00:00:18 he's not really indecisive, though? 00:00:20 just angry 00:00:48 One could probably make the case that donald is indecisive, but it's still not thematically appropriate dialogue 00:00:55 huh. I tried to open the objstat output, and xcode crashed 00:01:03 -!- reaverb has quit [Quit: Leaving.] 00:01:04 xcode? 00:01:10 it's a text file 00:01:13 yes 00:01:13 tdt text 00:01:31 don't you have a spreadsheet 00:01:34 that's much better 00:01:43 or something that can recognize the tabs as fields at least 00:01:46 -!- valrus has quit [Quit: My iMac has gone to sleep. ZZZzzzā€¦] 00:01:51 not afaik 00:01:51 -!- reaverb has joined ##crawl-dev 00:01:55 so this is pretty useless to me 00:01:57 ah well 00:02:03 PleasingFungus: google docs then 00:02:08 mm 00:02:11 PleasingFungus: Are you serious about the whole "remove extended" idea? 00:02:16 Lightli: no 00:02:17 oh god 00:02:20 however 00:02:22 I am very serious 00:02:24 about removing Lightli 00:02:34 just making certain 00:02:54 -!- CSDCS has quit [Read error: Connection reset by peer] 00:02:58 -!- Keanan has quit [Quit: Leaving.] 00:03:25 hm. gammafunk, I told it to open the file, but then it happily uploaded & did nothing 00:03:28 gdocs 00:03:31 gdocs spreadsheet 00:03:50 oh, let me check right quick 00:04:01 import does the same thing 00:04:21 PleasingFungus: open calc, file->import 00:04:26 yes 00:04:30 and do upload 00:04:32 yes 00:04:36 I did all of those things 00:04:37 it doesn't import as a sheet? 00:04:41 it spins and then does nothing 00:04:58 PleasingFungus: ok, easy solution 00:05:02 web apps \o/ 00:05:02 it shows them in the 'recent' imports' 00:05:07 if I hit select, it spins, and then does nothing 00:05:08 rename the file to have a .csv extension 00:05:15 or even .xls would be fine 00:05:28 -!- Stendarr has quit [Quit: KVIrc 4.2.0 Equilibrium http://www.kvirc.net/] 00:05:40 it is kind of brain-dead in that respect, isn't it 00:05:42 a little silly 00:05:46 spins 00:05:48 does nothing 00:06:02 (with .csv. will try .xls) 00:06:27 PleasingFungus: oh, you're google docs is weird 00:06:30 when I open a spreadsheet 00:06:32 you'll never guess what just happened!!! 00:06:33 and to import from file 00:06:51 it imports it as a sheet 00:06:58 are you in a spreadsheet when you import? 00:07:06 yes 00:07:08 hm 00:07:10 got it to another page 00:07:11 ... 00:07:15 claims the xls is damaged 00:07:18 but this is still progress 00:07:31 PleasingFungus: you try as txt 00:07:33 ok. got the csv to import 00:07:38 *yeah 00:07:48 well, w/e, but when I create a new spreadsheet in google docs 00:07:51 and do import 00:07:54 and upload a txt 00:07:56 and the txt. 00:07:58 okay 00:08:01 do you want to know the culprit 00:08:07 Unstable branch on crawl.s-z.org updated to: 0.15-a0-1814-gad9ca24 (34) 00:08:08 is it newlines or something? 00:08:09 beyond "me being dumb" (this is the culprit) 00:08:18 the culprit is a lil' hit from the mid-00s 00:08:20 I like to call it 00:08:22 noscript 00:08:29 oh 00:08:30 ahah 00:08:39 one of the many scripts was being blocked 00:08:41 never really used that 00:08:55 welp, good to know I guess 00:08:57 it is kind of past its time 00:09:12 hm 00:09:14 so 00:09:21 PleasingFungus: So there are better script blockers now? Or is the concept outdated? 00:09:29 the latter, I suppose 00:09:30 idk 00:09:34 I use it anyway because I am an old fogey 00:09:50 On something completely unrelated: I was thinking about making Silent Speteres work like Torpor Slugs or Ancient Zymes, except with a Silence low duration enough to only effect the player, so monsters spellcasters aren't effected by them. 00:09:50 so uh 00:09:51 yeah, I think add blocker plus blocks most of what people care about 00:09:59 uhh 00:10:08 xp without big fish: 00:10:09 110 00:10:13 xp with big fish: 00:10:17 please guess 00:10:23 wait, that's 00:10:25 100 ? 00:10:32 265 00:10:33 however 00:10:34 110 where? 00:10:35 this is a trick 00:10:41 That can't be correct. 00:10:47 there is a pan lord 00:10:51 what are you looking at, I guess 00:10:51 worth 160 xp 00:10:52 Ha. 00:11:00 *135 00:11:07 I guess 135 on average 00:11:14 I think that entry vault is fucking with objstat 00:11:16 is this one level of the dungeon? from all monsters? 00:11:20 gammafunk, are you respecting no_rewards? 00:11:28 this is d:1-10 00:11:32 so it seems a bit awkward that you can no longer sacrifice a corpse if it is on an altar that isn't your own god's 00:11:36 from all monsters, yes 00:11:41 since you can't move them 00:11:43 minmay: I think I had a fix for that at one point 00:11:48 I forget if I ever pushed it 00:11:52 PleasingFungus: can you share that spreadsheet and give me the link? 00:11:59 how about just assume the player would rather sacrifice a corpse than abandon their god 00:12:05 share on the upper-right 00:12:07 -!- Keanan has joined ##crawl-dev 00:12:16 yeah I know the button 00:12:19 gammafunk: https://docs.google.com/spreadsheets/d/1OxJ8Cv6aUbxeVlZgiC9tqzfvlVhBpVMTpQAXGhkzAg4/edit?usp=sharing 00:12:19 Of course the real fix for that is to remove corpse sacs <_< 00:12:23 minmay: have you seen the relevant code 00:12:25 it is lovely 00:12:31 man that's too rude. it's not *that* bad 00:12:33 but it is kind of a mess 00:13:20 !source pray 00:13:21 http://s-z.org/neil/git/?p=crawl.git;a=blob;f=crawl-ref/source/player-reacts.cc;hb=HEAD#l70 00:13:21 -!- CSDCMS has quit [Read error: Connection reset by peer] 00:13:28 that is... incorrect 00:13:35 !source godprayer.cc:280 00:13:36 http://s-z.org/neil/git/?p=crawl.git;a=blob;f=crawl-ref/source/godprayer.cc;hb=HEAD#l280 00:13:37 !source godprayer.cc:380 00:13:38 http://s-z.org/neil/git/?p=crawl.git;a=blob;f=crawl-ref/source/godprayer.cc;hb=HEAD#l380 00:13:42 ^ 00:14:41 probably want to move the sacrifice code up and putz with the messages, I guess 00:15:20 PleasingFungus: yeah I do respect no_xp, but most vaults that place those "decorative monsters" don't mark them as no_xp 00:15:28 hm 00:15:46 fixorate? 00:16:19 -!- buddhastalin has quit [Ping timeout: 246 seconds] 00:16:21 -!- omnirizon has quit [Ping timeout: 255 seconds] 00:16:33 huh. looks like neither sharks nor fish were really spawning in d1:10 anyway 00:17:04 Even before I killed their random spawning there? 00:17:31 yes 00:17:43 Hmm. 00:22:14 yeah they're rare 00:22:33 there's some variation in XP per level in these two 00:22:44 but it's probably variance due to weird vaults 00:24:34 yeah, 90% of the difference is that pan lord 00:24:35 which is p funny 00:25:08 unfortunately that one is a bit touch to account for 00:25:09 hrm 00:25:16 does that vault use a special tag 00:27:31 ah, just the arrival tag 00:27:34 that's not much help 00:28:02 -!- Kalir has quit [Quit: CADENZA-CLASS CLOCKWORK KNIGHT--OPERATION TERMINATED] 00:29:22 I guess I should run it with 500 iterations at some point to help with that variance a bit more 00:29:28 so looking at arenasprint the only outdated thing i notice is that the hungry ghost wave still has harpies 00:36:59 -!- DayBay has quit [Ping timeout: 264 seconds] 00:38:01 -!- HamsterSaurusMex has quit [Ping timeout: 246 seconds] 00:40:13 -!- magicpoints has quit [Ping timeout: 244 seconds] 00:40:53 -!- PsyMar has quit [Ping timeout: 252 seconds] 00:41:22 -!- magicpoints has joined ##crawl-dev 00:44:08 -!- PsyMar has quit [Client Quit] 00:44:27 03PleasingFungus02 07* 0.15-a0-1815-ge40f12a: Mark items visibly held by monsters as seen 10(8 minutes ago, 1 file, 8+ 0-) 13http://s-z.org/neil/git/?p=crawl.git;a=commitdiff;h=e40f12a500ed 00:45:42 I just realized all the pulsating lumps in sprint1 are jellies now and are on top of KITEM: any good_item 00:45:52 hahaha 00:45:58 ChangeAj: Ha, wow. 00:46:00 !tell wheals ChangeAj I just realized all the pulsating lumps in sprint1 are jellies now and are on top of KITEM: any good_item 00:46:01 PleasingFungus: OK, I'll let wheals know. 00:46:54 that's legit funny as hell 00:47:35 -!- ProzacElf has quit [Ping timeout: 248 seconds] 00:47:35 -!- simmarine has quit [Ping timeout: 248 seconds] 00:47:38 honestly jellies are way more difficult to deal with there too 00:48:04 PleasingFungus: so, monsters aren't allowed to pick up items that their companions drop if the player saw them carrying those ever? 00:48:25 that seems reasonable 00:48:27 tbh 00:48:44 that seems like a much more niche & less problematic case than the current one 00:49:01 which is "monsters can pick stuff up from the middle of stacks, so you should walk over all stacks ASAP" 00:49:14 which is both inexplicable to the player 00:49:17 and weird 00:49:18 oh yeah 00:49:27 -!- johlstei has joined ##crawl-dev 00:49:37 -!- simmarine has joined ##crawl-dev 00:49:49 -!- Morik has quit [Quit: This computer has gone to sleep] 00:50:30 so now you lose scenarios like "see sonja, poison her & go to another set of stairs, she dies and her pal picks up her dagger". which is... okay 00:50:48 yeah, yeah, 00:50:52 I got it ;-) 00:50:56 sorry 00:51:02 didn't mean to berate you 00:52:09 -!- reaverb has quit [Quit: Leaving.] 01:00:45 berate me, I like it 01:01:48 -!- magicpoints has quit [Ping timeout: 255 seconds] 01:01:59 -!- Diabl0658 has quit [Ping timeout: 248 seconds] 01:02:05 gammafunk: you are a terrible person, and your code smells of herring. 01:02:16 tyvm 01:02:24 np 01:02:25 you will never have the high elf highscore 01:02:30 dang, cold 01:02:35 ok now that's simply lying 01:02:58 Bah, herring? his code spells of Surströmming!!! 01:04:00 -!- magicpoints has joined ##crawl-dev 01:10:11 -!- rast has quit [Read error: Connection reset by peer] 01:10:44 -!- rast has joined ##crawl-dev 01:11:34 -!- rast- has joined ##crawl-dev 01:11:50 -!- WalkerBoh has quit [Remote host closed the connection] 01:13:43 -!- rophy has quit [Ping timeout: 248 seconds] 01:15:32 -!- rast has quit [Ping timeout: 260 seconds] 01:15:38 -!- rast- is now known as rast 01:19:14 god damn it, I thought i --amended that commit 01:19:15 rip 01:19:27 03PleasingFungus02 07* 0.15-a0-1816-gf986df7: Allow sacrificing corpses on other gods' altars (minmay) 10(4 minutes ago, 1 file, 69+ 44-) 13http://s-z.org/neil/git/?p=crawl.git;a=commitdiff;h=f986df79bd9d 01:19:27 03PleasingFungus02 07* 0.15-a0-1817-gcafc426: Prayer tweaks 10(27 seconds ago, 1 file, 6+ 2-) 13http://s-z.org/neil/git/?p=crawl.git;a=commitdiff;h=cafc426c7a5b 01:20:31 -!- magicpoints has quit [Ping timeout: 244 seconds] 01:20:41 -!- demiskeleton_ has quit [Quit: Leaving] 01:25:05 -!- Euph0ria has quit [Remote host closed the connection] 01:25:46 -!- UncertainKitten has quit [Quit: Connection reset by pier.] 01:30:00 -!- mong has quit [Quit: ChatZilla 0.9.90.1 [Firefox 30.0/20140605174243]] 01:30:33 -!- lobf has quit [Quit: lobf] 01:31:54 haha, trolls are special-cased to not get messages for rotting meat 01:31:59 because "they're stupid and don't care" 01:34:57 1learn add racism 01:42:11 -!- PepeRC2 has quit [Remote host closed the connection] 01:42:44 -!- ussdefiant has quit [Ping timeout: 244 seconds] 01:42:53 hm. is it safe to assume that there are no corpses in inventories? 01:43:09 -!- ProzacElf has joined ##crawl-dev 01:43:49 eh. if a few people are left with unrotting, eternal corpses in their inventories, that's ok 01:48:25 -!- Keanan has quit [Quit: Leaving.] 01:54:58 03PleasingFungus02 07* 0.15-a0-1818-g3d0ae94: Make carried skeletons/corpses never rot 10(4 minutes ago, 1 file, 19+ 88-) 13http://s-z.org/neil/git/?p=crawl.git;a=commitdiff;h=3d0ae94e5589 01:54:58 03PleasingFungus02 07* 0.15-a0-1819-g0448553: Remove hints referring to carried corpses 10(3 minutes ago, 1 file, 10+ 23-) 13http://s-z.org/neil/git/?p=crawl.git;a=commitdiff;h=0448553f169f 01:55:05 Selecting 0.14 on CDO gives you 0.13 instead; who maintains the server? 01:55:49 Or rather, whom should I tell about this? :) 01:56:06 -about 01:56:25 ??cdo 01:56:25 is cdo down[1/1]: Not especially! 01:56:28 huh 01:56:43 -!- ontoclasm has quit [Quit: Leaving.] 01:56:53 someone fucked up 01:56:56 !learn del cdo[1 01:56:57 Deleted cdo[1/5]: see {is cdo down} 01:56:59 ??cdo 01:57:00 cdo[1/4]: Crawl server (also running 4.1), located in Germany, crawl.develz.org, telnet port 345 or ssh port 22, ssh-username: crawl, ssh-key necessary: http://crawl.develz.org/cao_key (openssh) or http://crawl.develz.org/cao_key.ppk (putty) 01:57:26 anyway, I'd probably ask |amethyst or someone 02:01:24 -!- Staplegun has quit [Ping timeout: 240 seconds] 02:02:06 -!- xnavy has quit [Read error: Connection reset by peer] 02:04:20 !learn e cherub s/$/ In 0.15, will rouse nearby holy enemies, increasing their damage significantly. 02:04:21 cherub[1/1]: Boring holy monster, usually coming with ranged weaponry. In 0.15, will rouse nearby holy enemies, increasing their damage significantly. 02:06:36 -!- xnavy_ has quit [Read error: Connection reset by peer] 02:07:03 !learn add you_hear_a_strangely_unpleasant_explosion See {hellfire} 02:07:03 Term name exceeds the maximum length of 30 02:07:06 hm 02:07:07 -!- xnavy has quit [Ping timeout: 240 seconds] 02:08:53 -!- ChangeAj has quit [Quit: Lost terminal] 02:09:26 -!- Kramin has quit [Ping timeout: 252 seconds] 02:09:49 -!- Nomi has quit [Ping timeout: 272 seconds] 02:11:19 -!- Aeleron has joined ##crawl-dev 02:11:30 Testing... hi! 02:11:35 hello 02:12:13 I was looking for the source code that runs the "watch" feature that you run on your servers. Where can I find it? 02:12:49 * -->The watch feature that allows you to watch other peoples games. 02:13:46 I was curious to see how it was implemented. Surely you don't send the pixel information of all frames to each person who is watching it, do you? 02:14:48 http://s-z.org/neil/git/?p=crawl.git;a=tree;f=crawl-ref/source/webserver it's probably somewhere in here 02:15:15 I don't know that section of code so that's about as much as I can help 02:15:35 It uses tornado... 02:15:51 server.py 02:16:27 Whoever wrote this... this is really clean. 02:16:59 edthiolol or something like that, I think 02:18:31 This might be the script that handles the irc communication... There doesn't seem to be any comments, though. 02:18:49 Unstable branch on crawl.beRotato.org updated to: 0.15-a0-1819-g0448553 (34) 02:18:51 clearly it must be self-documenting code 02:18:59 -!- ProzacElf has quit [Ping timeout: 264 seconds] 02:20:12 If one is familiar with writing these kinds of scripts I suppose that it would be... 02:20:45 I would have to play with tornado for several hours to get a feel for how it works... 02:22:40 03|amethyst02 07* 0.15-a0-1820-g6c5f5d6: Remove a sewer entry. 10(3 minutes ago, 1 file, 0+ 9-) 13http://s-z.org/neil/git/?p=crawl.git;a=commitdiff;h=6c5f5d6e7898 02:22:42 There is a README 02:25:28 I'm sorry, I am really don't intend to try your patience. 02:25:51 * strike am 02:30:11 I really wish I understood this. Um: one possible memory model. If one creates a temporary cache to be loaded on the client side that includes all of the possible tiles in the game that one loads when they initiate communication with the server in the beginning (which creates a long load time in the beginning..) 02:30:23 -!- PleasingFungus has quit [Quit: ChatZilla 0.9.90.1 [Firefox 33.0a1/20140620030201]] 02:32:00 then one could just send tile information that includes the names and addresses of the tiles; which the client then uses to generate the graphical representation of the play session. 02:33:45 <|amethyst> Aeleron: it's not even a temporary cache 02:33:47 <|amethyst> it's a sprite sheet 02:34:26 <|amethyst> the server sends JSON messages to the client with the feature type, floor tile, etc 02:34:34 <|amethyst> and the client composites them in Javascript 02:36:07 <|amethyst> the client side is in webserver/static/scripts (stuff that doesn't change when Crawl is updated) and webserver/game_data/static/ (stuff that does) 02:36:35 -!- Danei has quit [Quit: Those who tell the truth shall die, those who tell the truth shall live forever.] 02:36:55 So you load all of the textures into a huge sprite sheet. Send that thing client side? Then pass the dungeon features through http? Which the client then puts together via javascript to generate the image? 02:37:29 <|amethyst> Aeleron: It's actually through websockets, not pure http 02:37:46 <|amethyst> but essentially, yes 02:38:44 I don't know enough about how this stuff works. I don't really know the distinction between websockets and regular http communication. 02:38:53 <|amethyst> {"msg":"map","clear":true,"player_on_level":true,"vgrdc":{"x":0,"y":0},"cells":[{"x":1,"y":-6,"f":7,"mf":2,"g":"#","col":6,"t":{"bg":871}},{"f":7,"mf":2,"g":"#","col":6,"t":{"bg":871}},... 02:38:55 Google. I've got that... 02:39:07 <|amethyst> is what the beginning of the map looks like 02:39:39 Is it passing the huge sprite sheet that accounts for most of the lag that the players experience? 02:39:49 <|amethyst> no, because that happens only once 02:39:59 <|amethyst> most of the lag comes from network latency 02:40:03 (also console doesn't have tiles and it still lags) 02:40:23 <|amethyst> if you have 100 ms latency to the server, that means you're waiting 1/10 of a second to get a response at least 02:41:06 What is the regular server load? How many players at maximum are on the server at once? 02:41:16 ...roughly. 02:41:51 I am seeing a maximum of 100 connections in the config.py script... 02:42:12 <|amethyst> probably 50 concurrent users at most, not all of whom are using webtiles 02:42:24 <|amethyst> but that's really only during tournaments 02:42:29 <|amethyst> right now there are 20 users on CSZO 02:42:40 <|amethyst> load is 4 02:43:05 How many processors do you have server-side? 02:43:42 <|amethyst> CSZO is four-core plus hyperthreading 02:44:00 <|amethyst> that load is not really from CPU though 02:44:47 Oh, that's what I just asked with the server load question. That has a more specific meaning than I was attaching to it... 02:45:11 -!- DrinkMachine has quit [Ping timeout: 252 seconds] 02:45:12 So you have 8 threads maximum... 02:45:35 <|amethyst> simultaneous, yes 02:45:59 <|amethyst> 50% usage right now 02:45:59 So your server is pretty much just a single mid range probably $600 PC. 02:46:18 <|amethyst> no, it's a Xeon 02:46:23 <|amethyst> it's in a rack 02:46:32 <|amethyst> but it's not top-of-the-line, no 02:46:40 <|amethyst> Intel(R) Xeon(R) CPU E31230 @ 3.20GHz 02:48:14 does it only run crawl? 02:48:26 and chei I guess 02:48:31 Is it donated gratis one of the developers? 02:48:33 and the %git repo I bet 02:48:37 <|amethyst> Chei actually runs on a different server 02:49:10 <|amethyst> it runs crawl, dobrazupa.org (serves data files), and sizzell (the announcement bot) 02:49:20 <|amethyst> it also runs the tournament scripts at tournament time 02:49:30 <|amethyst> I don't actually own the hardware; it belongs to the host 02:49:54 <|amethyst> I pay about $100/month for the hosting, including the dedicated server and essentially unlimited bandwidth 02:50:00 -!- simmarine has quit [Quit: Leaving] 02:50:14 yea. crawl seems to sip bandwidth for me 02:50:15 -!- minmay has quit [Ping timeout: 248 seconds] 02:50:24 the HD space is what I have to worry about :/ 02:50:35 I see. Do you recieve enough donations to cover the cost? 02:50:36 https://clientarea.ramnode.com/cart.php?gid=22 CBRO runs on the 2048 MB version of that 02:50:42 <|amethyst> /dev/sda3 455G 248G 185G 58% / 02:50:49 but CBRO is not busy at all :) 02:51:43 -!- zinny has quit [Ping timeout: 246 seconds] 02:51:43 <|amethyst> johnstein: only 2 GiB? 02:51:59 of RAM, yea 02:52:01 <|amethyst> CSZO is using 2 GiB right now 02:52:11 each crawl game seems to take up about 2% 02:52:39 usually less 02:52:51 I figured I would run into trouble if I had more than 50 concurrent players 02:52:55 -!- DrinkMachine has joined ##crawl-dev 02:53:00 but it survived the tournament 02:53:15 I think I'm definitely on the low-end with respect to the idea crawl server specs 02:53:57 maybe swapping is more tolerable with an SSD 02:54:28 -!- Patashu has quit [Ping timeout: 260 seconds] 02:54:32 most of the players are from mefightclub anyway. I have another Rotatell bot that announces boring milestones in #octolog that are relayed to our IRC room. so we can watch people kill terrance 02:57:23 -!- DrinkMachine has quit [Ping timeout: 264 seconds] 02:59:07 Interesting how this works. How much more expensive would it be to run on your own server? I suppose that your IP would probably charge more for the extra bandwidth usage on your end? 02:59:22 <|amethyst> you mean from home? 02:59:25 Yes. 03:00:18 -!- Patashu has joined ##crawl-dev 03:02:34 <|amethyst> I suspect you probably don't have adequate upstream bandwidth, because few home plans do 03:02:47 <|amethyst> not per month, but spot rate 03:03:00 <|amethyst> 95th percentile network bandwidth: 1.95 Mbps in, 3.02 Mbps out 03:03:02 I have flatrate full-duplex 100Mbit/s at home 03:03:06 <|amethyst> oh 03:03:12 -!- skele-tan_ has joined ##crawl-dev 03:03:23 Or different question. I've never bought a server before, so I'm not sure how the ownership concepts work exactly. When you buy a server; is it treated as "your" server that you can ssh into and run whatever daemons you please? Or is it more like a cluster where you send in "pbs scripts" to request operations to be run? 03:03:30 but I don't like the idea of running anything remotely responsible under my desk 03:03:37 <|amethyst> Aeleron: depends 03:03:55 <|amethyst> Aeleron: "dedicated hosting" is what I have 03:03:55 -!- skele-tan has quit [Ping timeout: 240 seconds] 03:04:12 <|amethyst> Aeleron: I rent a server, but it's all mine, I have root, and I can do whatever I want with it 03:04:38 dedicated servers are usually pretty expensive 03:04:50 <|amethyst> Aeleron: or you could get a shared server account, but then you're limited by what the other people on that server are doing 03:05:00 I see. So dedicated hosting is like the first concept. "shared hosting" is like the second concept? 03:05:22 <|amethyst> Yeah; though the implementation details of shared hosting may differ from one host to the next 03:05:24 there are also VPSes, which are usually VMs that are completely yours, but which share resources with other people's VMs 03:05:57 Shared hosting could be more efficient, in theory, if you had a huge server to work with and the costs were distributed among a large number of people? 03:06:26 that's the theory behind e.g. amazon ec2 03:06:36 it relies on the assumption that the average resource utilization of an individual server is not very high, indeed 03:07:52 So, in theory; that could be cheaper. In practice, I guess your mileage may vary? 03:08:24 yes, it's cheaper, but it's usually more resource-constrained, and relies on (unpredictable) other users' resource utilization when not 03:08:46 <|amethyst> I think a crawl server would get kicked out by most VPS hosts 03:08:54 <|amethyst> for overutilization 03:09:20 <|amethyst> not EC2 03:09:40 <|amethyst> but EC2 running 24 hours a day seems to be not any cheaper than what I'm paying for now 03:09:49 yeah, EC2 is quite expensive 03:10:52 even their cheapest offerings aren't as cheap as digital ocean or linode 03:11:30 <|amethyst> still might make sense for spot allocation during tournaments 03:11:46 indeed 03:11:52 -!- rast- has joined ##crawl-dev 03:12:07 Linode is advertising SSD root access server hosting for about $5/month... but this is the VPS that would likely kick crawl for overutilization? 03:12:41 -!- TAS_2012v has quit [Ping timeout: 252 seconds] 03:13:22 Aeleron: my VPS is also rented but I have root access and can do whatever. 03:13:40 I see linode's cheapest offering at $10/month, and it has only one core 03:13:49 -!- rast has quit [Read error: Connection reset by peer] 03:13:52 -!- rast-- has joined ##crawl-dev 03:13:55 -!- rast-- is now known as rast 03:13:56 ah, there's 'view smaller plans' 03:13:59 I haven't been kicked out of my VPS yet, but when there's some runaway crawl processes that peg CPU usage to 100% for an hour, I get emails 03:14:03 ah, nevermind 03:14:21 and sometimes they reboot things. I need to set up a cron job that monitors that and pings me on IRC when that happens 03:15:07 but no, I don't think linode would kick you out, but you'd still be resource-constrained on their cheapest offerings 03:15:23 What types of processes tend to generate that CPU usage? 03:15:41 I'm pretty happy with my setup, but again, CBRO isn't very busy so I know it's got a limited max size. 03:15:54 Tornado is based on forking processes? 03:16:39 If there was a path that had tornado continuously fork processes; that could cause a runaway issue? 03:17:37 <|amethyst> the webserver itself is single-threaded 03:17:48 -!- rast- has quit [Ping timeout: 260 seconds] 03:18:01 <|amethyst> but each player has their own copy of crawl running 03:18:19 <|amethyst> so if crawl goes into an infinite loop or something... 03:19:14 -!- Blazinghand has quit [Read error: Connection reset by peer] 03:19:43 I see. 03:20:24 How does the server know where to look to run crawl client side? 03:24:19 -!- skele-tan_ has quit [Ping timeout: 240 seconds] 03:24:32 -!- skele-tan has joined ##crawl-dev 03:24:35 I could see that if DCSS had a standard place to install things and if the server always looked there first and opened a socket communication where the server controlled the world and passed information to all of the client sockets Round-Robin fashion... 03:24:39 <|amethyst> I mean, their own copy of crawl running server-side 03:24:50 <|amethyst> the only thing that runs on the client is some javascript 03:25:07 Oh. That's what causes the lag, then.... 03:27:21 Loop-->Generate Monsters-->Move monsters-->Move turn--> Player command interrupt ---> Place interrupt on queue--> Round Robin execute interrupts --> Loop,,, err something like that? 03:28:48 That seems like a lot of information to keep track of. Mostly in the dungeon generation process when a player is moving down to a new floor. 03:29:41 50 players moving to a new floor at the same time == "a not insignificant amount of stuff to process" 03:30:42 That's really cool! It's interesting to hear how this is implemented. :) 03:32:13 If online play becomes really popular in the next year or two, and you have an order of magnitude more players on the server; it will probably be unfeasible... 03:33:22 -!- category has quit [Ping timeout: 246 seconds] 03:33:38 crawl has had online play for many releases now 03:33:41 fun fact, server power is cheaper than hiring coders/engineers by an order of magnitude 03:34:07 <|amethyst> crawl runs like it normally does locally, except in addition to sending stuff to the screen it also sends those JSON messages to a socket that it uses to communicate with the webtiles web server 03:34:27 <|amethyst> the web server forwards those packets on to the player and all the watchers 03:34:55 <|amethyst> but, yes, 50 people loading Depths:$ at the same time would be a lot of load 03:37:10 <|amethyst> Crawl itself is while (true) _input() 03:38:07 I've got to get a server! I need to figure out a way to make money to afford it... I could pick up the skills to learn to run a server quicker-first. Learning this kind of thing should open up all kinds of doors in the tech world 03:38:14 thank goodness the input function is staticalized 03:39:24 ABS, A-Always B-Be S-Staticalizing. Always Be Staticalizing! 03:39:45 -!- its_jenna has quit [Ping timeout: 272 seconds] 03:40:35 -!- ts__ has quit [Ping timeout: 264 seconds] 03:43:08 -!- giantbat has quit [Quit: Leaving] 03:47:42 -!- tupper has quit [Ping timeout: 245 seconds] 03:56:00 -!- Insomniak has quit [Remote host closed the connection] 03:57:07 -!- NTRAFF has quit [Quit: Page closed] 03:59:23 -!- scummos| has quit [Ping timeout: 272 seconds] 03:59:26 -!- bmfx has quit [Ping timeout: 252 seconds] 04:03:11 -!- Diabl0658 has quit [Ping timeout: 272 seconds] 04:04:26 Anyways, I will need to study your python scripts in-depth after I figure out how tornado works. I had trouble with the module the last time I tried; I have advanced a bit more since that time, so this time it should be better for me. 04:05:42 Thanks for talking about your experiences with servers. 04:07:33 -!- Azzkikr has joined ##crawl-dev 04:07:48 is the maintainer of the develz server present? 04:08:26 He/she might have gone to sleep. There were two: amethyst and johnstein. 04:09:32 Ok, thanks, I'll check back later then. FYI the reason I wanted to contact either is that selecting v0.14 gets me a v0.13 game on that server 04:09:45 -!- archaeo has joined ##crawl-dev 04:09:54 -!- Basil has quit [Ping timeout: 255 seconds] 04:10:28 -!- TAS-2012v has quit [Read error: Connection reset by peer] 04:10:31 Oh, I remember you mentioning that issue earlier. 04:10:39 Do you have access to the git version? 04:11:23 What kind of an OS are you running? 04:13:07 -!- paxed has quit [Ping timeout: 240 seconds] 04:13:58 -!- ZRN has quit [Read error: Connection reset by peer] 04:14:52 <|amethyst> Napkin: looks like CDO still has the 2013 dgamelaunch.conf but the 2014 menus, so trying to run 0.14 gives you 0.13, 0.13 gives you 0.12, etc 04:15:30 heh 04:16:23 re-login, try again, please 04:17:02 <|amethyst> looks good 04:17:08 It's kinda funny: If one could get crawl running on a GPU, in theory, you could have access to around 800 scalar processors each handling different jobs simultaneously for the price of roughly $400. I have no idea if anyone has actually built that kind of flexibility into gpu computing... 04:17:12 thanks, |amethyst 04:17:17 <|amethyst> thanks, Azzkikr 04:18:45 <|amethyst> Aeleron: it would be easier to rewrite Crawl in assembler than to rewrite it to use GPUs I think :) 04:18:52 That would require a massive overhaul of the code, however... it would be a *very* difficult project... 04:19:05 <|amethyst> to use GPUs for most of the stuff, that is. Local tiles uses the GPU of course 04:19:22 <|amethyst> but not any custom shaders or anything like that 04:19:50 <|amethyst> I guess rewriting crawl in assembler is trivial: g++ -S 04:19:57 <|amethyst> Haskell, then, let's say :) 04:20:34 It would be really useful if there were a way to build an abstraction layer around a typical application designed for a single cpu and to get it to run and self organize on a gpu. A library that could do that would be... beautiful. 04:20:55 <|amethyst> a library that could do that would be a massive feat of AI 04:21:29 I think the main problem would be the way that the memory is typically handled on a GPU... 04:21:31 -!- paxed has quit [Ping timeout: 240 seconds] 04:21:33 <|amethyst> since even the best programmers have a hard time doing that; doing it automatically just isn't feasible 04:22:20 <|amethyst> also, most tasks aren't easily parallelisable 04:22:41 -!- paxed has quit [Changing host] 04:22:59 <|amethyst> For example, in crawl, you couldn't make monster actions run in parallel, because each depends on the results of the previous ones 04:23:23 <|amethyst> I mean, you *could*, but it would end up working differently from current Crawl 04:23:51 Well, I was thinking about it from the perspective of: You currently have 8 CPUS-1Server. 04:24:33 <|amethyst> Aeleron: I think I'd be better off using the GPUs to mine bitcoin and using that to pay for more CPUs :) 04:25:04 If you could instead treat it as: 8 SPs-100 Servers where each server acts independently doing whatever your server normally does. 04:25:09 <|amethyst> not saying a highly parallelisable roguelike is a bad idea 04:25:19 <|amethyst> but I wouldn't try to graft it onto an existing game 04:25:36 <|amethyst> software needs to be designed from the ground up to take advantage of that kind of stuff 04:26:12 Well, bitcoin; I saw the way that those things scaled. It's a factorial function. The greatest value for your processing time has already been mined. 04:26:58 I'm fairly certain that we are going to see that particular economy collapse pretty soon. I might be wrong... 04:27:21 <|amethyst> yeah, it probably won't be long before ordinary GPUs are too slow to make a profit mining bitcoin 04:27:32 <|amethyst> maybe we already reached that point 04:27:40 <|amethyst> we definitely have for CPUs 04:27:41 Yep. 04:28:20 Unfortunately, the greatest value that you get for parallel computing is a linear speed up based on the # of processors... 04:28:51 It gets better if you have a multi-layered set of processes that operate in logarithmic time. But bitcoin mining is not one of those... 04:29:14 <|amethyst> bitcoin mining is embarassingly parallel 04:30:49 <|amethyst> (which is as good as it gets) 04:31:58 Sorenson Sieve gives you: O(N/((log N)^Llog log N) complexity. 04:33:28 Embarrassingly parallel with Sorenson on M processors gives you: O(N/(M*(log N)Llog log N) 04:34:28 That doesn't look so hot... when N gets huge. The yield curve has already passed it's hump... 04:35:54 Then again; fiscally conservative as I am, I really wouldn't want to touch bitcoins with a 10 meter pole... but that's just me... 04:35:55 <|amethyst> Sorenson also described a parallel sieve in O(log n) time using O(n/(log n * log log n)) processors 04:36:19 -!- bd has quit [Ping timeout: 240 seconds] 04:37:38 Let's see what that curve actually looks like... 04:37:40 <|amethyst> embarassingly parallel would mean you could do it in O(1) time using O(n) processors 04:38:59 -!- ruwin has quit [] 04:39:13 Let's say that you had $1e6. 04:39:37 One high power GPU costs maybe $1e3. 04:40:02 $1e3~=1e3 SPs 04:40:16 <|amethyst> SP? 04:40:22 Scalar processors. 04:40:56 Which means that 1e3 GPUS per Fermi-cluster. 04:41:32 Which means 1e6 SPs per $1e6. 04:41:57 I'm tired and I don't trust my head for math right now, so spot me if I do something dumb... 04:43:10 I'm not sure how big max{N} is practically for bitcoins at the moment. 04:43:44 http://blockchain.info/charts/total-bitcoins 04:43:47 <|amethyst> A K20 is ~$3200 for ~2500 shader threads 04:44:14 <|amethyst> so that sounds about right depending on what you mean by "scalar processor" 04:45:04 |amethyst, Napkin : thanks, its fixed :) 04:46:17 I'm pretty sure we are at a point right now where an individual who is doing mining is getting swamped by those who have *much* bigger pocketbooks. I'm pretty sure that at least one moron on wall street thought that they could make a quick buck with a high-risk, high-reward investment in mining... 04:46:41 <|amethyst> People are using custom hardware to do their bitcoin mining these days 04:46:55 <|amethyst> you can get by with GPGPUs but probably not for much longer 04:47:20 Compared to their resources, individuals are just going to get crushed by the riptide of those with bigger pocketbooks. 04:47:59 But that's just my impression; I only really eyeballed the math... 04:48:58 <|amethyst> This is true of pretty much any economic activity though :) 04:49:13 <|amethyst> economies of scale favour consolidation 04:50:30 Yeah, I guess you're right. I think that with a high powered GPU one could get bigger bang out of their buck by picking up shading languages and using those skills to build something to sell to the masses; than wasting that computational power on bitcoin mining. But I may be wrong... 04:52:48 Or doing the banking thing with compound interest. There is more interesting math in that direction... it's been making people rich for over 500 years. :P 04:53:33 -!- Azzkikr has left ##crawl-dev 04:53:38 You've got to know what your doing with that though. An incorrectly designed automated algorithm could make you lose more than your shirt... 04:56:37 The problem with bitcoin mining is: A guy who has 1e12 processors is going to generate far more coins far faster than the guy with 1e3 processors... And unless you can pick your range of numbers to test and pick a sweet spot with a high density of primes; you are not going compete. Period. Full stop. 04:57:41 <|amethyst> The problem with investing is: a guy who has $1e12 is going to generate far more money than someone with $1e5 in their retirement account 04:57:58 <|amethyst> that doesn't mean you should keep your money under the mattress :) 04:58:14 -!- minmay has joined ##crawl-dev 04:59:20 <|amethyst> but neither bitcoin mining nor investment banking is something I'm particularly interested in, or likely to do 04:59:23 Plus the value of those coins that you "mine" is highly volatile. They might be worth $600 a piece one moment, and less than a couple of cents at another. 05:00:11 Plus the algorithmic complexity assures a limited supply sometime in the near future... 05:00:46 <|amethyst> I don't think the market is quite that volatile: if it goes to a couple of cents it will stay there because no one will re-enter the market 05:01:03 <|amethyst> and that's better than Zimbabwean dollars :) 05:01:11 lol :) 05:02:56 The thing is; I think that bitcoins were intentionally designed as an inverted pyramid scheme; they just have a little encryption "twist" to get nerds like us interested. 05:03:40 Again: Highest yield near the beginning of the mining process; Significantly diminishing returns as time goes on... 05:04:01 <|amethyst> Yes, that was intentional. 05:04:46 When mining visibly no longer produces significant returns; won't interest in the currency wane? Won't faith in the currency wane as well? Without faith in the currency; it's value will collapse? 05:05:01 <|amethyst> Aeleron: the idea is that you switch to transaction fees 05:06:12 -!- bmfx has quit [Quit: Lost terminal] 05:06:35 I'm pretty sure that whoever came up with the idea is definately holding the biggest stack of cards. When mining no longer produces many more coins; I am not sure what will happen to the liquidity of the currency. 05:06:49 -!- st_ has joined ##crawl-dev 05:08:24 <|amethyst> well, the reward halves every four years 05:08:35 <|amethyst> computing performance doubles more quickly than that 05:08:44 <|amethyst> *but* electricity efficiency maybe not 05:09:01 <|amethyst> s/maybe // 05:09:08 Hmm. Yeah. Cooling is become a significant problem, or so I've read... 05:10:43 What kind of thing do you do for a living? 05:11:00 <|amethyst> teaching CS, and doing bioinformatics 05:11:16 Grad student? 05:11:26 <|amethyst> not for a few years 05:11:33 <|amethyst> not faculty though 05:11:39 Tenure track? 05:11:41 <|amethyst> no 05:11:53 Oi. That is not a fun position to be in... 05:12:14 Err, unless it's a postdoc... 05:12:32 <|amethyst> Not exactly a postdoc, but similar in scope I guess 05:12:53 <|amethyst> I'll probably be looking for a full-time position in the next year or so 05:12:56 Temporary-ish research assistant, then? 05:13:13 How many publications? 05:13:37 <|amethyst> My title is "IS technical support specialist IV" 05:13:45 <|amethyst> err 05:13:55 <|amethyst> maybe "specialist" wasn't in there 05:14:03 <|amethyst> I have no idea what it's supposed to mean 05:14:07 Actually, that *is* a pretty good position to be in 05:14:12 <|amethyst> well 05:14:12 -!- rast- has joined ##crawl-dev 05:14:14 <|amethyst> I am grant-funded 05:14:28 On another PI's grant? 05:14:33 <|amethyst> which is a pain in the current fiscal climate 05:14:43 -!- scummos| has quit [Ping timeout: 240 seconds] 05:14:44 <|amethyst> Aeleron: yes, several different grants over the years 05:15:12 -!- rast-- has joined ##crawl-dev 05:15:44 <|amethyst> The grant I previously was on expired... 6 hours ago I think. But there's another for the next year anyway. 05:15:47 -!- rast has quit [Read error: Connection reset by peer] 05:15:49 -!- rast-- is now known as rast 05:15:52 Yeah, I'm not sure how we are going to get out of this financial quagmire. I don't think that we've felt the strongest effects of the sequester. 05:17:04 <|amethyst> eh, if it gets bad enough there will be other places to move 05:17:12 <|amethyst> as an academic I mean 05:17:28 <|amethyst> places that actually fund the sciences 05:18:24 Somehow we are going to need to generate more value and more trade. The reason why the unemployment rate is so high right now is because most of the major corps rely on credit stream to hire people with. After 2008; the credit stream went from a "flooding" state to a slim trickle. Hence our unemployment rate... 05:18:58 *Oi, that was redundant. 05:19:08 -!- rast- has quit [Ping timeout: 260 seconds] 05:20:27 High unemployment rate --> depressed individual income --> less buying --> slimmer credit stream potential --> lower unemployment ---> very bad feedback loop. 05:22:13 -!- nooodl has joined ##crawl-dev 05:22:30 It's interesting: because of the way we are handling our research at the University level at the moment, it's keeping our PIs distracted with constantly trying to generate more grant revenue. 05:23:50 If in the 60s, the average PI had to spend 5% of their time in grant funding activities; over time if the average PI ended up needing to spend 60% of their time generating grant revenue through a process of generating ever higher amounts of publications... 05:24:30 <|amethyst> I would not count writing publications as "time generating grant revenue" 05:24:43 <|amethyst> time writing grant applications, on the other hand :( 05:25:05 Higher # of publications --> more time spent generating lower quality research --> PIs spinning their wheels pumping out ever more marginal research. 05:25:50 ---> less time spent fixing the internal structure which is causing this leak in the fundamental structure of how we do research. 05:26:43 The funny thing is... this process of generating funding is also eating into the time of our politicians. They are stuck in the same position as PIs. 05:27:01 Except the formula looks a little different. 05:27:37 A Senator or a Representative spends much of their time trying to generate the revenue needed to run for their next campaign... 05:27:49 <|amethyst> /nick rentseekingbehaviouralist 05:28:31 That means that there is less time spent actually trying to govern because that time is monopolized by revenue seeking activities. 05:29:08 I doubt very much that this system evolved this way by design... 05:29:13 -!- twelwe has quit [Ping timeout: 246 seconds] 05:29:47 The system that we live in is pretty complex... 05:30:08 <|amethyst> but in general those in power have to expend resources to maintain their power 05:30:22 If we think of our activities as being dictated by rates that control our behaviors.... 05:30:26 -!- Philonous has quit [Read error: Connection reset by peer] 05:30:28 <|amethyst> it's been that way since the beginning of life :) 05:31:00 -!- Stelpa has quit [Read error: Connection reset by peer] 05:31:23 That means that our system could abstractly be described as one great big honging system of partial differential equations; with a twist thrown in... 05:31:31 *honking 05:31:43 -!- bonghitz has quit [Remote host closed the connection] 05:31:58 <|amethyst> That's what economists do for a living, at least the more mathematically inclined ones 05:32:00 Rates and new systems can be generated by spawning more rates and more subsystems.. 05:32:25 <|amethyst> but to have a DE you can solve you need to simplify a lot 05:32:31 <|amethyst> and even simple DEs have chaos 05:32:37 <|amethyst> s/have/can have/ 05:32:37 Feedback loops develop, evolve, and change.... 05:33:08 <|amethyst> Aeleron: now you're starting to sound like Hofstadter's conception of the mind 05:33:14 <|amethyst> :) 05:33:42 So, an analogy: If our world economy works on rates, systems with subystems that generate more subsystems:: I can think of another system that does this... 05:34:03 The human body constantly generates and repairs cells. 05:34:25 But if some bug in the system generates the wrong kind of feedback loop; a cancer begins to develop and kills the host... 05:34:43 Can such a thing happen economically on a worldwide scale? 05:40:03 <|amethyst> I think Marx had something to say about that :) 05:40:40 Yeah, well those who tried to follow in his footsteps managed to cause a mass starvation in Russia. :P 05:41:20 <|amethyst> They couldn't compete with the global capitalist system 05:41:25 they were explicitly going their own way, though, not following anybody's footsteps 05:41:50 economicscrawl: the most dismal crawl 05:42:15 Same thing happened to the folks who tried to follow in his footsteps in China. To this day news stations in China try to avoid portraying an empty bowl, because people have bad memories of what happened in the 50's. 05:42:28 -!- flowsnake has quit [Ping timeout: 244 seconds] 05:44:01 I just wish that I knew how to fix it. I would kind of like to see a Jetsons scenario as an old man... 05:45:20 Portal Projectile doesn't work with missiles fired via 'F' 13https://crawl.develz.org/mantis/view.php?id=8749 by Sprucery 05:45:20 if you figure it out, let somebody know 05:45:35 not all of us are brilliant academics like |amethyst, who can just follow the winds of science 05:46:19 Yeah, well, I've got to find a solution to my own rent-seeking-behavior, too. 05:46:54 <|amethyst> I'd be happy enough just not to have 30s-style nationalism come back to power within my lifetime 05:47:05 <|amethyst> or within the next 10-20 years :/ 05:47:29 hi what is this conversation ???? 05:47:32 Yeah, the golden dawn party is spooky. You know Greece is in the same position that Germany was in right before WWII. 05:47:48 nooodl: welcome to ##econ-dev 05:47:53 <|amethyst> nooodl: About running a crawl server 05:47:53 -!- flowsnake has joined ##crawl-dev 05:48:16 Sorry, we kinda got off on a weird tangent... 05:55:46