00:02:33 03advil02 07[stone_soup-0.24] * 0.24.1-1-gd1d2a07: Work around a seed stability issue in regret_index_temple_of_mockery 10(4 minutes ago, 1 file, 51+ 29-) 13https://github.com/crawl/crawl/commit/d1d2a07e1b2f 00:03:06 03advil02 07* 0.25-a0-730-g633e9c3: Work around a seed stability issue in regret_index_temple_of_mockery 10(5 minutes ago, 1 file, 51+ 29-) 13https://github.com/crawl/crawl/commit/633e9c3c14be 00:04:45 we tagged it for the security vuln I think but I didn't do a release 00:04:54 ah I see 00:12:28 advil: https://imgur.com/ek62wKD.png 00:13:32 hm maybe reload? 00:13:49 spectating you I don't see those glitches 00:14:27 Yeah, I just wanted to stop and preserve the state in case it was relevant. 00:15:05 there were a couple of minor webtiles changes that could lead to temporary visual glitches like that, maybe 00:15:49 I think it's just in the ui, not the game itself 00:17:45 Unstable branch on crawl.kelbi.org updated to: 0.25-a0-730-g633e9c3c14 (34) 00:17:47 though I'm not sure what's with the spurious chat msg 01:31:22 yeah not being able to restart webtiles without disconnecting games is going to become a problem 01:32:28 ordinarily you could just have multiple instances and do a rolling deployment, but that doesn't work with long-lived socket connections 01:33:29 not to mention that webtiles stores a bunch of stuff like logged in state in memory, which wouldn't work with multiple instances 01:36:00 maybe webtiles instances being drained can notify the client js, which would transparently reconnect to the new socket on the replacement instance 03:21:43 Unstable branch on crawl.beRotato.org updated to: 0.25-a0-730-g633e9c3 (34) 03:55:10 Stable (0.24) branch on crawl.kelbi.org updated to: 0.24.1-1-gd1d2a07e1b 03:56:00 Fork (bcrawl) on crawl.kelbi.org updated to: 0.23-a0-2999-g24818c1c7b 03:58:18 Fork (bcadrencrawl) on crawl.kelbi.org updated to: 0.22.1-2574-g139baa18eb 04:12:08 or some way to start new games with a new process and give old games like 5mins to finish up 04:58:50 wtbh i'm not that fussed by this limitation 04:58:54 and fixing it seems Hard 05:19:48 -!- Tiobot is now known as Guest87744 06:48:31 Forums seem to be having a Moment 06:48:52 Perhaps CDO in general 10:19:35 it looks like cdo rebooted, and most services didn't start up automatically? 10:20:22 possibly including napkin's irc bouncer 10:21:39 I don't have permissions to start apache 10:22:30 crawl is still working though! 11:25:18 Napkin is taking a look at CDO 11:40:41 alexjurkiewicz, yeah that sounds straightforward; I suspect the hard part would not be making clients reconnect to the socket, but things like dis/reconnecting ttyrec recording 11:45:29 but even that simpler method would require some form of orchestration on the server 12:15:17 I was thinking about whether something like this would be possible at one point, but my experience so far is that every time I restart the webtiles server there's non-trivial downtime while I sort something or other out 12:16:08 if we were in a position where the server was regularly restarted without problems it would be nice, just not sure that's where we're are 12:17:15 and it's a bit hard to imagine the multiple container failover type of situation any time soon :-) 12:17:41 saving is actually pretty transparent in a way, just requires a login, so all our current system might really need to do is come up with a way for token-based login to keep working across a restart 12:18:09 and of course figure out why cpo crashes the games 12:19:18 i.e. rather than a live failover thing, it might be better to figure out a smoother save + reload process where the smoothing is mainly in the browser rather than on the server 12:21:14 right; something like a temporary popup in the browser while it tries to reconnect, and the server reopens all crawl games on start up? 12:23:35 CDO services are back up; Napkin says that he'll take CDO dgl game server offline at some point due to lack of version updates and general interest 12:24:22 Cdo is SSH only, right? 12:24:25 yes 12:34:39 03advil02 07* 0.25-a0-731-g0472a16: Suppress builder dprf spam from main message window by default 10(11 minutes ago, 3 files, 44+ 10-) 13https://github.com/crawl/crawl/commit/0472a16d6ab8 12:34:59 We found ourselves using it during the CXC interregnum so could be tricky for Europeans if CXC goes again 12:34:59 re login across restarts, the traditional solution is to store tokens in a database instead of in memory 12:35:30 I wonder to what degree napkin is tired of running all of this stuff 12:41:23 advil: well, he did mention that he's busy with IT stuff and has limited motivation; I'm sure he's ok with running the website, but would probably be happy to be less responsible for things 12:44:57 definitely +1 to less dprf spam for debug builds 12:47:26 at one point I dreamed of turning the message log into an accordian list 13:01:16 is there any documentation on triggerables besides what's on the dev wiki? 13:01:35 I am finding this API a little bit baffling 13:03:22 Unstable branch on crawl.kelbi.org updated to: 0.25-a0-731-g0472a16d6a (34) 13:07:05 hmm, I always use debug-lite simply to avoid the dprf spam, and step through with a debugger when needed 13:10:06 advil: didn't even realize that page existed! I think I've used them in various vaults, and always just looked at other examples 13:10:34 yeah that's more or less what I did for the thing I fixed yesterday 13:11:04 definitely seems like one of those APIs that is tailored towards the most complicated of complicated things you could want to do 13:12:23 gammafunk: do you happen to know why a triggerable still gets triggered after you remove the marker it is associated with? Or even, does that seem like a bug? 13:13:16 hrm, how are you removing the marker? 13:13:27 I recall vaguely that there is a way to deactivate markers 13:13:46 I tried `triggerable:remove(marker)` where triggerable is the 2nd argument to the callback 13:13:52 that's from an example on the wiki 13:13:55 gotchat 13:13:57 *gotcha 13:14:02 I also seem to recall that a lot of those marker triggerables just set state to disable subsequent callbacks 13:14:08 I tried to find cases in the code that did this, but everything I found sets custom state 13:14:17 yeah 13:14:20 as in the callbacks examine this state and do nothing under the right conditions 13:14:21 that's what I ended up doing 13:15:04 I also tried setting `repeated` to be false, though tbh I never really figured out whether that param is what it sounds like 13:15:44 with both removing the marker and with that, the callback continues to run, and I got errors caused by the callback running with a removed marker 13:15:46 %git %git d1d2a07e1b2f 13:15:46 Could not find commit %git d1d2a07e1b2f (git returned 128) 13:15:50 %git d1d2a07e1b2f 13:15:50 07advil02 * 0.24.1-1-gd1d2a07: Work around a seed stability issue in regret_index_temple_of_mockery 10(13 hours ago, 1 file, 51+ 29-) 13https://github.com/crawl/crawl/commit/d1d2a07e1b2f 13:15:55 very cargo-cult code in there 13:18:32 oh, interesting form of seed stability bug 13:18:32 yep 13:18:32 !source lm_trig.lua:39 13:18:32 https://github.com/crawl/crawl/blob/master/crawl-ref/source/dat/dlua/lm_trig.lua#L39 13:18:32 I need to do a canvas of other des things to see if there's more like it 13:18:58 not sure if this comment will lead to an answer to your original question 13:19:21 hmm 13:19:38 I have to admit when I first read that my eyes glazed over on the master/slave stuff, but yeah it does seem like it should answer 13:21:00 though I still don't know what a marker being "removed" entails 13:21:26 could just be that ""the master marker being automatically removed when all slaves are gone." is not working though 13:28:08 Stable (0.24) branch on crawl.akrasiac.org updated to: 0.24.1-1-gd1d2a07 13:28:20 also a very basic question, is it possible to have triggerables without a marker? seems like no? 13:29:38 yeah, not sure how you'd run an arbitrary lua function at any time that wasn't vault placement without a marker 13:30:28 well, I was wondering because there's all sorts of things you can do with these that don't rely at all on being tied to a specific square in the dungeon 13:30:34 can a single location have multiple markers? 13:30:38 yes 13:30:41 ok good 13:31:34 so I can just stick a pile of markers at 1,1 or something if I want to ignore needing to be tied to a location 13:31:38 markers are classified into types, and iirc you can have multiple markers of each type at a given position 13:31:52 and can get a list of all markers of a given type at a given position 13:48:35 Unstable branch on crawl.akrasiac.org updated to: 0.25-a0-731-g0472a16 (34) 14:58:09 aha I bet the trigger remove thing I was asking about is a bug, that is caused by having multiple instances of the same marker, because remove works perfectly when there is just one 14:59:10 or maybe multiple markers with the same triggerable instance 15:03:41 oh, that's possible? each instance of the marker doesn't make its own object? 15:04:32 hola 15:09:54 well well well, if it isn't pizza tornado 15:10:53 gammafunk it seems to be possible, not sure whether it is intended to be posssible; it's yet another aspect of that code I copied from somewhere else kind of blindly 15:13:04 twelwe: we reworked magic but I lost the pt patch notes so I couldn't add pizza magic 15:13:07 sorry 15:17:37 (the real story is that ebering didn't want pizza magic to overshadow his own work) 16:16:35 i want to greet each and every one of you but take this message instead as gesture enough 18:02:26 -!- jfcaron_ is now known as jfcaron 18:22:40 Unstable branch on underhound.eu updated to: 0.25-a0-731-g0472a16d6a (34) 19:17:17 Difficult To Accept Beogh 13https://crawl.develz.org/mantis/view.php?id=12235 by Boozermonkey 19:22:37 Stable (0.24) branch on underhound.eu updated to: 0.24.1-1-gd1d2a07e1b 19:34:54 %git stone_soup-0.24 19:34:54 07advil02 * 0.24.1-1-gd1d2a07: Work around a seed stability issue in regret_index_temple_of_mockery 10(20 hours ago, 1 file, 51+ 29-) 13https://github.com/crawl/crawl/commit/d1d2a07e1b2f 19:35:25 %git 0.24.1 19:35:26 07Aidan Holm02 * 0.24.1: Update changelog 10(2 weeks ago, 2 files, 17+ 0-) 13https://github.com/crawl/crawl/commit/3008d5f674b2 19:35:30 ah, right 19:35:45 we just didn't make that one for a long time, but the security fix prompted aidanh to make it 19:36:00 so I guess I should actually go and make those builds 19:36:15 well, I guess we should make a 0.24.2 at this point 19:36:54 might be some other fixes in trunk we could cherry-pick by now as well