00:30:08 -!- ais523 has quit (""Going home""). 00:35:54 -!- sebbu has quit ("je re"). 00:41:28 -!- timotiis has quit (Remote closed the connection). 00:43:39 -!- slereah_ has joined. 00:44:38 -!- sebbu has joined. 00:46:52 -!- Slereah has quit (Read error: 104 (Connection reset by peer)). 01:12:51 -!- oerjan has quit ("Good night"). 01:25:46 -!- calamari has joined. 01:30:16 -!- BMeph69 has changed nick to BMeph. 01:45:33 -!- Tritonio_ has joined. 01:45:41 -!- Tritonio__ has joined. 01:45:49 -!- Tritonio_ has quit (Read error: 104 (Connection reset by peer)). 01:45:57 -!- Tritonio__ has quit (Read error: 104 (Connection reset by peer)). 01:46:13 -!- Tritonio_ has joined. 02:01:59 -!- RodgerTheGreat has joined. 02:08:12 Ugh, I just realized that if I do in fact make this language, I am going to make my own profiler since it cannot use C calling convention. Sigh. 02:08:40 -!- ehird has quit ("Konversation terminated!"). 02:08:44 Ugh. 02:08:59 Ugh. 02:09:23 * Sgeo has learned a LOT about DNS today 02:11:03 today, i pood, ate and coded java 02:11:19 (you figure out the parse tree.) 02:11:20 Heh, 02:11:35 I assume coding java and pooing are a single process. 02:12:17 cute, very cute 02:12:24 Eh, that doesn't make as much sense as it did in my head. 02:12:32 I apologize. 02:12:39 :D 02:13:01 sometimes i take the comp with me to the toilet. 02:13:13 if you have to ask 02:14:01 Well I was trying to make a simile between eating poo and coding java. 02:17:37 :D 02:17:50 i've made that multiple times 02:18:03 i mean 02:18:04 that simile 02:18:29 Except the words came out in senseless babble. 02:18:53 I guess that is the result of spending too much time writing a language spec. 02:19:16 My typing system is turning into a clone of C. 02:19:26 Pointers and such are exactly the same. 02:19:57 The only additional type I expect is a closure. 02:20:58 But every other aspect of the language is in no way similar to C. 02:20:59 At all. 02:22:15 is it esolicious? 02:22:57 What does that mean? 02:24:50 :) 02:25:01 ? 02:25:04 do you considered it esoteric? 02:25:24 Yes. 02:25:34 eso- is something you add when something is esoteric, -licious is just something you add when something is hot and steamy 02:25:49 the latter can also be a nop, depends on the situation 02:25:51 Heh. 02:30:50 I am trying to decide if I should even implement closures. 02:30:52 They are so slow. 02:31:24 And it isn't that much work to just create a function. 02:31:38 But it makes it even more confusing. 02:34:04 what kinda closures? 02:34:28 There is only one type of closure, no? 02:38:44 umm, closure = function copies surrounding namespace? 02:38:47 The US Government uses ZoneEdit? 02:39:09 oklofok: Basically. 02:39:46 Deformative: and that is slow because? 02:40:11 You need to stick all the local variables on the heap somewhere. 02:40:14 you just compile that into giving the params explicitly instead of taking them from the surroundings. 02:40:46 Yeah, but you need to malloc every single time you pass a closure. 02:40:57 And in my language they could EASILY become overused. 02:42:25 The user would likely end up creating a closure for the true and false of every single if statement. 02:42:34 And that is bad. 02:44:26 Plus it would massively uglify the syntax. 02:45:05 -!- Tritonio_ has quit (Remote closed the connection). 02:45:36 I am also trying to decide to have anonymous functions or not. 02:45:43 It is easily implemented. 02:45:46 But ugly. 02:50:38 anonymous functions are sex 02:50:48 you have them, and people will like you 02:53:10 Heh. 02:53:16 Yeah, I will implement those. 02:53:34 But I think closures are a nono. 02:54:43 Should I use <> or [] for function definitions? 02:54:59 I think <> so that i could use [] for arrays. 02:55:35 Even thought vice versa looks better, [] for arrays will conform with all other languages. 02:56:52 Then I will use {} for structures. 02:56:56 And I do not know what for comments. 03:03:34 <> is cute 03:03:56 i use stuff with hyphens 03:04:04 <-- --> for oklotalk, -- for most 03:04:49 -- is from haskell, <-- --> is my own invention based on haskell + nopol, but it is also a html comment perhaps 03:05:00 not sure, i've never used comments in my html :) 03:07:18 Well I cannot use <> that is function definition. 03:07:45 I guess I could use /* */ but it is so difficult to type. 03:10:29 Heh, if I could make everything with more than 2 lines of indentation commented. 03:10:37 Because there will never be more than one. 03:10:38 Ever. 03:10:40 Which is nice. 03:11:34 s/if I could/if I was stupid I could 03:13:39 nono 03:13:44 make one indentation comment 03:13:47 0 and 2 are program! 03:13:59 xD 03:18:13 Maybe ~comments~ 03:18:48 Or `comments` 03:18:59 Eh, the ` char is too small. 03:19:16 Not enough characters on the keyboard, seriously. 03:33:27 that's why they invented tokens 03:53:07 Hmm. ~comments~ would logically be not comments. :p 03:55:58 heh 03:56:05 "...colorForth can not help but be the one operating system in the world capable of being totally open and at the same time the most quickly understandable by children." 03:56:40 "...if you realize that this is possibly a once in a life time chance to make computing easier for the entire world in one project..." 03:56:49 (somebody wants to put colorForth on OLPC) 04:03:59 it's perhaps the most insane page i've ever seen. http://colorforthray.info/ 04:20:42 I'm having difficulty understanding why colors are necessary or useful in FORTH. 04:23:29 and lament- if that's the most insane page you've ever seen, you should investigate http://www.timecube.com/ 04:39:49 I like colorforth. 04:39:57 Could be better though. 05:04:33 That reminds me. . . 05:04:46 cherez: Going to do more of your dramatic readings of Time Cube? 05:06:25 Wow. The Time Cube stuff has even gotten worse recently. . . 05:07:14 The craziness goes up to eleven now. 05:09:11 the TimeCube page gets updated? 05:10:13 Yeah. 05:10:18 "Believer is far more EVIL than a False God, for Google 05:10:19 cut back my Site from 34,000,000 to 4,000,000 in 1 night 05:10:19 for the above Statement. 1 Day1God exists only as Evil. 05:10:19 I thought Google was free of such evil bias, predjudice 05:10:19 and shenanigans that block real truth from being known. 05:10:19 Once before, Google cut back my site from 89,000,000 to 05:10:21 34,000,000 in a single act for something I said, that/s Evil" 05:10:23 sorry 05:10:30 WTF is that supposed to MEAN? 05:10:41 He's now discussing +0- antipodes. . . 05:10:43 WTF is that? 05:11:18 http://uncyclopedia.org/wiki/Slime_Cube :D 05:11:39 BTW, a slime cube with a bunch of potions in it? *Dangerous*. 05:18:54 heh 05:19:23 I always figured it would digest any potions spilled within, but that could be interesting... 05:20:05 Potions activating at random. 05:20:08 *shudder* 05:24:04 knowing my campaigns, a couple random potions are instant, bottled hilarity 05:24:25 I was always proud of the "plaid potion". 05:26:24 ... Plaid potion? 05:26:33 Do I *want* to know what that does? 05:26:58 a particularly deranged alchemist made most of my players fear potions with odd, misspelled labels like the plague 05:27:05 pikhq: it does what it sounds like 05:27:59 ... it turns stuff plaid. 05:52:20 LMAO 05:52:45 * pikhq needs to know more of this campaign 05:53:35 I ran my D&D games very fast and loose, and there was a lot of comedy and running gags 05:54:25 "Melf" was a dangerously inept wizard/alchemist/probably pothead that made the craziest shit 05:55:23 sometimes my notes would be just a sheet of paper with "there are some caves, here's a random encounter table and here's a list of the insane magic items the players will find" 05:55:58 we could spend hours with them just figuring out what stuff did, fighting some goblins and solving a puzzle room or two 05:57:39 Even in the more serious adventures, it wasn't unusual to run across some "Melf's Acid Apples" or a deck of "Stinky Pete's Cards of Fortune" 06:00:09 the "bag-O-charisma" was a fun one. Strangle someone with it, and it steals a portion of their CHA, and then transfers it to the next person that dons the bag, temporarily. It took players a while to figure it out, and then caused trouble because nobody recognized the players afterwards until the effects wore off. 06:00:56 Fortunately one player, thinking quickly, beat himself in the face with a rock until he was ugly enough for the townspeople to recognize 06:01:21 good times, good times 06:04:56 y'know, I think if I could find my old notes to refresh my memory, I could make some pretty funny comics about our exploits back in the day 06:07:52 My God. 06:08:01 Hmm. 06:08:13 Of course, you need a few Decks of Everything. 06:08:27 Erm. 06:08:33 I forget the name of that again. . . 06:08:48 Deck of Many Things. 06:10:21 * pikhq goes off, sticks of Bag of Holding in a Bag of Holding. 06:10:57 Speaking of which. . . 06:11:05 RodgerTheGreat: You should have given out a bag of devouring. 06:12:09 One of my players sewed a bag of devouring into his pocket, hilariously slaying several pickpockets 06:12:46 LMAO 06:12:52 BRILLIANT! 06:13:07 and the books make it clear that storing bags of holding or portable holes in themselves or each other will tear a hole in existence and suck shit inside before disappearing 06:13:37 Fun to mess with magic objects. 06:14:41 I am *intimately* familiar with all the core rulebooks, particularly magic items. I'm good at memorizing these things, which is why I was a good DM. They're still all up here in my skull. 06:15:08 Damn. 06:15:10 After a year or two I never needed the books at all unless somebody had to roll up a character 06:15:26 Hmm. What's the funniest one you can think of? 06:15:30 just me, my notes, and dice. 06:15:37 uhm... 06:16:00 well, the hand and eye of Vecna aren't exactly funny, but they're interesting 06:16:26 * pikhq is not familiar with the magic items. . . 06:16:35 but you can't just sling those around willy nilly, because they're supposedly unique 06:16:36 Didn't come up much in campaigns I've played. ;) 06:17:14 I made loads of Vorpal quarterstaffs and intelligent +2 hafling skiprocks and goofy stuff like that 06:17:40 LMAO 06:20:19 hell, I could run an adventure over skype for old times sake someday if you can get a decent group together 06:20:44 If I can, sure. 06:21:08 what edition do you use? 06:22:02 I'm used to 3rd. 06:22:15 I might have a copy of 2nd around here, though. 06:22:20 good, good. Would you be up for trying 1st edition sometime? 06:22:30 the term "fucking brutal" doesn't do it justice 06:23:00 Sure, I'd be up for trying it. 06:23:01 I've always wanted to run an adventure for it, and I have a handful of classic modules I could snazz up 06:23:04 I lack rulebooks, though. 06:23:35 I can scan a couple pieces, explain the rest and preroll characters for people 06:23:49 I even have a table of pregenerated characters for this purpose 06:23:52 Mmkay. 06:24:12 Everyone will probably get 2 characters because they'll die like flies 06:25:08 example: in 1st edition, if you're disabling a poisoned lock and miss the DC, it stabs you. Fort save to resist poison. Fail the DC there, and your character is dead. period. stats are irrelevant. 06:25:53 Holy shit. 06:25:59 holy shit indeed 06:26:21 I wanna try that trick with the bag of devouring, though. . . 06:26:25 Hmm. 06:26:33 Something more unique would be nice, though. 06:26:36 in 1st edition a wizard's starting HP is lower than an average housecat. There is no difference between money and XP. Everything, EVERYTHING is fucking hardcore. 06:27:03 O.O 06:27:38 Melf's Acid apples are cool. Bite into one and it'll dissolve most of your head. They're best used as grenade-like things once you know what they actually are. Good for making holes in castle walls 06:27:55 (one of my inventions, btw) 06:28:34 I think the damage was like 2d10+5+ 1d4 CHA damage if you ate it. 06:28:44 but really, improvisation is key here 06:30:29 * pikhq discovers an online copy of the d20 ruleset. w00ts. 06:30:51 (and yes, I know there's a difference between CHA and Rifts-style P.B.- you're less persuasive and convincing when you don't have a lower jaw.) 06:31:21 A Cloak of Poisonousness. . . 06:31:22 Hmm. 06:31:31 That could actually be used as a deadly weapon. 06:31:36 did I mention that aside from encyclopedic knowledge of D&D I know about 6 other rulesets as well? 06:31:51 (put it on. Unless you succeed a fortitude check of 28, you die.) 06:31:53 Wow. 06:32:45 This all contributes to why I play with out books, and make up most stuff on the fly- I know how everything works, and I know all kinds of ways it *could* work. 06:33:45 eventually, you realize the books are just there to point you in the direction, and if you take enough of a running jump on your own, you can learn to fly. 06:35:21 Well, of course. 06:35:28 it's like the DM/GM equivalent of becoming "the one". The rules of a roleplaying game are just like the rules of a computer system- some can be bent. Others, broken. 06:35:38 The game is just a conduit for storytelling. 06:36:00 Most of the time, I'd play using nothing but my character sheet. . . The DM would look things up from time to time, but most of the time, it was just sheer amusement. . . 06:36:15 * pikhq remembers the final day of his first campaign. . . A giant chicken ended that. XD 06:37:25 The DM stuck a rubber chicken and some miniatures on the table and said 'There. You're fighting that.' 06:37:29 'I summon a giant bee.' 06:37:30 :D 06:37:41 A good DM knows the only rules necessary to play are the following: 1) Be consistent. 2) Be fair. 3) You want the players to succeed, but they don't have to know that. 4) keep 'em guessing. 06:37:44 * pikhq <3 druids, BTW. . . 06:38:14 follow those principles and you can have a grand time doing *anything* 06:38:18 Another campaign ended in apocalypse. For the hell of it, he was rolling an apocalypse check every time we did something really, really unlikely. 06:38:23 (3 20s in a row, for example.) 06:39:49 best moment for me as a player EVER "I rolled a 20! I succeeded" "No, you didn't" "What if he rolls a nat 20 too?" "Yeah, like that's gonna happen" (rolls a nat 20 without blinking) "... fuck you guys. It explodes." 06:40:38 "I rolled a 20. Thrice." "That. . . Can't die. . ." 06:41:11 new galleries on my site for games (http://rodger.nonlogic.org/games.php) and comics (http://rodger.nonlogic.org/comics.php), btw 06:41:13 "Fine. It dies. You see it again, behind you." 06:42:00 w00ts. 06:42:07 "I hide and make the sounds of a babbling brook to throw the owlbears off my trail" (critical failure) "BAAAASSS! BAAASSS!" 06:42:53 LMAO 06:44:24 and then there was the time my character used weighted dice and some insane bluff checks to win EVERYTHING the guards to this fortress OWNED while the other characters snuck in and locked them out. 06:44:51 http://rodger.nonlogic.org/comics/Comic019.png Brilliant. 06:45:27 "halt who goes there" "Flashbang, gnomish smith of great renown. Up for a game of chance? I have *shiny* things..." 06:45:54 I thought CL#24 would be topical 06:45:58 :D 06:47:45 have you tried out "Polyp", my new game? 06:47:50 http://rodger.nonlogic.org/comics/Comic011.png Sorry, but I won. 06:47:53 I'll have to in a bit. 06:48:09 that XKCD was so uncool. 06:48:20 I disagree. 06:48:35 I think The Game just became more positive. 06:49:30 I think ryan just has a less rigorous strain of the virus- mine is pretty explicit about what constitutes a win condition. 06:49:54 "A STRANGE GAME; THE ONLY WINNING MOVE IS NOT TO PLAY. *klick*" 06:50:40 and goddamnit why hasn't Aaron updated yet. 06:50:57 Because he must be executed. 06:51:10 Nice 'weekly' schedule, no? 06:51:25 s/weekly/semi-monthly 06:51:35 If we're lucky. 06:53:05 I gather you're starting to take my point of view on the matter 06:53:20 Yes. 06:53:28 It was one thing when he didn't update during the flu and moving. . . 06:53:40 "If you do a strip for a living, MOTHERFUCKING TREAT YOUR RELEASE SCHEDULE AS MORE THAN A VAGUE SUGGESTION" 06:53:46 But god dammit, it's halfway through March, and we've had one update in the past two weeks. 06:53:56 s/weeks/months/ 06:54:08 there we go 06:55:00 at that rate, 6 strips per YEAR. Dresden Codak is well done, but I could draw something that well blindfolded if I had two months in which to make it. 06:55:29 http://rodger.nonlogic.org/comics/Comic001.png 06:55:41 that one's a classic 06:55:45 Truly. 06:55:53 I'll need to stick that one on my wall or something. 06:56:17 based on a true story no less 06:57:46 Wow. 06:57:56 I had a fun orientation week. 06:58:39 Heheh. 07:00:22 -!- Sgeo has quit (Read error: 104 (Connection reset by peer)). 07:03:21 anyway, frickin' try Polyp before I go to sleep 07:03:42 Mmkay. 07:05:31 Brilliant. 07:05:48 thankoo 07:05:56 under 700 lines of code 07:07:31 -!- calamari has quit ("Leaving"). 07:13:47 Damn. 07:14:42 feel free to download the source and fiddle. It's all WTFPLed 07:34:31 I'm out 07:34:41 -!- RodgerTheGreat has quit. 07:47:40 -!- BMeph has quit ("calls it a night"). 07:59:59 -!- clog has quit (ended). 08:00:00 -!- clog has joined. 09:14:24 -!- RedDak has joined. 09:37:24 -!- Corun has quit ("This computer has gone to sleep"). 10:20:03 -!- RedDak has quit (Read error: 104 (Connection reset by peer)). 10:25:04 -!- RedDak has joined. 11:32:46 -!- marshmallows has joined. 13:14:04 -!- slereah_ has quit (Remote closed the connection). 13:14:31 -!- Slereah has joined. 13:20:56 what is a "low-order copy"? 13:21:33 it's mentioned in some befunge fingerprint specs I'm trying to implement, but I got no idea what it is 13:26:16 C ('bracelet') pops three vectors off the stack and performs a low-order copy of Funge-Space. 13:30:49 IIRC the way I interpreted that is start from the end, not the beginning, or vice versa 13:31:00 i.e. go from top left to bottom right versus bottom right to top left 14:04:16 ah so not a high order byte copy? 14:08:03 -!- RodgerTheGreat has joined. 14:08:33 -!- bsmntbombdood has quit (Read error: 110 (Connection timed out)). 14:09:18 -!- bsmntbombdood has joined. 14:09:24 no, probably not 14:10:37 howdy 14:10:55 yo 14:11:16 AnMaster: if I recall correctly, !Befunge does it the same way - or at least tries to, the mycology test fails 14:11:36 heh 14:12:02 Deewiant, you need to update that chart, try my last version, it got concurrency 14:12:12 I need to update it anyway 14:12:16 Deewiant, pretty fast still with my super-optimize script 14:12:29 no idea if you *want* to see it 14:12:37 probably not ;-) 14:12:52 Deewiant, it's not that bad 14:13:09 Deewiant, http://rafb.net/p/ipoHrq23.html 14:13:18 need to be run from top checkout dir 14:13:25 ie, out of tree builds = not supported 14:14:12 Deewiant, also skipping GC = saves about 0.01 seconds or so 14:14:26 * AnMaster is down around 0.077 here so :D 14:14:35 with that script 14:15:20 Deewiant, anyway what do you think? try adding your own -march and such in CHOSTFLAGS 14:15:34 I think that's overkill :-) 14:15:35 I want to know how well it performs with that 14:15:47 Deewiant, yes but I want to see how fast I can get it! 14:16:04 you haven't compared startup times yet? i.e. just replace the first char of mycology with @ 14:16:06 Deewiant, oh and you may want to add -DCONCURRENT_FUNGE to DEFINES variable 14:16:17 Deewiant, right, I'll test that 14:17:03 * AnMaster ponders optimizing if it can be determined that part of the program won't get loaded, ie, if first char is @ just quit :D 14:17:55 real 0m0.066s 14:18:03 Deewiant, hrrm, looks like loading takes most of the time 14:18:11 for cfunge at least 14:18:19 as I thought 14:18:40 the program is completely IO bound and you're optimizing CPU usage like crazy :-) 14:19:06 Deewiant, well without my optimizing it wouldn't be IO bound 14:19:19 * AnMaster goes to look at how to make loading faster 14:19:20 no, I still think it would be 14:19:41 if a program spends 90% of its time doing IO it's IO bound alright ;-) 14:20:00 Deewiant, because in CCBI the difference is much larger 14:20:19 how big? 14:20:28 real 0m0.188s for ./ccbi --disable-fprints ~/src/cfunge08/mycology/mycology.b98 > /dev/null 14:20:32 real 0m0.117s 14:20:35 and keep in mind that CCBI is always doing IO during mycology as well :-) 14:20:36 with it disabled 14:21:09 is that the latest CCBI? 14:21:29 Deewiant, well notice that I got it down from around 0.150 to 0.070 with my cpu optimizing, with -O0 -ggdb3 it takes around 0.500 14:21:54 Deewiant, no, not the very last, will download last 14:22:06 this one had the typo in --help 14:22:11 so we've got a program that used to take 0.2 seconds to run now takes 0.1 seconds to run, and it's doing IO for 70% of that :-) 14:22:23 Deewiant, for what is left yes 14:22:43 I don't know, optimizing such just seems like so much overkill :-) 14:22:53 I'd go optimize bashfunge instead 14:23:12 Deewiant, bah, it would fail at doing x 14:23:18 the way it was implemented 14:23:57 Deewiant, the last ccbi zip file from your site is slower 14:24:00 $ time ./ccbi --disable-fprints ~/src/cfunge08/mycology/mycology.b98 > /dev/null 14:24:00 real 0m0.179s 14:24:07 with @ at start 14:24:10 real 0m0.244s 14:24:13 without @ at start 14:24:14 -!- timotiis has joined. 14:24:15 heh 14:24:29 I don't know, it's about 50% faster for me :-) 14:24:37 Deewiant, probably CPU? 14:24:42 hence we see that optimizations are once again a bit... dubious 14:24:49 Deewiant, I downloaded the last ccbi_linux.zip 14:24:55 Deewiant, not if you do profiling! 14:25:04 I do do profiling 14:25:07 oh? 14:25:10 gcov? 14:25:17 that took the time for funge-space access down by 30% or so 14:25:21 here, at least. 14:25:41 Deewiant, yes indeed, changing to not pass fspace pointer around helped for me too 14:25:56 and it wouldn't be needed without TRDS I suspect 14:26:51 Deewiant, if I profile for a mycology with a @ at the start I can get time down to less than 0.06 seconds XD 14:26:57 * AnMaster runs (it was a joke!) 14:27:21 of course, what I don't do is profile-guided optimization. :-P 14:27:34 Deewiant, hehe indeed 14:28:25 I got an idea at least, hm can you get the size of a file without using stat *looks at man pages* 14:28:47 Deewiant, what I'm thinking about is doing some optimizing of initial allocation size 14:28:58 to select maybe 1.25 * program size 14:29:15 instead of hard coded defaults 14:34:34 -!- oerjan has joined. 14:54:57 Deewiant, it looks, according to gprof, like the bottle neck is inserting into hash table 14:55:09 yep 14:55:23 so, faster hash library then? 14:55:43 or admit you're at the limit :-P 14:56:06 Deewiant, nah, current hash library I selected because it was 1) lgpl 2) small 3) easy to use 14:56:34 I've seen better that are 1) custom GPL compatible 2) large 3) harder to use 14:57:04 http://www.sunrisetel.net/software/devtools/sunrise-data-dictionary.shtml <-- going to investigate that one 14:57:30 Deewiant, anyway I can speed up current hash library a bit 14:57:41 by hard coding it to handle FUNGEDATATYPE 14:57:51 instead of handling pointers to data 14:57:53 should help eh? 14:57:58 :-P 14:58:11 Deewiant, you don't think so? 14:58:37 I'm not going to say anything, I'm just going to sit here and laugh at your desperate attempts to squeeze every last bit of performance from a Befunge interpreter :-D 14:58:39 Deewiant, also it use a function pointer to a hash function, hard coding it to use one of the hash functions instead would also help 14:59:08 Deewiant, it's just a matter of seeing "how far can I take this" 14:59:28 Deewiant, it looks like most code in mycology only executes once, right? 14:59:34 so a JIT wouldn't help much 15:01:25 Deewiant, looks like gprof works sub-optimally on binaries with much inlining hrrm 15:05:17 Deewiant, without inlineing I see it is "search_in_bucket" private function of hash table that is slowest 15:10:05 here a GC function is the one that takes most time 15:10:23 AnMaster: what style of GC is the boehm one? generational, mark-and-sweep, what? 15:10:42 Deewiant, mark-and-sweep iirc 15:11:00 or some variant of that 15:11:41 yep mark-and-sweep 15:12:04 Deewiant, anyway that was a build without GC I tried to check what was going on in hash 15:12:21 looks like it does some generational collection as well 15:12:34 the search_in_bucket takes 80% of the time of running when mycology starts with a @ 15:12:37 :/ 15:12:48 and it already looks quite optimized 15:13:19 here gc.mark takes 30%, followed by input, followed by hashing, followed by input, followed by gc.fullCollect, followed by gc_calloc 15:13:31 Deewiant, calloc, hrrm 15:13:39 -!- slereah_ has joined. 15:13:45 Deewiant, anyway I did LEAK_MODE to check without gc 15:13:51 AnMaster: D always initializes all variables 15:14:03 might cause slowdown somewhere with arrays, can't be bothered to find where 15:14:06 Deewiant, aha, well I think boehm does too 15:14:54 the difference is that where you do int[40] x; D also does memset(&x, 0, 40 * int.sizeof); or something :-) 15:14:54 at least you can safely do: #define calloc(x,y) GC_MALLOC(x) * (y)) 15:14:55 iirc 15:15:03 Deewiant, ah I see 15:15:19 but I don't really use static arrays much anyway 15:15:34 static arrays would not be initialized would they? 15:15:42 that was a static array 15:15:45 and yes, it would be. 15:15:46 nor ones allocated on the stack, right? 15:15:50 hm 15:15:52 yes, they would be. 15:15:56 everything is initialized to something. 15:16:02 they wouldn't in C I think 15:16:07 well static array might 15:16:08 depends 15:16:09 correct, they wouldn't. 15:16:22 C initializes nothing implicitly. 15:16:56 -!- Slereah has quit (Read error: 104 (Connection reset by peer)). 15:16:59 Deewiant, that still doesn't explain why an array on stack ended up as { 0, 0, 0, 0 } without initializing heh 15:17:11 dumb luck 15:17:20 -!- oklofok has quit (No route to host). 15:17:41 Deewiant, probably, anyway I filled it just one line after so didn't matter, but gdb told me every time I hit it that it was that 15:19:18 * AnMaster thinks DESCRIPTION and RETURN VALUE of man memcmp are somewhat redundant, telling the same info 15:20:02 aha, gcc does a __builtin__ for memcmp, explains why that instruction took so much 15:21:19 Deewiant, btw kprof is very nice 15:22:03 Deewiant, and it turns out according to it, the real problem isn't loading when running without @, but getting from hash 15:23:58 Deewiant, of the 0.8 seconds to run without the @, about 0.4 is spent searching in hash 15:23:59 :/ 15:24:05 so not really IO bound 15:24:13 but hash table lookup bound 15:24:28 and of course insert on loading program needs to look up too 15:33:56 -!- slereah__ has joined. 15:40:31 -!- oklofok has joined. 15:40:50 -!- slereah_ has quit (Read error: 104 (Connection reset by peer)). 15:44:50 Deewiant, for now I'm not going to try to optimize hash, but the way it is coded it should be easy to replace with another hash library later 15:53:53 -!- faxathisia has joined. 15:54:51 -!- marshmallows has quit (Nick collision from services.). 15:55:03 -!- faxathisia has changed nick to marshmallows. 16:00:37 -!- Slereah has joined. 16:03:30 -!- slereah__ has quit (Read error: 110 (Connection timed out)). 16:13:28 * SimonRC curses his net connection 16:18:14 Should I have whitespace separate type and var name the way C does? 16:18:40 Deformative in what context? 16:18:42