 00:28:07 i cant think high level 00:28:12 COM is a binary standard 00:28:20 even when i code c i code low level 00:28:21 like VFAT or ext2 00:28:37 vfat and ext2 are complicated beasties 00:28:52 BUT low-level stuff can get incredibly complex ... 00:28:57 yes 00:29:05 the purpose of high-level code is to make things vastly simpler 00:29:06 bot alot simpler also 00:29:09 erm 00:29:10 no 00:29:19 it tries to put alot of shit into a small space 00:29:31 tht isnt a simplification :P 00:29:48 ok, look at the interface for the "open" syscall in linux ... 00:30:03 it is a VERY simple interface 00:30:20 a string filename, a few mode bits, maybe 10 at most 00:30:55 but that is the tried and tested interface to one of the most complex software systems the world has EVER seen 00:32:38 well 00:32:53 it still packs alot of complicated shit into one small word but 00:33:15 what you have is ALOT of simpler problems that it covers 00:33:18 opena file 00:33:21 open a socket 00:33:24 open a device 00:33:25 etc 00:33:29 yes, a terminal ... 00:33:32 you GOT it 00:33:44 but each of these is a simpler problem than the whole 00:33:46 it's absolutely beautifull in conception 00:34:07 if there were a seperate call for each case, it would be a NIGHTMARE 00:34:14 treat everything as a file 00:34:20 forth treats everything as a word 00:34:22 or a number 00:34:26 right 00:35:02 I have a similar situation, and I expecting to write a high-level interface to COM that is simple and straight-forward 00:35:38 theoretically, it should be doable any any platform ... although some forths are to broken to do it 00:36:40 I take it your bored of the discussion? 00:36:58 no hehe 00:37:25 ok, would you like a 10 minute overview of what a COM object is? 00:37:30 sure 00:37:37 what does com stand for again ? 00:37:39 so you can see what bits of info are needed? 00:37:47 Component Object Model 00:38:02 dcom = Distributed Component Object Model 00:38:05 oh hang on before u go on.... 00:38:15 is the standard for com freely available ? 00:38:29 and is it a microsoft invention ? 00:38:37 yes and yes 00:38:39 hrm 00:38:49 then how come you ahve to PAY ms for the omf file format 00:38:50 grrr 00:38:53 that sux 00:39:06 it actually works pretty good, it is what IE5+ is written with 00:39:26 including all the mutimedia stuff 00:39:35 i dont like any ie versions at all 00:39:45 it sure beats the HELL out of DLLs 00:39:46 not because they are ms 00:39:52 but because i dont like the interface 00:39:56 oh, ok 00:40:02 netscape looks better and feels better to me 00:40:19 how come they give away com and charge for omf ? 00:40:21 I use netscape myself, simply because IE never let me preview pages before I printed them 00:40:28 what is OMF? 00:40:43 if i want my simulator to support object modual format then i have to PAY microsoft 00:40:45 grrrr 00:40:48 object model format? 00:41:11 you mean .OBJ files? 00:41:12 .obj files are usually omf 00:41:21 .obj is the extention 00:41:24 omf is the file format 00:41:34 ok, go buy the MSDN disks, they're cheap 00:41:41 no 00:41:44 not a fscking chance 00:41:52 not 1 in 100 billion chance of that ever happening 00:42:06 the .OBJ formate is part of of the Portable Execution specification 00:42:10 why not? 00:42:11 not if i got 100 billuion cd's for 1 cent 00:42:26 because i will NOT support microsfot AT ALL 00:42:36 not one red cent 00:42:38 I only suggested it because it's much faster than goig to the MSDN website 00:42:51 it's all online, but it's faster off your HD 00:42:58 if i could get a guarnateed 100 percent profit per day for the rest of my life for buying 1 share of ms 00:43:05 anyway, the .OBJ format is online 00:43:05 i wouldnt fscking buy a sinbgle share 00:43:17 omf needs to be licenced 00:43:25 also, it is a bastardization of .COFF format 00:43:41 no more than .EXEs need to be licensed 00:43:56 erm 00:44:03 bally gaming 00:44:12 whome i worked for 00:44:17 BUT you may need a license to place the "Microsoft Approved" software tag in the .OBJ file 00:44:24 paid a HUGE ammount of money for the omf file format documentation 00:44:34 heh, it's online 00:44:40 what year was this? 00:44:52 came in leather-bound binders? 00:44:55 i worked for them a cpl of years ago 00:45:03 erm no just a ring binder 00:45:07 actually 00:45:09 they paid 00:45:12 with each page in a sheet protector? 00:45:16 then got access to download the pdf 00:45:26 and printed ONE copy thereof 00:45:34 thats all they were entitled to print 00:45:56 ok, want me to answer why ms published COM? 00:46:20 yea 00:46:55 ok, here's the marketing scene ... 00:47:07 in the PC market, there's ms and MacOS 00:47:56 after gates friend, Steve Jobs, sued him for espienage (and all the maneuerings in the 80s) they settled down to a pretty comfortable truce 00:48:22 ms was for business aps, and targeted corporate 00:48:33 Apple was for home users, and targets schools 00:48:57 everything else was mini and main frames, not enough to write software for 00:49:13 now along comes Linux 00:49:24 and shortly thereafter, FreeBSD 00:49:53 at first, Gates and Jobs ignored them altogether 00:50:08 but they got bit in the ass for their ignorance 00:50:12 :) 00:50:41 one day, they woke up and discovered that 8% of all PCs were running neither Windows or MacOS 00:50:56 (I think it's like 12-15% now) 00:51:10 they saw they were in serious trouble 00:51:31 now, it WAY too late for ms to try simplifying WinNT 00:51:52 so ms starts writing apps for Linux 00:51:53 :) 00:51:58 they did ? 00:51:59 (or saying it's going to) 00:52:02 ms writes linux apps ? 00:52:04 yes, they do 00:52:15 where ? 00:52:18 does anyone use them ? 00:52:19 not sure they sell them yet 00:52:38 not that I know of, hard to compete against open freeware 00:52:44 :) 00:53:01 sure is :) 00:53:04 but all that code sitting around at MS is useless on linux or freebsd 00:53:28 whether it's good code or bad code is irrelevent 00:53:49 AND ms was having serious problems with versioning in DLLs 00:53:52 nono EVIL code hehe 00:53:57 yes 00:54:00 they were 00:54:26 applications would randomly replace a DLL with the "newest" version they knew about, 5 years old 00:54:39 all the new apps would suddenly quit with system erros 00:54:56 yup 00:54:59 it simply wasn't working 00:55:11 which is REALY stupid 00:55:22 replaceing a newer dll with the newest YOU know about 00:55:25 ok, so ms went pure object-orientation, at the binary level 00:55:33 right, I4 00:56:10 and they want an object to work EXACTLY the same way on Linux, FreeBSD, MacOS, etc 00:56:26 even the amiga didnt do that, if the installed app cant use the one in the libs dir it stored the one IT wanted in its OWN dir 00:56:40 because eventually, they're likely to start porting their code 00:57:00 they will have to 00:57:10 but nobody will BUY their code any more 00:57:15 not linux ppl anyway 00:57:38 to port ie5 to XWin, for example, will involve mostly writing wrapper-macros and recompiling the COM objects 00:58:03 (and getting sued again by Netscape, most likely) 00:58:03 ug! 00:58:09 hehehe 00:58:22 netscape has a patent on linux web browsers ? 00:58:27 why woyuld that cause a law suit ? 00:58:39 so they wrote AND published the COM and DCOM standard 00:58:52 I think its a good one 00:59:21 it's been tested effective in Win95, Win98 and WinNT 00:59:23 i hear a "but" on its way 00:59:39 well, if you want one ... 00:59:53 but it's ms, so Linux hackers are reluctant to play with it 01:00:02 :) 01:00:09 anyway, you logging? 01:00:13 no 01:00:17 i never log but clog is 01:00:21 clog is a channel logger 01:00:30 COM = Component Object Model 01:00:32 i believe all this goes on the #tunes web page 01:00:42 DCOM = Distributed Component Object Model 01:01:30 an object is a ethereal thing floating in memory somewhere and can only be reached by "Interfaces" 01:01:55 an object need not be in the current memory space, nor even on the same computer 01:02:14 this is where me and OOP have a parting im affrade 01:02:33 most objects are in the current process, of course, but the params need to be moved, it's up to the interface to do the work, not the object user 01:03:00 a class is just a structure. you create instances of this class and each instance is a structure. its wasteful of memory 01:03:05 all I, the programmer/object user need to know about is the interface 01:03:29 yes, they did much to reduce that waste, though 01:03:41 k 01:03:46 ok, so I want to create an object of type BLAH 01:03:50 ok 01:04:02 i like that type, its almost as good as type FOO hehe 01:04:12 blah could be a stack, a list, a hardware interface driver, anything 01:05:06 now, objects of type blah have an identifier that makes them seperate from all other object types 01:05:53 lets store that "type" in the variable "BlahID", however it's specified 01:06:13 ok 01:06:21 now, to access a blah object REQUIRES an interface 01:06:52 an interface being a set of functions with wich to manipulate and interact with BLAH type objects 01:07:06 yes 01:07:15 constructors, destructors etc 01:07:21 all objects understand at least on interface, IUnknown 01:07:40 or they aren't COM objects, by definition 01:08:27 the IUnknown interface has exactly three functions, called AddRef QueryInterface and Release 01:08:53 and so every com object ever created understands at least those functions 01:09:17 ok 01:09:34 in addition, the first three functions of EVERY interface must implement AddRef QueryInterface and Release, in that order 01:10:39 so let's say that the author of the BLAH type objects wrote two other interfaces, which he chose to call BlahSet1 and BlahSet2 01:11:19 now we want to create a blah object, and access it with the BlahSet1 interface 01:11:25 follow so far? 01:11:39 yes 01:11:51 each of those interface has a completely different set of functions, sometimes not even related to each other 01:12:42 like IUnknown certainly has little to do with whatever Blah objects do, only in creating and destroying blah objects and blah interfaces 01:13:14 ok 01:13:29 so we call a function that specifies 1. the type of object we want to create and 2. the particular interface we want to use 01:13:53 now, this is a regular function call for whatever OS the implementation is on 01:14:25 one Windows, it CoCreateInstance (I think in comobj.dll) 01:14:44 in linux, it's prob Create in /lib/comlib.xxx 01:15:23 in my Forth interface, it will be COM-Create 01:15:53 now, to prevent versioning probs... 01:16:16 each object type is globally usique 01:16:19 unique 01:16:32 and each interface type is globally unique 01:16:59 and an object can also be globally unique, but the user doesn't need to worry about that, just the object does 01:17:21 guess what happens when you need a globally unique identifier? 01:17:28 erm 01:17:32 u gotta register it 01:18:11 whe u say globally i assum3e u dont mean globally on that machine 01:18:13 actually, quite close 01:18:15 u mean GLOBALLY 01:18:28 there is a published standard on how to create global identiers 01:18:35 there has to be an issuing authority 01:18:37 (they needn't be registered) 01:18:51 registering is used for distribution, though 01:19:07 do you have a windows machine handy? 01:19:12 my laptop 01:20:10 ok, lemme find the progname 01:21:09 ok ill get coffee while u do that 01:21:34 ok, it's called UUIDGEN.exe 01:21:53 but it doesn't come with windows, but with one of the developer products 01:22:17 if you just run it, it spits out a number that is guaranteed to be unique throught the universe 01:22:38 -n lets you spit out X many of them, 5 ... 100 whatever 01:23:03 guaranteed ? 01:24:11 yes 01:24:37 the OSF spent a LOT of time coming up with an algorithym that would never collide 01:24:39 ok 01:24:59 if you have a network card in your machine, it's guaranteed never to collide 01:25:04 can tehy PROVE it will NEVER collide ? 01:25:09 yes, they can 01:25:13 erm 01:25:15 with a network card 01:25:29 erm each network card is ASSIGNED an id 01:25:40 each card has a 6byte serial number guaranteed to be unique for each card 01:25:43 and there are a finite number of id's 01:25:48 yes 01:26:01 its called a mac number 01:26:07 andthose numbers are allocated in blacks to manufacturers who ask (or pay) for them 01:26:09 and THAT is your TRUE internet address 01:26:18 yes 01:26:19 in blocks, rather 01:26:26 i can read perfect mistype :) 01:26:43 BUT 01:26:43 so part of you UUID is your network card serial number 01:26:59 you arentaliking about running a program that spits out unique values 01:27:04 another part is a timestamp 01:27:15 what if you DONT have a nic 01:27:17 another part is a serialization 01:27:34 right, they don't guarantee uniqueness without a network card 01:28:18 but they use the ROM serial number and other neat things to hash out a 6 byte number is incredibly likely to be unique 01:28:34 or the pentium id 01:28:39 the serial id on the chip 01:29:17 and, on top of that, the timestamp nearly guarantees that two machines that happen to has to the same serial id certainly won't produce GUIDs in the same second 01:29:22 erm but not guaranteed. 01:29:27 I'm not sure which serials they hash 01:29:53 but, for all intents and purposes, GUIDs *ARE* unique 01:30:03 ok 01:30:17 1 in 100 billion is good enough for government work :) 01:30:22 and I'd bet my life that if you remove my ethernet card, the resulting number will never collide 01:30:50 GUIDs are 128 bits 01:31:15 that's why you see the BIZARRE and incredibly long string given to Interface: 01:31:20 a humongous number 01:31:52 because it needs to know EXACTLY which interface it's creating on the object 01:32:24 so the CoCreateInstance function gets pointers to TWO of these monstrosities to create your object for you. 01:32:42 one for the object type, one for the interface type 01:32:49 guess what you get back? 01:32:56 er let me guess 01:32:58