02:58:07 Monster database of master branch on crawl.develz.org updated to: 0.23-a0-419-g0174fd8 06:19:58 Arcadio (L1 FoHu) ASSERT(you.equip[slot] == armour.link) in 'delay.cc' at line 772 failed. (D:1) 09:21:37 Stable (0.22) branch on underhound.eu updated to: 0.22.1-2-g9c5b331383 09:27:33 !crashlog Arcadio 09:27:34 1. Arcadio, XL1 FoHu, T:21 (milestone): https://crawl.xtahua.com/crawl/morgue/Arcadio/crash-Arcadio-20181112-111942.txt 09:28:34 o.O( 13:32:01 Which stats affect the damage/accuracy of crossbows? I am having a hard time looking through attack.cc and ranged-attack.cc 15:20:08 LexAckson (L12 BaFi) ERROR in 'tileweb.cc' at line 219: Socket write error: Resource temporarily unavailable (Lair:2) 15:20:19 Gumsho (L7 TeFi) ERROR in 'tileweb.cc' at line 219: Socket write error: Interrupted system call (Temple) 15:26:25 geekosaur: weird indeed 15:26:40 it'd be cool if crashlogs could print the most recent N keys you pressed 15:30:24 When rotted, only the players modified max HP is shown, not the base max HP 13https://crawl.develz.org/mantis/view.php?id=11759 by NormalPerson7 16:01:07 but i can't find a single unified place all keyinput goes through, which would be a nice place to add that keylogging 16:07:52 eg if you type Te, _get_next_cmd sees T but the e key is read as part of the removing-armour command 16:31:43 is crawl determinsitic at all 16:31:53 i know there are replays via console recorders 16:32:00 but could a replay file ever exist, as it currently is 16:32:13 sure would help with debugging 16:32:21 yes and there are tests that use it. you can specify an rng seed 16:38:33 well, don't use response files that I know of 16:52:43 it's deterministic but "unstable", i think? i don't know the right word for this. a single action performed differently alters everything that comes after it. compare to, say, binding of isaac, where the structure of each floor is fixed by the seed 16:52:44 in crawl, the randomness used by monsters mingles with the randomness used for building floors, so even with the same seed you won't get the same dungeon as another player unless all actions you perform are identical 16:54:34 but that qualifies as deterministic as far as replay files go right? 16:55:27 crawl has a number of issues related to how it generates levels that make it hard to guarantee identical level generation given a specific seed regardless of player action 16:55:42 even if you separate the RNGs out 17:05:42 |amethyst: johnstein: anything wrong with CAO right now? People on discord reporting lag spikes and it had some WebTiles failure messages randomly yesterday and the day before 18:44:44 it's not just a webtiles problem - i was able to reproduce lag pretty easily in console 20:46:04 -!- amalloy is now known as amalloy_ 20:53:04 gammafunk, are those issues documented? i'm curious 20:53:10 replays would be dank 21:04:12 Temple branch being cloned into Abyss 13https://crawl.develz.org/mantis/view.php?id=11760 by Ultraviolent4 21:12:57 -!- amalloy_ is now known as amalloy 21:49:52 Bringing a conversation from ##crawl over here -- I've run into an interesting Elf:2 map, where there are literally no permanent staircases into the main area (including the Hall of Blades) from either Elf:1 or Elf:3. I'm on CAO if anyone's interested enough to examine the map. 22:05:16 so where do the staircases go? a disconnected segment of E:2? 22:08:47 I hope Tapin hasn't vanished. I looked at the game and confirmed. They supplied https://pastebin.com/mCxcLxRL - all the non-hatches are in the explored area. 22:09:08 Sorry, still around just lagged to hell 22:09:18 Yes, three separate disconnected areas 22:09:48 I eventually broke down and /oDig'd into the main area, which I'm cautiously exploring right now when my lag lets me 22:14:31 ...and now I'm gone for a little while, my train has arrived at the station. Happy to answer questions/etc later if anyone's got any more. 22:36:11 -!- mike_ is now known as Guest90846 22:45:10 are there any documents related to the single sign on idea? 22:46:49 e.g. what why when where or how 22:49:01 floraline: not really anything well developed, no 22:49:29 are there any notes or comments at all? 22:49:47 maybe the irc logs will have something useful 22:49:51 a few conversations in this channel over the years 22:50:32 some issues include preventing games from being unplayable when the signon server is down for whatever reason 22:50:55 and how servers might want to be able to cache account information to that end 22:52:24 there are probably also issue related to ownership and administation of any SSO server 22:52:29 *issues 22:53:30 historically we have server admins run and be responsible for their own servers, and for other kinds of services, they're often hosted and maintained by some dev who's implemented and hosted the service; mostly talking about something like sequell 22:54:38 it's possible for one person to implement the technology and perhaps a server admin/dev could take responsibility for hosting it, but that depends on what the person implementing a SSO service wants 22:55:09 and I suppose it also depends on what server admins want; what are we going to require of them if they plug into this system? 22:56:09 but on a purely technical level I think discussion has been pretty high level, mostly just musing about what might be a good approach and how do we make sure that it doesn't prevent games from being unplayable if it has to go down for some length of time 23:07:26 gammafunk: what are the reasons why we would want to use it? 23:08:04 would it really be necessary to have a full typical SSO implementation? 23:08:42 floraline: if nothing else it sucks that someone else can register an account on (for example) CPO with your username and lose games, on purpose or accidentally 23:09:03 one of the main issues is having to register your account on each server both as you play on each server and in order to protect your scoring information from griefers 23:09:17 right, that's the only point of pain that i'm aware of 23:09:33 If it's not too Captain Obvious, hardfought (NetHack) has a master server; you register there and push accounts to every other server, which is simpler and there's no single point of failure for playing, only account creation. 23:09:33 i wonder if there is some other clever design that would solve that problem 23:09:47 ^ like that.. that makes a lot of sense 23:10:02 the issue is only really at the time of account creation 23:10:34 & (as you know, Professor) CAO is already a single point of aggregation for scoring so that would be an obvious place for it 23:11:05 don't agree that it's an obvious place for it for at least a couple reasons 23:11:16 but does this server also coordinate password resets etc? 23:11:40 gammafunk: Hardfought? Yes. 23:11:40 I think another issue related to all of this are the scoring pages themselves 23:12:08 the CAO scoring pages are basically unmaintained in terms of fixing bugs, certainly in terms of adding new functionality, and are pretty fragile 23:12:26 we sort of badly need to rebuild the database due to past logfile corruption on e.g. cjr 23:12:30 but doing so takes a really long time 23:13:22 so floraline might be right that a simpler idea that only handles account registration could fix most major problems, but we also have a big problem with the scoring pages, I guess 23:13:32 other devs might have some other opinions/input 23:14:23 re: CAO, r.ax owns this server, doesn't have free time to devote to upgrading it, and it also hosts a (somewhat fragile) webtiles server that's pretty important in terms of overall traffic 23:15:14 she may be willing to give access to someone working on a new registration service but depending on what the needs are of this service it may be impractical to say the least to develop, test, and run it there 23:15:37 ... forget the suggestion of CAO, I didn't know all that 23:15:39 I was always imagining it running on some other server, and preferably scoring might run elsewhere as well 23:18:51 one other thing that's useful to think about for longterm planning is things like account registration filtering and possibly providing some very basic server administration capabilities to block malicious users from new registrations on an IP level 23:19:05 it's kind of a hassle for admins to manage that stuff now, I think 23:19:22 this might just need to be something that's part of webtiles, but it might relate to any account registration system 23:20:50 Account registration filtering does raise an awkward question for centralised registration a la hardfought; suppose there is a problematic user who I'll call MR, but only a subset of server admins think MR is persona non grata. (I think the benefits will greatly outweigh this, but I can see potential for a spat over that nevertheless). 23:22:16 right, that as far as registration goes, I was more thinking of just "offensive username"; for server access that probably is something we should leave to each server admin 23:23:41 however for scoring pages/sequell that's a service where perhaps there could be something like a code of conduct with users denied access. however historically we've really just left that up to whoever runs the service 23:24:20 just like how we've let server admins run their servers as they see fit so long as they don't make logfile/milestone data that's bad or do something horrific in terms of content they host, I guess 23:25:35 And yeah, I'd like to see more ability to manage access to some of those services but it's not that simple, since again individual devs/admins develop and pay for these services 23:26:12 But filtering out usernames containing offensive language seems like something where we could get agreement 23:26:38 I think I'm worrying more about the opposite scenario where centralised account registration means you have de facto control but this vexes server admins who after all have de jure control 23:28:53 Right, well that's why I say it's not so simple to control access to those services even if most devs would like to have something basic in place there, conduct-wise 23:31:52 Something like preventing registration of usernames containing slurs seems like the best case for there being general agreement between server admins and devs 23:39:52 i feel like the centralized account creation still allows server admins to have full control of their system 23:40:29 admins could even optionally allow 'offline' accounts to be created that are only local to their server and don't take part in scoring 23:42:55 Suppose central sends "create", "change password", "destroy"; server account data structure mostly unchanged. Server admin can just create an account that doesn't exist centrally; central scoring daemon ignores scores for accounts that aren't on its list (not a big change to the existing scoring daemon, I think). Server admin can delete a centrally created account they personally don't like, that's 23:43:01 their fiat as server admin. 23:44:40 So frex if I find a bunch of dodgy words in Korean and register all these accounts centrally, the Korean server admin can jump on them just as fast as they can now. 23:45:56 frankly the scoring server could be totally separate from the account creator 23:46:06 and whatever is responsible for creating accounts may not even need to keep a roster of accounts itself 23:46:34 it would help this discussion to understand how scoring currently works 23:46:52 It could but (assuming scoring moves off CAO) it seems convenient to put them in the same place 23:46:55 there are two ways of scoring, sequell and the CAO pages 23:46:55 ebering: I think I do? 23:47:15 i definitely don't 23:47:16 the account creator 23:47:17 23:46 < floraline> and whatever is res 23:47:21 this is currently the case 23:47:30 oh arg a line came under my copy buffer 23:47:40 the scoring server is currently totally separate from accounts 23:47:44 yes 23:48:16 it should stay that way 23:48:45 I run my own scoresd and had a look at it when I did 23:49:26 floraline: Well, the scoring knowing what accounts "exist" is an obvious way to keep locally created ("offline" in your terminology above) games out of the scoring, because the scoring just grabs the logfile from every server and smooshes them together. 23:49:27 CAO pages are a set of python scripts that manage a postgres database and pull from official server logfiles (and milestone files), updating player/overview/category html pages based on data already in the database and on the new data recieved 23:50:24 sequell does a similar thing but his its own implementation in perl/ruby and is hosted on a greensnark's server 23:50:43 So if "Local200" exists on (say) CXC but not elsewhere and isn't meant to be scored, either a) CXC can post an expurgated logfile or b) the central scoring can know what accounts exist and that "Local200" isn't one of them. 23:50:54 and of course it's providing an irc front-end to said data, not generating static webpages 23:51:57 anyway I for one look forward to signing into crawl with my Yahoo! account 23:52:10 as we'll definitely standardize on them for 3rd party auth 23:52:16 oh and we have tournament scripts that we use every tournament, which work similary to the CAO scoring pages but also have their own implementation 23:53:03 Arguably Sequell might continue to provide results for such accounts anyway 23:53:31 incorrect, what will happen is that Yahoo! will be switching over to use WebTiles for its account registration/authentication next year 23:54:03 soon webtiles chat will be available on all major social media platforms 23:54:36 finally I can webtiles chat from my Okurt 23:56:42 wow, I only learned just now that Google+ is shutting down