Category:Self-modifying

00:00:04 I also like Magic: the Gathering puzzles and Pokemon card puzzles, and I like to compose Pokemon card puzzles, including the one requiring retreating at least three times, knocking out your own cards, and so on. 00:01:03 I play chess puzzles too, as well as tsume shogi, too. 00:01:11 zzo38: I saw a player use Restoration Angel to flash a token at the GP 00:02:01 Can you give me the text of that card, and the details of the situation? 00:02:54 too much effort 00:05:30 can someone copy the list at http://esolangs.org/wiki/Category:Self-modifying and email it to me? 00:06:22 `fetch http://esolangs.org/wiki/Category:Self-modifying 00:06:25 2013-03-05 00:06:24 URL:http://esolangs.org/wiki/Category:Self-modifying [18477] -> "Category:Self-modifying" [1] 00:06:48 `mail 00:06:49 /home/hackbot/hackbot.hg/multibot_cmds/lib/limits: line 5: exec: mail: not found 00:06:51 coppro: Do you not remember? 00:07:19 -!- DHeadshot has joined. 00:09:10 I remember 00:09:17 I'm just too lazy. You can look up card texts yourself 00:09:27 @google mtg gatherer restoration angel 00:09:28 http://gatherer.wizards.com/pages/Card/Details.aspx?multiverseid=240096 00:09:29 Title: Restoration Angel (Avacyn Restored) - Gatherer - Magic: The Gathering 00:10:23 OK 00:10:39 What was the situation, however? 00:10:48 * oerjan wonders what kind of strange setup makes Arc_Koen unable to do the tranfer himself 00:11:01 *transfer 00:11:09 well the whole point is to compare what I see with what others see 00:11:20 ah 00:11:39 zzo38: there was a spell targeted at the token which he did not want to resolve 00:11:54 O, OK, yes, that would work. 00:12:10 `url Category:Self-modifying 00:12:13 http://codu.org/projects/hackbot/fshg/index.cgi/raw-file/tip/Category%3ASelf-modifying 00:12:35 error: Category:Self-modifying@f368e91512e6: not found in manifest 00:12:38 I like that kinds of ideas. 00:12:43 hmm yes definitely different from what I see :p 00:12:44 hmph 00:12:53 `cat "Category:Self-modifying" 00:12:55 cat: "Category:Self-modifying": No such file or directory 00:13:03 `run cat 'Category:Self-modifying' 00:13:04 cat: Category:Self-modifying: No such file or directory 00:13:13 ho hum. 00:13:24 `undo 2374 00:13:29 patching file Category:Self-modifying 00:13:34 try now :P 00:13:36 I deletedetedeted it 00:13:51 thank you! 00:14:01 oh nooooo 00:14:01 It says "that card" but the first part doesn't say it has to be a card. 00:14:05 `run cat 'Category:Self-modifying' 00:14:07 \ \ \ Category:Self-modifying - Esolang \ \ \ \ Under the current rules, does the token come back, though? 00:15:00 well it appears to be exactly what I see 00:16:01 `? ais523 00:16:02 Agent “Iä” Smith is an alien with a strange allergy to avian body covering, which he is trying to retroactively prevent from ever evolving. On the 3rd of March, he's lawful good. 00:16:12 ah 00:16:47 Arc_Koen: but what if your browser has a "trusting trust" problem, what then! 00:16:55 haha 00:17:29 "Warning: I cannot be trusted to display this webpage correctly. Display webpage?" 00:18:01 so how does mediawiki handle category page? for instance if you add a page to a category, when is the category page updated? 00:18:14 "eventually" 00:18:27 i thought it was pretty quick... 00:18:37 on esolang, anyway 00:18:50 I remembered you told me that 00:19:00 Jafet: wtf are you doing to rainwords :P 00:19:04 If it is a SQL query, it should do immediately if there isn't a cache of the query. 00:19:42 coppro: Do you know the answer to my rules question? They change the rules so I don't know what it is 00:20:42 oerjan: making it work on bin/rainwords 00:20:52 `run rainwords $(which rainwords) 00:21:08 -!- fungot has quit (Ping timeout: 248 seconds). 00:21:21 YOU KILLED FUNGOT 00:21:23 No output. 00:21:38 i sense problems. 00:22:08 also exactly how didn't it work before... 00:23:08 `ls /usr/bin/python 00:23:10 /usr/bin/python 00:23:10 raw_input() reads one line 00:23:18 (I think we learned this recently) 00:23:23 oh. i was wondering more about the r thing 00:23:32 maybe that's not new 00:23:40 ooookay 00:23:50 `which python 00:23:52 /opt/python27/bin/python 00:24:07 `/usr/bin/python --versino 00:24:09 Unknown option: -- \ usage: /usr/bin/python [option] ... [-c cmd | -m mod | file | -] [arg] ... \ Try `python -h' for more information. 00:24:10 `/usr/bin/python --version 00:24:10 I just used my magic powers to update the category page, now it has 82 languages listed 00:24:11 Python 2.6.6 00:24:35 `run python -c 'r=[4,7,8,9,2,13,6]; print " ".join(chr(3) + c + ":-)" for c in r)' 00:24:37 Traceback (most recent call last): \ File "", line 1, in \ File "", line 1, in \ TypeError: cannot concatenate 'str' and 'int' objects 00:24:45 `run python -c 'r=[4,7,8,9,2,13,6]; print " ".join(chr(3) + str(c) + ":-)" for c in r)' 00:24:46 :-) :-) :-) :-) :-) :-) :-) 00:25:01 13 isn't very indigo, unfortunately 00:26:16 oh it's actually rainbow order :P 00:26:43 `rainwords What does this do then 00:27:01 i think there is a problem yes 00:27:14 No output. 00:27:59 `cat bin/rainwords 00:28:01 #!/usr/bin/python \ import random; w=[l.split() for l in open("/dev/stdin").read().split("\n")]; r=[4,7,8,9,2,13,6]; print "\n".join((lambda s: " ".join(chr(3) + str(r[(i+s)%len(r)]) + l[i] for i in range(len(l))))(random.randrange(0, len(r))) for l in w) 00:28:14 oh wait 00:28:22 -!- sirdancealot has quit (Ping timeout: 250 seconds). 00:28:33 `run echo I think I remember the issue | rainwords 00:28:35 I think I remember the issue 00:28:38 I'm kind of confused as to how regular expressions are a single production rule (in the formal grammar sense) 00:28:48 `run yes | rainwords 00:29:00 hum maybe not ideal 00:29:03 kallisti_: ac*b generates ab, acb, accb, acccb, etc 00:29:09 Traceback (most recent call last): \ File "/hackenv/bin/rainwords", line 2, in \ import random; w=[l.split() for l in open("/dev/stdin").read().split("\n")]; r=[4,7,8,9,2,13,6]; print "\n".join((lambda s: " ".join(chr(3) + str(r[(i+s)%len(r)]) + l[i] for i in range(len(l))))(random.randrange(0, len(r))) for l in w) \ MemoryError 00:29:13 Bike: no I understand that 00:29:16 oerjan: what do you think this is, haskell?? 00:29:24 Take your infinite streams and begone 00:29:26 `run yes | head -50 | rainwords 00:29:28 y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y \ y 00:29:41 oh wait maybe I misunderstand what a regular grammar is 00:29:44 to Wikipedia 00:29:45 -!- Frooxius has quit (Ping timeout: 240 seconds). 00:29:46 !!! 00:29:59 it's just alternation and kleene star 00:30:07 oh and concatenation obv 00:30:14 regular expressions? yes 00:30:20 I'm trying to find the link between regular expressions and regular grammars 00:30:21 basically 00:30:47 oh I think I misread 00:30:48 "In regular grammars, the left hand side is again only a single nonterminal symbol, but now the right-hand side is also restricted. The right side may be the empty string, or a single terminal symbol, or a single terminal symbol followed by a nonterminal symbol, but nothing else." 00:31:06 so a regular grammar can have multiple production rules right? 00:31:13 yeah 00:31:16 okay 00:31:24 I misread and thought that it meant the grammar had only one production rule 00:31:28 which seemed... uh, strange. 00:31:31 http://upload.wikimedia.org/math/5/6/e/56e5e8888abe83d0f4f731bde2611d72.png wikipedia's example 00:31:40 `run ghc -e 'putStr $ replicate 100 "_,.-~^~-.,"' | rainwords 00:31:40 (for a+b+ in regex syntax) 00:31:44 \ :1:24: \ Couldn't match expected type `Char' with actual type `[Char]' \ In the second argument of `replicate', namely `"_,.-~^~-.,"' \ In the second argument of `($)', namely \ `replicate 100 "_,.-~^~-.,"' \ In the expression: putStr $ replicate 100 "_,.-~^~-.," 00:31:47 ...that didn't work 00:31:53 http://en.wikipedia.org/wiki/Formal_grammar#Regular_grammars that. 00:32:07 Jafet: concat/join 00:32:21 `run ghc -e 'putStr $ take 999 $ cycle "_,.-~^~-.,"' | colorize 00:32:26 _,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.- 00:32:37 `run ghc -e 'putStr $ take 200 $ cycle "_,.-~^~-.,"' | colorize 00:32:42 _,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-., 00:32:50 ah okay, I simply misread what a regular grammar is. 00:33:20 `run python -c 'print("_,.-~^~-.," * 100)" | colorize 00:33:21 bash: -c: line 0: unexpected EOF while looking for matching `'' \ bash: -c: line 1: syntax error: unexpected end of file 00:33:22 `run yes ":D " | head -50 | tr -d '\n' | rainwords 00:33:24 :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D :D 00:33:27 `run python -c 'print("_,.-~^~-.," * 100)"' | colorize 00:33:29 File "", line 1 \ print("_,.-~^~-.," * 100)" \ ^ \ SyntaxError: EOL while scanning string literal \ Traceback (most recent call last): \ File "/hackenv/bin/colorize", line 4, in \ w=raw_input() \ EOFError: EOF when reading a line 00:33:32 so this concept for an esolang I have 00:33:33 uuuuugh 00:33:38 is essentially a context-sensitive grammar 00:33:44 `run python -c 'print("_,.-~^~-.," * 100)' | colorize 00:33:45 where the production rules are operating on themselves 00:33:46 _,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.-~^~-.,_,.- 00:33:51 whatever. 00:34:25 `tcc 00:34:26 /home/hackbot/hackbot.hg/multibot_cmds/lib/limits: line 5: exec: tcc: not found 00:34:34 What 00:34:35 `gcc 00:34:36 gcc: no input files 00:34:43 `icc 00:34:45 /home/hackbot/hackbot.hg/multibot_cmds/lib/limits: line 5: exec: icc: not found 00:34:50 `clang 00:34:51 /home/hackbot/hackbot.hg/multibot_cmds/lib/limits: line 5: exec: clang: not found 00:34:59 `fasm 00:35:00 /home/hackbot/hackbot.hg/multibot_cmds/lib/limits: line 5: exec: fasm: not found 00:35:07 `nasm 00:35:09 /home/hackbot/hackbot.hg/multibot_cmds/lib/limits: line 5: exec: nasm: not found 00:35:12 `g++ 00:35:14 g++: no input files 00:35:17 `as 00:35:35 hm is there an esolang based on linker command language 00:35:39 -!- sebbu has joined. 00:35:39 -!- sebbu has quit (Changing host). 00:35:39 -!- sebbu has joined. 00:36:08 Bike: that seems redundant 00:36:20 No output. 00:36:21 Is there an esolang that randomly picks other esolangs to try running your code in? 00:36:40 Bike: colorize doesn't add colors beyond hackego's output limit of 350 bytes 00:36:48 boooooring 00:36:58 -!- augur has joined. 00:37:19 -!- Frooxius has joined. 00:37:31 Rainbows all the way across #esoteric 00:38:30 Fabulous. 00:40:01 Is coppro even on anymore? 00:40:31 -!- TeruFSX has joined. 00:46:46 -!- sirdancealo2 has joined. 00:51:37 Is there a way in Linux to access the text mode VRAM without root? 00:52:29 Furthermore, can you capture the text mode VRAM of a different virtual console in this way? 00:56:15 chmod /dev/mem and then you can :) 00:56:33 I don't want to access all of the memory though. 00:58:33 Can you put ACLs on /dev/mem 00:58:48 "there could be a legit use for this" 00:58:51 you can write a setuid helper for reading parts of /dev/mem 00:59:10 over a UNIX socket with SCM_CREDS 00:59:18 bad ideas itt 00:59:34 What would be the addresses of such screens anyways (also based on the numbers of rows and columns)? 00:59:51 dunno, you may have to source-dive the linux kernel 01:00:06 Source-dive the linux documentation 01:00:25 -!- doesthiswork has joined. 01:00:25 Are they the standard PC text mode formats? 01:01:14 I was thinking to have the program which writes the text VRAM of one virtual console to a file, with a header specifying the width/height. 01:01:37 Doing this is MZM layer mode. 01:03:10 Some people may want to have a ANSI screenshot of Linux screens, so you could then use MZM->ANSI to do that. 01:03:30 Does the linux console support 256 colors 01:03:54 As far as I know it is PC text mode. 01:07:53 I thought it used the framebuffer 01:08:12 can do either 01:08:14 I think it supports that too 01:08:57 I suppose you would need the framebuffer or other things though, if you are running Linux in the computer which is not PC. 01:11:16 You would need the framebuffer unless you enjoy running htop in 80x25 or whatever the VGA text mode is 01:12:31 i enjoy it very much 01:12:33 it's so easy 01:14:42 The default text mode is 80x25, but it seems to be able to run in various other sizes too. 01:20:13 Hrmph. Still no new OOTS 01:22:22 oots is the worst campaign ever 01:22:33 Who runs a campaign for ten years 01:24:26 How long do *you* run a champagne^Wcampaign? 01:25:00 Nobody lets me run a campaign 01:25:09 I'm not sure why 01:27:22 -!- DHeadshot has quit (Read error: Connection reset by peer). 01:27:28 -!- DH____ has joined. 01:27:30 Then you must learn. 01:40:34 -!- GOMADWarrior has joined. 01:42:08 "In May 2006, Intel announced that 80386 production would stop at the end of September 2007." 01:48:34 -!- Jafet has quit (Quit: Leaving.). 02:06:49 -!- Phantom_Hoover has quit (Remote host closed the connection). 02:25:20 In Dungeons&Dragons game, some people forced us to give them much of our things; much of which is not important but some are, so we need to get it back. 02:25:52 It seems they don't understand magic very well, so if we can somehow convince them that the amulet of anti-magic is actually a amulet of illusions? (note: My character is currently invisible) 02:27:06 -!- GOMADWarrior has quit (Ping timeout: 264 seconds). 02:27:57 Or if I find the single one, alone, which has the item we need, then I could safely stun them even though becoming visible, if I can surprise them. 02:28:46 I do think of the possibility that they are just trying to make us think they don't know magic and actually they do, but I find this not likely at this time. Nevertheless, I must consider all of the possibilities. 02:33:19 -!- oerjan has quit (Quit: leaving). 02:33:38 -!- Arc_Koen has quit (Quit: leaving as well). 02:48:56 -!- oerjan has joined. 02:49:13 `olist 02:49:17 olist: shachaf oerjan Sgeo 02:50:51 Did I missed something else? 02:51:14 oerjan, ty 02:51:24 And I take it you got your cache issues sorted out? 02:51:40 no, i just went to the right number directly 03:08:06 -!- madbr has joined. 03:08:13 man 03:08:22 the 65816 is so impossible to pipeline 03:08:28 it's like it's designed against it 03:13:06 @hoogle m (x -> y) -> x -> m y 03:13:07 Control.Applicative (<*>) :: Applicative f => f (a -> b) -> f a -> f b 03:13:07 Control.Monad ap :: Monad m => m (a -> b) -> m a -> m b 03:13:07 Control.Applicative (<**>) :: Applicative f => f a -> f (a -> b) -> f b 03:13:23 :t (??) 03:13:25 Functor f => f (a -> b) -> a -> f b 03:13:48 @hoogle m (x -> m y) -> x -> m y 03:13:49 Prelude (=<<) :: Monad m => (a -> m b) -> m a -> m b 03:13:49 Control.Monad (=<<) :: Monad m => (a -> m b) -> m a -> m b 03:13:49 Prelude (>>=) :: Monad m => m a -> (a -> m b) -> m b 03:17:53 thoerjan 03:32:46 -!- monqy has joined. 03:45:42 zzo38: it is a card because once in exile, it is just a card 04:04:58 -!- abumirqaan has quit (Read error: Operation timed out). 04:07:06 -!- iamcal_ has quit (Ping timeout: 245 seconds). 04:07:56 -!- surma has quit (Ping timeout: 256 seconds). 04:08:00 -!- ssue has quit (Ping timeout: 264 seconds). 04:08:11 -!- jix has quit (Ping timeout: 255 seconds). 04:13:22 -!- jix has joined. 04:15:59 -!- ssue_ has joined. 04:20:41 -!- abumirqaan has joined. 04:34:14 -!- FireFly has quit (Excess Flood). 04:34:31 -!- FireFly has joined. 04:38:31 -!- monqy has quit (Quit: hello). 05:08:00 coppro: I thought it is a token? 05:37:51 -!- oerjan has quit (Quit: leaving). 05:47:07 -!- iamcal_ has joined. 05:52:21 -!- monqy has joined. 05:55:13 -!- TeruFSX has quit (Ping timeout: 248 seconds). 06:01:57 -!- surma has joined. 06:20:15 -!- nooodl has joined. 06:31:25 zzo38: tokens cannot be returned to the battlefield once exiled 06:31:32 so that sentence wouldn't apply to them anyway 06:37:35 coppro: Yes it is what I thought might be the case 06:38:15 Do they cease to exist due to lack of initial state? How do the new rules work with this? 06:40:23 -!- mtve has quit (Ping timeout: 255 seconds). 06:40:49 zzo38 : got any interest in synthesizer design? 06:41:17 madbr: Some. 06:48:18 Can you make a Pokemon card puzzle which involves retreating more than three times? What is the maximum number of times you can make it have? There is win in 1, win in 2, win in 3, even as much as win in thirty, can there be made a win in sixty? 06:49:54 Or make one involving a not 100% win but maximized chances? 06:55:07 madbr: What was your ask of interest in synthesizer design? 06:58:02 writing a synth atm 06:58:06 vst 06:58:58 zzo38: a token, once it has left the battlefield, cannot again change zones 06:59:08 it ceases to exist the next time state-based actions are performed 06:59:10 hhh38 07:01:42 coppro: Yes, those are the rules I thought but there are many things I dislike about the Magic: the Gathering rules that I think are very klugy; I don't like the rule of removing auras that are also creatures, for example, as well as other things. But, OK, now at least I can know what those rules are. 07:02:04 madbr: What synth are you writing? 07:02:13 physical model 07:03:48 I always use Csound for this kinds of stuff, it has physical models. But what specific physical model are you making? Maybe it is one I don't have. 07:05:03 going to have a couple "generic" models 07:05:12 one for wind instruments 07:05:26 and one for plucked strings with a variant for percussion 07:05:44 O, I already have things like that. 07:05:59 Still, I suppose yours might be useful for using with VST, I guess. 07:06:27 yeah but the ones in STK suck 07:06:45 dunno about the csound ones but they're probably not very good 07:07:16 O, OK, then. Are you able to license the files under the LGPL v2.1 and later version? 07:07:29 what files 07:09:04 The files to implement the physical model; in case it works better than the one in Csound then I can port it to Csound. 07:09:30 Actually Csound has a few generic models for wind instruments and plucked strings; one of the plucked string models is of a good quality. 07:11:30 Csound uses the LGPL v2.1 license, so my own plugins to Csound use the same license so that they could include it in the main program if they wanted to. 07:12:29 does that imply that the other plucked string models are bad 07:13:44 Not necessarily; it could also be that I configured them incorrectly or have bias due to other reasons, still, it can be useful for certain sounds nevertheless. 07:14:18 I just mean they do not seem to be as good as the specific one that I found to be much better than the others. 07:14:52 or maybe they just sound bad 07:15:00 like most physical modeling stuff 07:15:45 Yes. 07:17:16 Still, one of them is not bad, specifically "wgpluck" seems good. 07:22:17 I have gotten some special effects by using percussion sounds with waveguides to make a new sound 07:30:49 * madbr looks it up 07:31:04 doesn't look like it has inharmonicity 07:32:09 You might be correct. 07:37:09 -!- epicmonkey has joined. 07:40:05 How does a mark-and-sweep gc know which "objects are still known"? 07:40:08 i.e 07:40:36 root set? 07:40:45 void foo() { char* b =alloc .. char* a= alloc... bar(b); ...} void bar(b) { doStuff } 07:40:56 if my gc kicks in while b is currently in doStuff 07:41:06 how does he know that he should not delete b and a yet? 07:41:27 and how does he know he can delete them after foo 07:41:29 You mean a conservative GC? 07:41:38 usually you'd add the stack frame to the root set? 07:41:44 It'll just scan the stack/registers/etc. for pointers. 07:42:15 Yeah but since I'm compiling to C there might be other stuff on the stack too 07:42:28 I could pass stuff on an other stack structure 07:42:39 and then just scan the stack 07:42:42 that'd be the easiest thing 07:43:10 instead of compiling to bar(b); I would compile to push(someStack,b); bar(); 07:43:17 and bar pops arguments from someStack 07:43:24 but I'm not sure if there's not better way 07:44:09 what "other stuff" is there on the stack? 07:44:21 C stuff. 07:44:31 Like, return addresses? 07:44:44 somehow i doubt it's going to be a problem to have return addresses in the root set 07:44:47 Yes, but that would not be the problem. 07:44:53 since that's predictable. 07:46:55 You can do inline C stuff i.e you can write a function in my language and embed C in it 07:47:25 so? 07:47:31 That's unpredictable. 07:47:47 if somebody allocates something on the stack there 07:47:57 like 07:48:01 {{ int c; }} 07:48:30 I don't see the problem? The root set is what's not being collected, you can be greedy. 07:48:41 And like shachaf said it's gonna be conservative. 07:48:49 What root set? 07:49:08 The root set. The set of things you tell mark and sweep not to collect. 07:49:45 i.e. the "objects still known". 07:49:51 yes 07:49:56 but how do I know which objects I know. 07:50:20 Everything pointed to by the stack, and pointed to by that, etc 07:50:29 The stack is not in my control. 07:50:48 course not. 07:51:15 unless I use my own stack 07:51:29 -!- epicmonkey has quit (Ping timeout: 248 seconds). 07:52:22 I can't scan the c stack for pointers 07:52:30 how would I even know what's a pointer and what not 07:52:51 and how would I know which are mine and which are not my pointers 07:53:06 You don't. You assume everything is a pointer. it's okay to overestimate. 07:53:39 Oh. Ic. 07:54:17 but that would require that all pointers are aligned 07:54:17 If you want an exact GC you can use tags or do something funny like put all integers in some particular page, but, you're 07:54:20 doing C. 07:54:33 4byte ptr, 1byte char , 4byte ptr 07:54:43 ^- that'd mess things up big time if I do a ptr sweep 07:55:24 I thought the stack would usually be word-aligned anyway... 07:55:41 doesn't need to be 07:55:54 isn't it required by most ABIs...? 07:56:13 ... actually isn't it generally required that data types are aligned to their natural alignment? 07:56:17 like, on the stack or not 07:56:26 unaligned data is invalid C, I think 07:56:36 -!- heroux has quit (Ping timeout: 264 seconds). 07:57:15 The natural alignment for all my data types is Chaotic Evil. 07:57:20 That's what I thought. But if tha'ts not the case, maybe you could store stack alloc data along with compiled functions. 07:57:31 *giggle* 07:57:36 yeah normally everything should be aligned 07:57:42 the alignment for all my data types is neutral good 07:57:44 except long double 07:57:47 which is chaotic neutral 07:57:54 since most processors slap on massive alignment penalties on you if you don't 07:57:54 Actually maybe you could do that anyway and then you'd really know what's a pointer... hm 07:58:05 -!- heroux has joined. 07:58:11 except the lastest ridiculous intel chips but that's a detail 07:58:22 most? I think most non-x86 just outright segfaults ^^; 07:58:29 sure. miss alignment decreases performance 07:58:50 mroman: normally you don't even want to allow misalignment on a cpu 07:58:58 I guess a kernal trap counts as decreased performance.... XD 07:58:58 I know. 07:59:00 complexifies the design like mad 07:59:01 But still, you can do that. 07:59:05 ARM just gets rotated values for misaligned loads, doesn't it? 07:59:11 only old arms 07:59:22 newer arms have like 8 cycle penalty 07:59:33 I think you still have to use uanligned load instructions explicitly 07:59:34 otherwise it kabooms 07:59:45 how would you do that? i thought int x; char y; int z; would put three words on the stack (assuming a int is a word) 08:00:02 Bike: Yes. 08:00:13 -!- carado has joined. 08:00:14 So... 08:00:18 but you can still mess things up by pushing manually a single byte to the stack 08:00:19 I'm not sure how different compilers align the stack 08:00:30 You can do that in C? 08:00:31 but often you need 8 or even 16 byte alignment 08:00:43 I don't think you can even do that in asm 08:00:48 there's no "push byte" instruction in x86 <.< 08:00:51 I just, why would you even /want/ to do that. 08:01:04 fiora: aren't there a zillion size prefixes? 08:01:05 Fiora: there is 08:01:05 I mean of course you could like, manually move a byte to the stack and decrement it 08:01:14 PUSH reg16 ; o16 50+r [8086] 08:01:15 PUSH reg32 ; o32 50+r [386] 08:01:18 or you could just manually overwrite sp 08:01:24 "The operand-size attribute of the instruction determines whether the stack pointer is decremented by 2 or 4: this means that segment register pushes in BITS 32 mode will push 4 bytes on the stack, of which the upper two are undefined. If you need to override that, you can use an o16 or o32 prefix. " 08:01:38 or maybe that wasn't x86 08:01:46 there's no o8 prefix, so, I don't think push even supports it. which I guess makes sense because x86 never had <16-bit pointers 08:02:04 I'm thinking maybe if you're going to be this cruel to your computer maybe the GC should just break their knees. 08:02:09 XD 08:02:12 yeah but x86 is horrible so I was half expecting a push byte instruction :D 08:02:23 possibly even a 1-byte opcode for doing it 08:02:26 there's a "push imm8", but it pushes with the current bitness (or operand prefix bitness) 08:02:39 so it's just a nice short shortcut for pushing small numbers like zeroes 08:02:43 without using an imm32 <.< 08:02:54 mhm 08:03:04 There's also no push reg32 in long mode, while you can still do a push reg16 with the o16 prefix. 08:03:11 there's no point except in function call stuff 08:03:37 which is what stack instructions are for anyways, I guess 08:03:43 (So you can push 2 bytes, or 8, but not 4.) 08:03:52 fizzie: huh, did they do that to free up opcode space? 08:04:02 that's really weird 08:04:11 The manual doesn't give reasons, just facts. :p 08:04:20 True XD 08:04:21 ~arch design~ 08:04:34 I still love some of the really silly incongruencies 08:04:52 like the fact that nop is "xchg eax, eax" in 32-bit mode 08:04:56 but in 64-bit mode, they're not the same thing 08:05:06 because xchg eax, eax isn't a nop 08:05:12 that's because you done have to do the horrible pipeline design that comes with that :D 08:05:20 Fiora: Possibly they did it so that you don't need a REX.W prefix for every PUSH, though? 08:05:40 fizzie: couldn't they make it default to 64-bit and 32-bit requires the address size prefix? 08:05:41 isn't there one of the RISCs where mov is really OR rd, rs, rs 08:05:46 I remember there were some instructions that did that? 08:06:04 or something like OR rd, rs, rzero 08:06:04 Fiora: I don't remember any examples, but I guess that's possible. 08:06:08