00:00:11 AnMaster: that doesn't excuse putting core functionality in modules. 00:00:44 mib_8oomfc, I swear there are more modules than when I last read that page you linked.... 00:01:16 some are not included in core package hm 00:01:19 I should write my own damn bouncer. 00:01:26 And every piece of software. 00:01:27 mib_8oomfc, sounds nice :) 00:01:28 I can't trust others. 00:01:37 They suck. :P 00:02:06 I used to have my own bouncer, but znc was easier, and had what I wanted, rather than coding the missing features myself I changed to znc 00:02:17 Happy xmas. 00:02:21 -!- mib_8oomfc has set topic: Real christmas. 00:02:34 mib_8oomfc, happy day after xmas 00:02:36 miau is very simple, one network only (I use a script to launch multiple miaus) 00:02:43 -!- AnMaster has set topic: Real christmas | The day after the real one. 00:02:58 comex, ew 00:03:21 comex: does it do the playback? 00:03:34 mib_: yes, it has a "quicklog" 00:03:39 explain 00:03:51 mib_8oomfc, znc does playback on every channel, I'm not sure if you can turn it off for some channels, but it may be quite possible 00:04:09 * AnMaster looks in the web UI 00:04:15 when you log in, it replays the quicklog. I don't remember how configurable it is 00:04:29 comex: on the original channel? 00:04:40 yes 00:05:21 the thing is, IMO irc bouncing isn't exactly _hard_. when the user is connected, you do just forward everything. when they disconnect, log everything you get from the server. when they reconnect, send the log to them and erase it. 00:05:46 mib_8oomfc, znc can do that easily 00:05:49 oh and it does support multiplec onnections (although I assume all others do as well) 00:05:57 what I'm not sure is if it is per channel 00:05:59 as you requested 00:06:03 why do you need a frobtapulous perl modulized C++ elegant blaaaaah thing just to do that 00:06:17 mib_8oomfc, you *CAN* turn off the timestamps 00:06:18 it seems 00:06:24 comex: what about that weird away bug 00:06:25 by setting the timestamp format string to empty 00:06:27 :D 00:06:46 mib_8oomfc, you can disable that 00:06:50 i should, just lazy :p 00:06:50 mib_8oomfc, so that issue is solved 00:06:50 comex: neat. 00:06:53 i bet i could write a bouncer that makes me happy in like 50 lines of Haskell 00:07:06 do it then 00:07:13 not now :P 00:07:34 mib_8oomfc, one thing I like with znc is that you can make it automatically add/remove channels to the "auto join on connect" list if you want 00:07:39 that is configurable of course 00:07:47 AnMaster: but my client does that 00:07:50 :P 00:08:03 the bouncer should: 00:08:08 - play back things it saw when disconnected 00:08:21 - when it gets disconnected, reconnect, send out what commands you've put in its config, and join channels it was in 00:08:24 'sit 00:08:32 mib_8oomfc, znc can do those two 00:08:38 there is the perform module for the second 00:08:39 to be honest, I could probably write _everything_ I want in 50 lines of haskell 00:08:42 I use it to auto-oper up 00:08:45 on another network 00:08:49 for example 00:09:11 away can do the first 00:09:18 so yes znc can do those 00:09:25 and it will auto reconnect 00:09:29 i'll go with miau for now 00:09:30 it seems ko 00:09:31 ok 00:09:41 I haven't used the nickserv module with znc since I use other strange services 00:09:45 account based and such 00:10:10 --disable-ascii-art Disables fancy ASCII banner miau prints at start-up and when a client conencts to miau. 00:10:16 haha 00:10:17 that kind of thing is why I hated psybnc :P 00:10:25 mib_8oomfc, znc doesn't have such a thing 00:10:29 PSYCHOID AND THE MOST COOL LAM3RS GROUP EFNET 00:10:33 i know it by heart 00:10:39 ouch 00:10:41 that one HURT 00:10:43 I hated psybnc for some other reason, can't remember what it was exactly 00:10:45 BADLY 00:10:55 it confused my client :( 00:11:01 AnMaster: it prints that after it gives you the 1337 ascii art of "psybnc" when you start it up 00:11:13 mib_8oomfc, still I suggest you try several bouncers, including ezbounce and znc 00:11:33 I'll use this until I write my own bouncer... tomorrow. :-P 00:11:44 It occurs to me I could just start writing it now. 00:11:57 mib_8oomfc, as for znc starting I think it prints "starting znc\nloading modules\n forking into background" or something like that 00:12:00 a few lines of status 00:12:29 i shall call my bouncer bbbbounce 00:12:38 B4 for short. 00:12:46 mib_8oomfc, from the openbsd release song 00:12:48 for 3.9 00:12:50 I assume 00:12:52 pronounced "BU-BU-BU-BOUNCE" 00:12:55 AnMaster: er, no? 00:13:15 If you think I'd quote an openbsd release song you don't know me very well :P 00:13:17 mib_8oomfc, err they said "bu-bu-bu-bounce him on your knee" in that one 00:13:25 Eh. 00:13:27 so that is what everyone will think 00:13:36 sorry about that 00:13:43 Yes, because other people will know about this :-P 00:13:56 mib_8oomfc, 100% so far did :P 00:14:03 I meant, my bouncer 00:14:13 I'll be surprised if it leaves rutian/pastie.org 00:14:19 ah 00:14:35 -!- bsmntbombdood has joined. 00:16:09 mib_8oomfc, 0.6 mm screwhead rocks, yeah tiny indeed 00:16:18 lol. 00:16:25 yes there is one here 00:16:27 like that 00:16:28 heh, my bouncer will report server disconnections weirdly: 00:16:34 oh? 00:16:47 some nethack-like phrase? 00:16:56 :b4!b4@b4 PRIVMSG yourname :stuff here 00:17:05 where b4!b4@b4 doesn't actually exist, of course. 00:17:15 that would be seriously confusing 00:17:25 actually 00:17:28 that is what znc does 00:17:29 AnMaster: the reason it has ! and @ is to account for irc parsers 00:17:31 so they don't trip up 00:17:32 it messages from *status 00:17:34 saying: 00:17:34 actually, it'll be !n=b4 00:17:43 AnMaster: right, but *status has a reasonable host name there 00:17:44 and stuff 00:17:45 I assume. 00:17:48 and actually exists 00:17:50 (you can send to it) 00:17:59 *status!znc@znc.in 00:18:02 from thgat 00:18:03 that* 00:18:19 :*b4*!n=b4@b4 PRIVMSG :Disconnected from server, reconnecting... 00:18:33 sounds quite sane 00:18:45 it looks very similar to the znc one 00:18:56 I think it says: "Lost connection to server. Reconnecting..." 00:18:59 or something like that 00:19:17 don't remember on the top of my head (from the top? which is the English idiom?) 00:19:45 off the top 00:19:50 right 00:21:25 http://en.wikipedia.org/wiki/Poetic_License 00:21:26 I want to use this 00:22:01 in fact, I'm going to use it 00:22:19 I mean, why not? 00:22:25 "Aside from rhyming, the Poetic License is unique in its use of the first person, rather than passive legalese form, as well as the assurance of best effort. Many such licenses specifically distinguish between text and software, while the Poetic License may be applied to any work. Unlike other BSD-styled licenses, which explicitly require the copyright notice and 'this' notice to appear in all copies 00:22:25 of software and documentation, the Poetic License is vague as to the condition upon which these rights are granted. 'These rights, on this notice, rely' implies that the notice must remain in all copies, shared and/or modified." 00:22:25 hm 00:22:29 there may be issues there 00:22:41 Eh; I'm not planning to test this in court. 00:22:59 AnMaster: If the DO WHAT THE FUCK YOU WANT TO PUBLIC LICENSE, VERSION 2 has been testified to probably be valid in court by Debian lawyers, I'm sure this will to 00:23:03 1) assurance of best effort 2) the Poetic License is vague as to the condition upon which these rights are granted 00:23:07 (It has one clause, 1. You just DO WHAT THE FUCK YOU WANT TO) 00:23:22 mib_8oomfc, it has been found valid in a court? 00:23:23 heh ok 00:23:26 no 00:23:33 but the debian lawyers said they thought it would be valid 00:23:37 ah 00:23:40 which is pretty much all the gpl has had until recently too 00:26:08 haskell is delicious 00:27:18 *Scheme 00:27:27 no, haskell. 00:28:14 Haskell is bad. 00:28:56 yeah cuz monads are sooo complex 00:29:23 Also it has an ugly syntax. 00:29:31 * Slereah <3 parenthesises 00:29:45 Haskell's syntax is beautiful, maybe you're just bad at it. 00:30:16 I find infix more readable, especially with things like `isPrefixOf` 00:30:28 It's not about readability 00:30:32 It's about beauty :o 00:30:38 Haskell's syntax is beautiful 00:30:49 what you're saying is equivalent to 00:30:54 Slereah: scheme sucks, there's too many parentheses 00:30:57 i'd mess them up!! 00:31:00 uglyyyyyyyy 00:31:59 Don't diss scheme bro 00:32:41 Don't diss Haskell. 00:42:26 -!- mib_8oomfc has quit (Client Quit). 00:48:06 -!- oerjan has quit ("Good night"). 00:59:46 -!- Mony has quit ("Quit"). 01:36:14 -!- calamous has joined. 02:42:35 -!- calamous has quit ("Leaving"). 04:50:26 -!- Asztal has quit (Read error: 110 (Connection timed out)). 05:54:40 -!- Corun has joined. 06:13:35 -!- Corun has quit ("YES, NO W:ET."). 06:16:38 -!- bsmntbombdood has changed nick to bsmntbombgirl. 06:45:08 -!- moozilla has joined. 06:46:08 -!- moozilla has quit (Read error: 104 (Connection reset by peer)). 07:13:02 -!- Sgeo has quit (Read error: 110 (Connection timed out)). 07:24:23 -!- Corun has joined. 07:35:34 this shit is ridiculous 07:35:37 oops wrong window 07:43:07 -!- psygnisfive has joined. 07:59:59 -!- clog has quit (ended). 08:00:00 -!- clog has joined. 09:04:04 -!- Corun has quit ("This computer has gone to sleep"). 10:34:37 -!- oklopol has quit (Connection timed out). 10:54:31 -!- oerjan has joined. 12:05:42 -!- oerjan has quit ("leaving"). 12:10:35 -!- jix has joined. 13:25:18 -!- ais523|direct has joined. 14:35:28 heh 14:35:40 * ais523|direct wonders what AnMaster is hehing at 14:35:48 some stuff above 14:35:58 anyway hi ais523|direct 14:36:10 hi, and merry Christmas if it's today for you 14:36:12 it is for me 14:36:18 it was yesterday for me 14:36:26 we celebrate on the 24th here 14:36:44 yes, I know many countries celebrate on the 24th, wasn't sure whether you did or not 14:36:46 same in rest of Scandinavia 14:37:23 ais523|direct, seasons greetings to you 14:37:28 thanks 14:37:43 although in theory, I reckon you could get away with a season's greetings any time in Winter 14:37:46 * AnMaster is upgrading a remote freebsd server atm 14:37:53 although maybe it's an English idiom I don't really get either 14:38:01 ais523|direct, technically it could work at any point during the year 14:38:06 well, yes 14:38:07 just different seasons 14:38:12 I was thinking that too... 14:38:45 " although maybe it's an English idiom I don't really get either" <-- well it is the religious/customs neutral version basically 14:38:58 ah, maybe 14:39:09 equally meaningless in all traditions 14:39:17 oh yes and not offensive to anyone 14:39:20 just like the acronym UTC, which was chosen because it's wrong both in English and in French 14:39:30 hah indeed 14:40:33 incidentally, the hello world in Brainfuck I'm trying to debug atm is 929086 bytes 14:40:43 that's run-length encoded, too... 14:40:55 ais523|direct, wtf? 14:41:03 oh gcc-bf? 14:41:03 much bigger without, but that's partly because some of the pointer code has hundreds of thousands of >s in a row 14:41:04 right 14:41:05 and yes 14:41:20 there's a bug in my pointer handling that I know about, just haven't coded a fix yet 14:41:26 there are a lot more bugs I don't know about 14:41:48 ais523|direct, considering puts() fputs() and write() I did some tests recetly 14:41:52 recently* 14:41:57 anyway, the gcc-bf distribution, not counting the gcc or newlib sources, is 929086 bytes 14:42:03 um... 14:42:07 121507 14:42:09 copied the wrong number 14:42:10 someone claimed g++ -static generated much larger hello world than gcc -static 14:42:13 and that was true 14:42:26 until I added -nostdlib /usr/lib/libc.a 14:42:38 had to change to use write() too instead of puts() 14:42:48 and _exit(0); instead of return 0; 14:42:54 you can get gcc-bf programs a lot smaller by avoiding the stdlib 14:42:56 no idea why for the latter 14:42:59 I know why 14:43:02 it's to do with atexit 14:43:05 oh? 14:43:06 and the need to close open files when you use exit 14:43:09 ah 14:43:12 right 14:43:15 that means stdio needs to be linked in 14:43:22 gcc-bf's default runtime has an exit() in 14:43:38 which means it links stdio, that's why the hello world's so large 14:43:43 ais523|direct, anyway doing it that way got the size down to 6.6 KB for each. and *exactly the same binaries after strip* 14:43:46 from g++ and gcc 14:43:54 same md5sum 14:43:55 to get saner program sizes, I have a -naked option on the linker which doesn't include a limit 14:44:13 AnMaster: I'm not surprised, in theory C++ has no overhead compared to C if you don't use C++-specific features 14:44:27 on the other hand, good C style is bad C++ style 14:44:29 ais523|direct, yes I had to add -fno-exceptions to make it compile that way 14:44:36 for g++ 14:45:00 otherwise I got strange link time errors about symbols like __unwind_frame or such 14:45:03 don't remember details 14:45:16 anyway for bf stdio, that should be one area worth optimising a lot 14:45:24 yes 14:45:27 I already have a bf.h 14:45:30 maybe hand coding the stdio stuff to be as small as possible for gcc-bf 14:45:33 which only contains __bf_out and __bf_in atm 14:45:43 although I don't really need the double-underscores if it's in a dedicated header 14:45:47 ais523|direct, well it should be able to replace the newlib stdio and such 14:45:56 that's a project for later 14:46:04 anyway, here's a gcc-bf build command line: buildinto/bin/bf-gcc -Wl,-progress,-abi,-asm,-annotate,-map,-rle,-g,-trace tests/pointer.c --save-temps 14:46:17 that's due to me putting in all the debug options at once 14:46:19 because I bet that is one part that will save a lot if you replace 14:46:29 and yes, a new stdio would be nice, but difficult 14:46:30 -abi,-asm? 14:46:38 hm 14:46:38 -abi and -asm save two of the temporaries the linker uses 14:46:42 ah 14:46:44 makes sense 14:46:57 -map does just what it does in any other linker (although it often has a different name) 14:47:04 -progress shows progress bars, because it's slow 14:47:18 ais523|direct, why would a special bf stdio library optimized for size/speed be difficult? 14:47:26 I mean apart from bf always being a pain 14:47:26 because stdio itself is difficult 14:47:30 at least to pin down all the corner cases 14:47:36 I suppose you could have a nonconforming stdio-lite 14:47:43 ais523|direct, if I were to write it I would probably do something like C with inline bf 14:48:03 could be interesting 14:48:16 although there isn't all that much free tape atm 14:48:25 maybe parts could be pure bf even 14:48:35 well, most of gcc-bf doesn't generate pure bf 14:48:36 ais523|direct, you may need to increase memory then? 14:48:48 not really, I could just create extra memory 14:48:55 oh? 14:49:02 you said not a lot of free taoe 14:49:03 tape* 14:49:15 the problem's just that every tape cell is already used for something, well most of them 14:49:24 but I could just move everything to the right and use new cells at the left, for instance 14:50:14 hm? 14:50:15 what? 14:50:35 AnMaster: well, the BF tape is basically used to emulate a CPU 14:50:41 indeed 14:50:45 (the CPU was designed to be easy to emulate in BF) 14:50:51 hah 14:51:33 I also have a bf interp designed specifically for debugging gcc-bf 14:51:37 which can read its RLE output 14:51:39 hm nice 14:51:42 and reads its comments too 14:51:51 well, right, but can't you just do like the program does 14:51:52 so it knows quickly when something went wrong, and does a core-dump of the tape 14:51:53 malloc a block 14:51:56 and use it?? 14:52:01 yes, but mallocing's pretty slow 14:52:06 ah right 14:52:08 and besides, it involves pointers 14:52:10 and they're very slow 14:52:14 what about static buffers? 14:52:25 that doesn't need to involve pointers 14:52:35 you can have static buffers just fine 14:52:41 since you can pre-calculate where in the memory the buffer is 14:52:53 if you allocate a static buffer, gcc-bf will reserve every 6th element in a region of the tape (of its own choice) for it 14:52:56 you don't need any sort of indirection when accessing said buffer 14:52:57 and go there with > and < 14:53:03 directly 14:53:10 inded 14:53:10 so static buffers are pretty efficient in it 14:53:26 ais523|direct, that could be an issue if you don't know your exact current position? 14:53:30 say, in a recursive call 14:53:38 gcc-bf always knows its exact current position 14:53:47 really? hm ok 14:53:48 that's one of the main architectural choices I made in it 14:54:02 not only that, but -annotate writes the position as comments in the BF code 14:54:15 ais523|direct, how can it know that if the stack frame is of unkown size? 14:54:19 say due to using alloca() 14:54:28 it uses symbolic positions 14:54:34 hm ok 14:54:36 %sp means it's pointing to the top of stack, for instance 14:54:44 well I mean in the bf code 14:54:47 that is generated in the end 14:55:21 how can it know a fixed count of > or < needed to move to a static buffer and to get back 14:55:40 because it always moves via a particular location 14:55:47 ais523|direct, ah, and that is? 14:55:58 there's a dead zone of 6 cells, which is never used 14:56:02 at least, always 0 14:56:11 it always knows how to get to one of the cells in that range 14:56:16 (and by extension, any given cell in that range) 14:56:33 right, and this dead zone is used for scratch storage or? 14:56:38 during calculations 14:56:39 or 14:56:43 no, it's used for homing the pointer 14:56:51 many parts of memory are full of 1s 14:57:09 which means, for instance, starting at the stack pointer <<<<<[[<<<<<<]<<<<<<] will always land on a particular cell in the dead zone 14:57:27 ais523|direct, right 14:57:30 because every 6th cell contains a 1, apart from the start of a stack frame 14:57:41 and stack frames have to be at least 2 bytes, according to the processor ABI 14:57:53 ais523|direct, you have a copy of the stack frame every 6th byte!? 14:58:00 no 14:58:06 it wouldn't fit 14:58:10 what you mean then 14:58:19 the set of every 6th bytes together lets you determine where the stack frames are 14:58:27 they're 1 if a stack frame doesn't start there, or 0 if it does 14:58:55 ok 14:59:06 so you can easily search for it 14:59:07 right 14:59:09 yes 14:59:18 how do you know in what direction? 14:59:21 two 0s in a row mean you're either at top of stack (to the right) or in the dead zone (to the left) 14:59:34 and whenever the exact numerical location isn't known, we're to the right of the dead zone 14:59:39 to the left of that is just registers and temp cells 14:59:44 ah 15:00:18 so to move to a static area you just move to the dead zone then move a fixed number of cells from it? 15:00:22 yes 15:00:31 this makes me wonder how gcc-bf would handle buffer overflow 15:00:46 -!- cpu-jockey has quit ("leaving"). 15:00:47 it handles buffer overflow the same way as any other processor, it goes and overwrites memory 15:00:59 right, could it corrupt the internal state 15:01:04 weird stuff happens on buffer /underflow/, though, if it happened to be at the start of memory 15:01:08 -!- cruce has joined. 15:01:17 no, because the pointers are multiplied by 6 internally 15:01:19 also did I understand you right, for every 6 cells in memory there is one with actual program memory? 15:01:28 not exactly 15:01:33 no? 15:01:36 the program itself is compiled into BF, not into bytecode and interpreted 15:01:42 so the program is the program, it isn't stored on the tape anywhere 15:01:46 indeed 15:01:52 but I mean as in heap 15:01:53 every 6 cells, there are 2 which hold memory, one on the stack and one on the heap 15:01:54 or stack 15:01:59 ah 15:02:06 so stack and heap are interleaved!? 15:02:07 yes 15:02:14 and are separate segments? 15:02:15 heh 15:02:19 yes, separate segments 15:02:21 what does the pointers look like then 15:02:33 I mean, segment selector first or what? 15:02:42 0x00?????? for a function pointer, 0x01?????? for a stack pointer, 0x02?????? for a heap/static pointer 15:02:55 segment selector's in the MSB 15:02:55 hm interesting 15:03:10 byte or bit? 15:03:14 and the 4 bytes have weightings of 3, 65536*6, 256*6, 6 cells 15:03:19 they're byte pointers 15:03:23 hm right 15:03:41 I meant, MSB could be MS(Byte|Bit) 15:03:53 ah, ok, but it's byte in this case 15:03:58 the hex representations should have given it away 15:04:07 ah yes 15:04:23 ais523|direct, hm function pointers support jumping anywhere or? 15:04:28 yes 15:04:37 the entire program's basically in a massive switch statement, written in BF 15:04:44 that could probably be optimized away 15:04:48 at the end of any basic block, the address of the next basic block is set 15:04:56 and actually, that's very efficient with a decent interp and hardly uses any code 15:05:26 (the hyper-efficient way to end a program in gcc-bf is neither exit(0) nor _exit(0), it's goto *(void*)0;) 15:05:36 although the last is somewhat compiler-specific 15:05:40 and makes no sense in standard C 15:05:48 since you only need to jump to start of function, setjmp() calls, and for flow control inside functions (which could be a smaller switch for each function): labels, while, for, if and so on 15:06:21 being able to jump anywhere is pretty convenient, and it's not a computational order slowdown thing 15:06:42 it switches on the bottom 24 bits of the pointers, in blocks of 8 bits at a time 15:06:43 ais523|direct, I would assume it is undefined in C if you jump to any memory address that isn't a function pointer you got in a legal way 15:06:51 well, yes it is 15:07:01 but you'd still need to be able to jump to the address of any function 15:07:20 hm so each block, that means each line or each ; or what? 15:07:23 also, it would be nice if gcc-bf supported C++ too, and g++ does weird things with code pointers to handle exceptions 15:07:41 a block's a set of commands with linear control flow 15:07:53 as in, no jumping into or out of it 15:07:53 ah so you can't jump inside such a block? 15:08:02 no, nor call a function inside a block 15:08:04 but that makes me wonder 15:08:11 so they tend to start and end in the middle of statements 15:08:19 ais523|direct, ouch right 15:08:21 anyway 15:08:26 this makes me wonder a bit 15:08:27 can make it slightly confusing to debug 15:08:42 anyway, I have to go for a while, Christmas dinner 15:08:48