01:21:38 Experimental (bcrawl) branch on underhound.eu updated to: 0.23-a0-2457-gf0d5a3770a 01:24:33 -!- amalloy is now known as amalloy_ 01:48:33 Plant on stairs stopped auto-travel. 13https://crawl.develz.org/mantis/view.php?id=12068 by nubinia 02:23:56 03Aidan Holm02 07* 0.24-a0-641-g3ea5ddb: Remove support for multi-key macro binds 10(2 days ago, 1 file, 28+ 87-) 13https://github.com/crawl/crawl/commit/3ea5ddb52caf 02:23:56 03Aidan Holm02 07* 0.24-a0-642-g5a8975c: Simplify macro code 10(3 days ago, 1 file, 2+ 3-) 13https://github.com/crawl/crawl/commit/5a8975c3e223 02:23:56 03Aidan Holm02 07* 0.24-a0-643-g678ee33: Refactor _getch_mul() -> _getch_single() 10(2 days ago, 1 file, 9+ 13-) 13https://github.com/crawl/crawl/commit/678ee33c32bb 02:23:56 03Aidan Holm02 07* 0.24-a0-644-g049a785: Simplify macro_buf_add_with_keymap 10(2 days ago, 3 files, 10+ 9-) 13https://github.com/crawl/crawl/commit/049a78546b28 02:25:02 New branch created: format-fixes (3 commits) 13https://github.com/crawl/crawl/tree/format-fixes 02:25:02 03Aidan Holm02 07[format-fixes] * 0.24-a0-645-g9eab960: Remove unused and rename used fs_op members 10(13 hours ago, 3 files, 9+ 12-) 13https://github.com/crawl/crawl/commit/9eab9609ca65 02:25:02 03Aidan Holm02 07[format-fixes] * 0.24-a0-646-g1ce4416: Remove some formatted_string functions 10(12 hours ago, 4 files, 4+ 58-) 13https://github.com/crawl/crawl/commit/1ce4416f7569 02:25:02 03Aidan Holm02 07[format-fixes] * 0.24-a0-647-g3819d2a: Reduce unnecessary text widget allocations 10(12 hours ago, 3 files, 9+ 0-) 13https://github.com/crawl/crawl/commit/3819d2a53e4e 02:38:20 Unstable branch on crawl.kelbi.org updated to: 0.24-a0-644-g049a78546b (34) 02:48:42 Monster database of master branch on crawl.develz.org updated to: 0.24-a0-443-g80245de 03:31:58 Fork (bcrawl) on crawl.kelbi.org updated to: 0.23-a0-2457-gf0d5a3770a 03:55:36 The build passed. (format-fixes - 3819d2a #11913 : Aidan Holm): https://travis-ci.org/crawl/crawl/builds/589299064 05:04:52 -!- misterostroum is now known as Mister_Ostroum 09:24:24 Stable (0.22) branch on underhound.eu updated to: 0.22.1-59-ge965b3aba3 10:10:41 macro.cc:720:23: warning: braces around scalar initializer [-Wbraced-scalar-init] macro_buf_add({key}, false, false); 11:13:27 CBRO seems to have choked, can't start games 11:24:29 Stable (0.23) branch on underhound.eu updated to: 0.23.1-89-g277498a001 11:25:03 ^status 11:25:12 * geekosaur wonders if that even works these days... 11:46:18 03Aidan Holm02 07* 0.24-a0-645-g0e779d2: Fix some compile warnings (geekosaur) 10(3 minutes ago, 3 files, 5+ 3-) 13https://github.com/crawl/crawl/commit/0e779d289c51 11:50:31 Crash while trying to fast-walk downstairs 13https://crawl.develz.org/mantis/view.php?id=12069 by jackofblades 11:50:35 03Aidan Holm02 07* 0.24-a0-646-g0627d6e: Remove unused and rename used fs_op members 10(22 hours ago, 3 files, 9+ 12-) 13https://github.com/crawl/crawl/commit/0627d6e13b0d 11:50:35 03Aidan Holm02 07* 0.24-a0-647-g6ad1058: Remove some formatted_string functions 10(22 hours ago, 4 files, 4+ 58-) 13https://github.com/crawl/crawl/commit/6ad10583fd2f 11:50:35 03Aidan Holm02 07* 0.24-a0-648-geb4d282: Reduce unnecessary text widget allocations 10(21 hours ago, 3 files, 9+ 0-) 13https://github.com/crawl/crawl/commit/eb4d28254491 11:52:29 yeah, cbro is out of disk 11:53:02 03Aidan Holm02 07* 0.24-a0-649-g0b1efcb: Revert "Fix some compile warnings (geekosaur)" 10(17 seconds ago, 3 files, 3+ 5-) 13https://github.com/crawl/crawl/commit/0b1efcbfd730 11:59:15 03Aidan Holm02 07* 0.24-a0-650-gd158fba: Fix some compile warnings (geekosaur) 10(16 minutes ago, 3 files, 5+ 3-) 13https://github.com/crawl/crawl/commit/d158fba778ae 12:05:59 !tell rax sorry to keep bugging you but just want to send another ping about scoring, any chance you could grant access (or stare it yourself)? 12:06:00 advil: OK, I'll let rax know. 12:06:13 or |amethyst for that matter, though I know you haven't been around much 12:12:02 Unstable branch on crawl.kelbi.org updated to: 0.24-a0-650-gd158fba778 (34) 12:25:17 I wonder what is involved resource-wise in hosting scoring? disk space is probably high, but I wonder how bad the cpu is 12:29:57 advil: I think there could be high cpu when you rebuild the db 12:30:26 ??rax 12:30:26 rax[1/3]: Barracks, cmon, do I have to tell you guys everything? 12:30:30 ??rax[2 12:30:30 rax[2/3]: don't !tell; just email rachel at akrasiac dot org 12:30:42 ah 12:30:44 advil: re pinging rax, she doesn't really check !messages 12:31:43 and yeah I'd be happy to do some admin for cao as well 12:40:18 ok, I sent an email 12:42:27 i'm running a copy of scoring for cko and i just put it on a single cpu 12:42:27 i don't think it's very resource hungry on a daily basis 12:42:33 is yours just your server? 12:42:36 yeah 12:42:42 It's when it rebuilds afresh it takes forrrrrever 12:42:54 just one server and only like six months so far, it's at 3.5 gb 12:43:13 when does it have to rebuild from scratch? 12:43:22 there's a slightly absurd amount of milestones in the CAO one, so it might be different 12:43:33 apparently any time something goes wrong? 12:43:53 I'm planning on spending a bit of time trying to better understand the scoring and do some incremental improvements later this fall 12:44:00 floraline: Server reboots are the obvious example. Even Gooncrawl scoring takes a long time then, and they're not playing _so_ many games. 12:44:22 you mean when the game server reboots, or the scoring server? 12:44:30 When the scoring server reboots. 12:44:46 it just has to play catch-up though, right? 12:44:57 afaict it has to read in the whole thing 12:45:06 seriously? 12:45:08 It reads in the whole thing. 12:45:09 that's surprising 12:45:12 because milestone files just keep getting added to 12:45:21 trunk milestone files for some servers are massive 12:45:29 i thought it keeps track of where in the file it has scanned up to 12:45:54 yeah, I was hoping to look into some improvements along these lines 12:46:06 It does normally but this doesn't help it when scoresd starts afresh. Presumably it _could_, but it doesn't. 12:46:17 i see 12:46:24 IIRC the CAO scoring daemon was taking literally days to restart last time there was an issue 12:46:41 there are a bunch of cases where it has needed rerunning because the db became inconsistent somehow, and for those it might actually need to read everything 12:46:52 yeah that makes sense 12:47:06 it sounds like the whole concept of these milestone files should be revisited 12:47:13 indeed 12:47:22 the db is inconsistent right now due to game duplication from e.g. cjr 12:47:49 knowing they just continue to scale forever is sort of scary 12:47:49 but like all of these bits of infrastructure it's *incredibly* painful to change them 12:47:53 green.snark has a solution for sequell where he checks a hash of any logfile/milestone line he would add 12:47:55 yes, they make a lot more sense relative to the infrastructure of crawl circa 2008 12:48:09 and if it matches any existing hash, it doesn't process this line 12:48:36 this is necessary because due to previous admin errors, some logfiles contain lines that are duplicated in other logfiles 12:50:26 cbro trunk milestones is 2.3G for example 12:50:51 you can see this on my scoring page; starting from win 53 I have duplicated wins, and some of my wins are duplicated 5 times 12:51:04 Presumably something else could perform this check and append to sanitised logfiles that scoresd would look at 12:51:42 advil: how would you split up the files? 12:51:51 or I guess, what other kind of format would be better 12:52:15 the most minimal change I have come up with is simply to allow milestones for a version to be spread across multiple files and rotate them periodically 12:53:11 rotate as in delete data that's sufficiently old? 12:53:11 it doesn't seem too technically hard, but I haven't done a *lot* of feasibility checking 12:53:45 just spilling over into a new file is all i would ask for 12:54:00 then you could move them into way cheaper long term storage 12:54:01 rotate as in once a year (or whatever) copy the current milestone to milestones.2019 and start afresh 12:54:04 yeah 12:54:17 esp since milestones.2019 would never need to change once rotated 12:54:24 right 12:54:44 (rotate in the since of unix system log rotation) 12:54:52 *sense 12:54:56 And that would make it more feasible to process it once and shove the results into the database in a way that scoresd could read quickly on startup 12:55:02 yes, that too 12:55:16 good point 12:55:21 this is my hope and dream 12:55:26 you should never have to parse the whole file again 12:56:58 I guess unix log rotation will eventually cycle around, so I don't mean that 12:57:11 so archive rather than rotate 12:57:29 yeah, that's what I thought of when you said that initially 13:10:29 speaking of log rotate, something is rotating my webtiles log file and i haven't been able to figure out what it is 13:10:52 would tornado be doing that itself? 13:13:51 The build passed. (master - d158fba #11917 : Aidan Holm): https://travis-ci.org/crawl/crawl/builds/589523673 13:15:28 Unstable branch on crawl.akrasiac.org updated to: 0.24-a0-650-gd158fba (34) 13:40:03 Pinkbeast: Poking around, I found this: http://status.berotato.org/ - looks like there have been ~0 users for ~350 minutes - almost 6 hours, and at the same time the disk usage reached 100% 13:40:18 I don't know who runs cbro or how to contact them 13:40:35 jolamb: From scrollback here you'll see it's a known issue now 13:41:36 03Aidan Holm02 07[format-fixes] * 0.24-a0-651-g4b16227: More refactoring of formatted_string code 10(43 minutes ago, 2 files, 9+ 9-) 13https://github.com/crawl/crawl/commit/4b16227417ee 13:41:36 03Aidan Holm02 07[format-fixes] * 0.24-a0-652-g46ecf6d: Remove unused function chop_tagged_string() 10(36 minutes ago, 2 files, 0+ 56-) 13https://github.com/crawl/crawl/commit/46ecf6da7544 13:42:47 Oh, I skimmed scrollback but didn't connect that the discussion pertained to cbro. Reading_comprehension-- 13:43:16 Most of it hasn't but "<+advil> yeah, cbro is out of disk" 13:47:19 Are there any container images for hosted webtiles/ssh Crawl? It would be interesting to build out templates so that folks could stand up all the infra etc. using kubernetes/openshift 13:49:31 jolamb: there have been a couple projects like that, but the problem has been that official servers don't use them, so someone spends a bunch of time making something and then it sits unused 13:51:40 gammafunk: That's a shame. I know there's more than one way to skin a cat, but it would be handy if e.g. cbro was running under OpenShift, and one could just allocate another persistent volume claim to its storage when it gets close to running out 13:52:07 full disclosure: I work on OpenShift lol 13:52:38 I'm not sure I understand why public servers seem to be so very short of storage to begin with 13:54:39 jolamb: I think the last person who got interested in this made a full setup based on ansible 13:54:50 jolamb: I think the difficulty with that is either a) a human does it which is not appreciably better than a human compressing ttyrecs or whatever else they do now or b) it happens without human intervention so can consume money in sorceror's apprentice mode 13:55:19 Pinkbeast: seems like both are problems, but "b" is the bigger problem 13:55:37 but yeah, our servers are a patchwork with different people owning the instances/servers and each having their own preferred approach and each having different amounts of free time to devote to management and deploying new/better technology 13:56:05 Right. So what I'm saying is, since a human would have to be involved, it wouldn't really represent an improvement to be able to allocate more disc vs to have to free some up 13:56:06 the admin of cbro, johnstein, has extremely little free time to work on improving cbro's setup 13:56:25 Are there any servers that implement HTTPS yet? 13:56:32 yes, many do 13:56:37 cpo, cko 13:56:42 cue 13:56:42 I guess if I was running an official server with this problem I'd write a cronjob to reap the oldest ttyrecs altogether 13:56:48 ??cue 13:56:49 cue[1/4]: European Crawl server, located in Amsterdam, Netherlands. http://underhound.eu:8080/ or underhound.eu, port 23, username: terminal, key: http://crawl.develz.org/cao_key http://crawl.develz.org/cao_key.ppk See {putty} for Windows users. Runs 0.10-trunk (DCSS, Zot, Sprint, Tut). Formely known as CLAN. 13:56:49 ??cko 13:56:50 cko[1/1]: Crawl server owned by floraline hosted in NY, USA. Webtiles at https://crawl.kelbi.org/#lobby or console via ssh to crawl.kelbi.org with port 22 and username 'terminal' and using CAO keys (found at https://crawl.kelbi.org/crawl/). 13:56:53 ??cpo 13:56:54 cpo[1/1]: DCSS server in Sydney. Run by chequers aka alexjurkiewicz. https://crawl.project357.org/ 13:57:02 hrm 13:57:30 yeah cue is using https 13:58:00 https://underhound.eu:8080/#lobby 13:58:00 i have a cron job that compresses player data and offloads it to object storage, which is way cheaper per GB than the disk on my VPS 13:58:04 i think cpo does the same thing 13:58:05 !learn edit cue[1] s/http/https/ 13:58:06 cue[1/4]: European Crawl server, located in Amsterdam, Netherlands. https://underhound.eu:8080/ or underhound.eu, port 23, username: terminal, key: http://crawl.develz.org/cao_key http://crawl.develz.org/cao_key.ppk See {putty} for Windows users. Runs 0.10-trunk (DCSS, Zot, Sprint, Tut). Formely known as CLAN. 14:03:22 with that automated, there is very little to do manually in order to keep disk space under control 14:03:22 huh, cko looks good to me 14:03:25 cbro would probably benefit from something like that 14:03:53 Yeah, it seems like using something like S3 for records would be nice and cheap 14:04:18 -!- amalloy_ is now known as amalloy 14:05:07 i'm using digitalocean's "spaces" which i believe is supposed to be their imitation of s3 14:05:54 yeah, johnstein just hasn't had a lot of time to do a lot of server revamping lately, he is interested in doing some more work eventually 14:06:20 ime most of the ways in which crawl infrastructure struggles are about time, not money or tech 14:06:55 most of these things were set up when someone had a lot of time, but then suddenly it's 8 years later and life is different 14:07:11 yeah, which is totally understandable 14:07:14 yeah 14:08:44 true that 14:09:57 ??docker 14:09:58 docker[1/2]: https://bitbucket.org/TZer0/crawl-docker - functioning prototype with external storage settings (and thus import) working. 14:10:00 ??dgl[3 14:10:00 dgl[3/5]: ansible version in development by Mattias at https://github.com/mattias/ansible_playbook_webtiles 14:10:08 ah, yeah, those are indeed the two projects 14:10:15 those are the last ones that I know about 14:10:38 advil: how are you feeling about tournament schedule, by the way? I'm probably not too far from trying a fedhas merge, just a couple smallish decisions I need to make. The end design will be a bit wonky, but there's some time for tweaking if we have a couple weeks after merge 14:10:45 tzer0 is the cue admin so it could even be that he's using that? 14:10:59 and then there are those tournament changes we had been talking about, which I haven't started on 14:11:39 I don't think I have anything major to try to do before release, perhaps some small tweaks to the seed interface 14:12:24 I haven't really had so much crawl time in the last month, it turned out 14:12:40 I know how it goes :) 14:12:57 We can certainly do a standard tournament this time if need be, all the infrastructure for that is set up 14:13:16 would be nice to at least try for some changes, even if small 14:13:20 sure 14:14:11 how does oct 25th sound for release/tournament? could do maybe a week earlier if need be, or certainly later is no problem 14:14:12 I don't really have anything major scheduled for the next couple months, so release/tournament soonish works for me 14:14:31 er 14:14:47 I'd like to merge my newgame-menus branch, at long last 14:14:51 dang 14:15:02 I assume you're not far from doing that though, aidanh? 14:15:15 oct 25 for release? or tourny 14:15:33 for both; we usually do same day; we could do release two days prior to 25th if you prefer 14:15:37 just needs some testing on windows console 14:15:42 I guess we usually do it before tourney these days 14:15:54 25th would be for tournament, since that's a friday 14:16:10 so beta somewhere around Oct 9? 14:16:17 if I'm remembering right 14:16:34 yeah, that would be fine 14:16:48 again, we can do later if that's too soon for people 14:17:49 I will be somewhat incommunicado from Oct 31 - Nov 5, but that's not a huge problem for the tournament really 14:18:33 yeah, sounds like you'd be able to do os x builds and maybe our ghost files an a minimum? 14:18:35 btw I got a very quick response from r.ax, thanks for pointing out that note about emailing 14:18:39 oh, great 14:20:24 that schedule should be fine, yeah 14:28:33 advil: ok, if you are alright with oct 25th tournament and (Say) oct 23th release, I can plan for that 14:28:44 s/S/s/ 14:28:46 yeah, that's fine 14:43:36 03Aidan Holm02 07* 0.24-a0-651-g4b16227: More refactoring of formatted_string code 10(2 hours ago, 2 files, 9+ 9-) 13https://github.com/crawl/crawl/commit/4b16227417ee 14:43:36 03Aidan Holm02 07* 0.24-a0-652-g46ecf6d: Remove unused function chop_tagged_string() 10(2 hours ago, 2 files, 0+ 56-) 13https://github.com/crawl/crawl/commit/46ecf6da7544 14:45:58 btw is there any interest in a better CoC for the tournament? I know this has come up before, CKO's (https://crawl.kelbi.org/crawl/code_of_conduct.txt) would be a place to start 14:46:57 advil: have you seen the current conduct? 14:47:40 it's really just based on the one ebering used for csdc 14:48:12 if you want to update it contain those parts from the one you linked regarding obscene names, advertising, that would certainly make sense 14:48:41 oh yeah I forgot we did that last time -- yeah the name rules were ones I was interested in adding at least 14:49:14 sure, sounds like a good thing to add 14:49:32 I extended the code to make banning users more feasible while the tournament is running 14:58:53 Unstable branch on crawl.kelbi.org updated to: 0.24-a0-652-g46ecf6da75 (34) 15:01:48 -!- Mister_ostroum is now known as Mister_Ostroum 18:23:10 Unstable branch on underhound.eu updated to: 0.24-a0-652-g46ecf6da75 (34) 19:07:11 jolamb: i'm interested in running crawl binaries and perhaps even the webtiles server in containers. I don't think k8s is the right solution though 19:07:27 The size of current deployments is on the order of 2-8gb total footprint 19:08:34 about scoring, there are some things that would make life much easier. Number one is definitely declaring bankruptcy on old game records and starting with a new serialisation/storage system 20:47:29 -!- amalloy is now known as amalloy_ 21:04:36 New branch created: pull/1156 (1 commit) 13https://github.com/crawl/crawl/pull/1156 21:04:36 03kitchen-ace02 07https://github.com/crawl/crawl/pull/1156 * 0.24-a0-653-g7637252: Change spell chances in Kiku's second book 10(14 minutes ago, 1 file, 5+ 29-) 13https://github.com/crawl/crawl/commit/763725249bc2 21:16:54 -!- amalloy_ is now known as amalloy 21:30:03 Coo, it turns out I'm one of gammafunk's "sycophants" :-) 21:48:25 -!- amalloy is now known as amalloy_ 22:58:44 Pinkbeast: is that in reference to some particular comment somewhere? 23:05:59 gammafunk: Reddit, Malcolm Rose - who else? 23:06:09 oh 23:06:47 well, fwiw, given how little we agree on various things, I gammafunk am formally declaring that Pinkbeast is *not* one of my sycophants 23:07:21 It did seem an unlikely scenario. 23:42:54 I mean, you're in the secret anti-MR discord that gammafunk set up 23:43:05 technically that counts as being a sycophant 23:44:17 I hope sycophanting at least comes with a free lunch or something