01:34:35 <Gretell> Unstable branch on crawl.develz.org updated to: 0.27-a0-74-gebd156f8b4 (34)
01:55:30 <Gretell> Windows builds of master branch on crawl.develz.org updated to: 0.27-a0-74-gebd156f8b4
02:16:07 <Cbrotell> Unstable branch on cbro.berotato.org updated to: 0.27-a0-74-gebd156f8b4 (34)
02:49:48 <Cheibriados> Autofight moves player towards monster behind runed door. 13https://crawl.develz.org/mantis/view.php?id=12434 by Yermak
02:53:51 <Gretell> Monster database of master branch on crawl.develz.org updated to: 0.27-a0-74-gebd156f8b4
03:30:53 <Kelbell> Fork (bcrawl) on crawl.kelbi.org updated to: 0.23-a0-3840-g9efd504fd6
03:40:00 <ebering> this clan name is borderline: https://crawl.develz.org/tournament/0.26/clans/kobold-kuckold-kings-vidiiot.html
03:40:09 <ebering> no, it's over the line
05:07:52 <Cheibriados> Net Duping 13https://crawl.develz.org/mantis/view.php?id=12435 by FakerFangirl
05:48:10 <Cheibriados> Formicid doesn't move after activating Dig if key was pressed fast. 13https://crawl.develz.org/mantis/view.php?id=12436 by Yermak
08:23:23 <Cheibriados> Some/all enemies missing in level examination with X ] [ 13https://crawl.develz.org/mantis/view.php?id=12437 by hittemvvvhard
09:39:13 <advil> ebering agreed
09:41:15 <advil> I guess our options are, ban the clan name, ban the player, anything else?
09:44:39 <advil> I don't know anything about that player except that the username is vaguely familiar-ish
09:50:20 <gammafunk> they're a long-time player. I've spoken with them a few times
09:50:20 <gammafunk> hrm, one sec
09:51:37 <gammafunk> actually, yes, they're a pretty awful troll, also active in the MR discord
09:51:40 <gammafunk> I would recommend a ban
09:52:16 <advil> that's fine with me
09:57:29 <advil> right, I do remember noticing their team name last time as ... odd, but not crossing any lines that I could detect
09:58:44 <gammafunk> I'm setting up ban now
10:22:06 <ebering> I think the name alone is a sufficent breach of good manners for the ban; no need to bring up discord memberships if there's an appeal just to avoid kicking the hornets nest harder than necessary
10:26:25 <gammafunk> It's not just membership, but activity/chat history (and not only in one discord). Won't state that as the reason for the ban if there is an appeal, like you say, but will make it clear that it's a mitgating factor (for me, as a tournament organizer) when deciding.
10:27:37 <gammafunk> In any case, we can't actually revert a ban even if there is an appeal, not without completely rebuilding the DB
10:28:05 <gammafunk> CDO is much faster now, but not so fast that we'd take the tournament down for many hours just to rebuild for one user
10:28:20 <gammafunk> On that note, I'm not sure if we do have a viable backup plan for a situation where we *do* want to rebuild the DB
10:28:52 <gammafunk> Old CDO just couldn't handle it in any way for anything more than a day of scores, not sure about new CDO, I but I do know that it wouldn't be lightning-fast
10:29:05 <gammafunk> Just judging from how quickly it processed 24 hours of data
10:29:20 <gammafunk> for something like 3 servers, and now we have all of them
10:30:30 <gammafunk> Hopefully it just won't come up, but I'm not sure what we'd have to do, maybe provision something on AWS with enough CPU to handle it. Our scripts are just not very optimized.
10:31:16 <advil> someday I can return myself to that 2 month fever dream where I optimized cao scoring
10:31:33 <gammafunk> hehe
10:32:38 <gammafunk> I was going to add a sentence that I'm not qualified to say where our scripts are optimized or not, just sort of took that for granted given how the old scripts behaved
10:32:47 <gammafunk> things are possible better with the new backend?
10:33:22 <advil> I suspect the biggest problem is shared with old scoring, here's my original diagnosis: "In general, a problem with logfile reading is that there are many, many
10:33:23 <advil> individual INSERT statements per logline (interspersed with other queries in some cases)."
10:36:15 <ebering> I think that issue is somewhat improved with new scoring
10:36:15 <ebering> I tried to base as many queries as possible off of the style I used for csdc scoring
10:36:26 <ebering> which has a 1:1 insert per log-line
10:36:26 <advil> does it use bulk inserts?
10:36:31 <ebering> yes
10:36:38 <advil> 1:1 is still pretty bad :-)
10:36:39 <ebering> well, csdc does
10:37:07 <advil> though on the scale of a tournament might be fine
10:37:30 <advil> some day I guess someone should do some actual profiling, I think I would have wasted a lot of time on cao scoring without that
10:37:32 <ebering> csdc used a feature of the sqlite driver that can do smartly optimized bulk inserts w/o developer interaction using its transactional system
10:37:52 <ebering> I spent a while arguing with MySQL EXPLAIN PLAN
10:38:12 <ebering> there are some comments in the schema where a query is written in a funny way because the optimizer couldn't figure it out
10:38:35 <ebering> for new tournament scoring
10:55:26 <advil> here's a brainstorming question: what would be the most expedient way to get a complete list of ttyrec files on a server for games that meet a certain criteria? (where the list is likely long)
10:55:46 <advil> sequell has this info but I don't think there's a direct way to get it out except on a fairly small scale?
10:57:01 <advil> (just trying to do some more emergency disk clearing on cao)
10:57:25 <gammafunk> hrm
10:57:45 <gammafunk> can you even link the files to game conditions at all? I guess you have the player name, since it's in that directory
10:58:03 <gammafunk> but the ttyrec filename is just the timestamp of when the recording began, so no actual connection to game logic, right?
10:58:19 <gammafunk> only way sequell knows is via a directory listing, as I recall
10:59:15 <gammafunk> presumably it has some logic about "file with the nearest timestamp before the milestone/logfile entry"
10:59:15 <advil> yeah, that's basically the problem
10:59:19 <advil> right, it's logic is something like that iirc, I looked into it once
10:59:38 <advil> so in principle sequell could probably be adapted to generate this list
11:00:28 <gammafunk> yeah, get a formated output with the timestamp of the entry in question, grouped by player
11:00:55 <gammafunk> then parse this list to do two splits, one overall, and an inner split to derive the player name and the set of timestamps
11:01:12 <gammafunk> then iterate over those and find nearest ttyrec for player p with even at time t
11:01:33 <gammafunk> but this is also just kind of error-prone (which sequell's ttyrec lookup is also error-prone)
11:02:06 <gammafunk> sequell is pretty willing to spit out many lines of output, so you could get quite a lot of results per query if limited to cao
11:02:45 <gammafunk> I guess you could just process the logfile/milestone file yourself
11:03:13 <advil> I have to admit I've been hoping there's an obvious readymade way to do this that I haven't thought of
11:03:31 <advil> though I do still have NotSequell kicking around
11:05:01 <advil> one of my various ideas is to delete all the most trivial ttyrecs on cao, which are not individually large but there's a lot of them
11:05:08 <advil> !lg * cao turns<100 -ttyrec
11:05:11 <Sequell> 564467. wrhowe, XL1 MiBe, T:62: http://crawl.akrasiac.org/rawdata/wrhowe/2021-01-11.15:12:09.ttyrec
11:06:10 <advil> hopefully even someone who really believes everything should be archived is not going to be too fussed about those getting deleted
11:06:57 <gammafunk> yeah, could be a good emergency thing for now, maybe have something a bit more selective later
11:08:57 <advil> like even a 62 turn game is still 66393 bytes
11:09:13 <advil> !lg * cao turns<100 x=avg(turns)
11:09:18 <Sequell> 564467 games for * (cao turns<100): avg(turn)=28.45
11:09:43 <advil> some back-of-the-envelope calcs suggest that is 20-30G right there, which currently would matter
11:10:44 <advil> *but*, it entails getting a list of 564k files
11:32:17 <advil> lol the rc file directory for 0.10 is 1.2G (similar for many older versions)
11:32:49 <advil> I wonder if deleting those would break things? iirc it might
11:34:38 <gammafunk> any RC file directories for pre 0.10? those could be deleted without any problems!
11:35:00 <advil> I just found a bunch of stuff for those versions and archived them onto the other drive, that was a few GB
11:36:27 <advil> oh, it looks like it autocreates them when you log in
11:37:09 <advil> so deleting the stock ones for these versions is probably safe
11:38:31 <gammafunk> brb implementing randomtiles for 0.10
12:08:15 <Henzell> Unstable branch on crawl.akrasiac.org updated to: 0.27-a0-74-gebd156f (34)
12:50:34 <Discord|> <12e​bering> hm. The combo high scores might not be handled well in the case of a ban, iirc those are computed sequentially. ugh
12:51:21 <advil> (for context, that is re somone on discord asking, question. If a player gets banned from the tournament and they had a combo highscore should it get removed too if possible?)
12:56:00 <Discord|> <10P​leasingFungus> oof, that is not a good clan name
12:56:23 <Discord|> <09g​ammafunk> *was not :wfrTeeth:
12:56:54 <Discord|> <09g​ammafunk> (I can't use my discord emotes from irc :whitehEye: )
12:57:41 <Discord|> <10P​leasingFungus> lol
13:57:59 <advil> %git d68fa41eda39
13:57:59 <Cheibriados> 07wheals02 * 0.26-a0-1197-gd68fa41: Defer redrawing the wield display if travel_delay = -1 (#1626) 10(2 weeks ago, 1 file, 19+ 22-) 13https://github.com/crawl/crawl/commit/d68fa41eda39
13:58:29 <advil> I can't figure out how the content of this commit matches the commit description
13:58:39 <advil> am I missing something?
13:59:03 <advil> it looks like it just refactors some bools in a way that is sensible but won't change behavior
14:03:09 <advil> wheals: I wonder if there was supposed to be more to this commit that got left out accidentally?
14:04:43 <Discord|> <03w​heals> look at the start of _print_stats_wp, for example: the function cancels out immediately if you're travelling, but before the change it would still set you.wield_change to false
14:05:25 <advil> ah I see
14:05:26 <Discord|> <03w​heals> so when you finish your travel, it would have no way of knowing that it needs to do a redraw that it's been putting off till now
14:06:08 <advil> thanks
14:06:36 <Discord|> <03w​heals> no problem
14:06:44 <Discord|> <03w​heals> it was pretty tricky to track down in the first place
14:07:18 <advil> there's apparently still weird issues with autotravel redraw
14:08:03 <advil> of the weapon
14:08:04 <advil> I wonder if that initial condition should be removed, I added it pretty recently and it was more by analogy than for concrete reasons
14:08:54 <Discord|> <03w​heals> is it the inventory letter disappearing? i think i remember noticing that back in 0.25 too
14:08:58 <advil> yeah
14:09:06 <advil> did  not know that was old
14:09:27 <advil> ok I don't post this to make fun of this player but I'm pretty entertained by this newquiver reaction "guys, i somehow mapped my f key to cast sting somehow, and cant fire my crossbow. How can i reset it?"
14:12:16 <advil> if it's old my immediate guess is that something else is overwriting the letter with spaces
14:26:10 -!- aidanh_ is now known as aidanh
14:27:52 <advil> ah, timeout on the tournament script isn't working because the function throws an IOError that isn't caught except in the global loop, so `reinit` restarts each time
14:30:24 <advil> also, the default number of retries is 20, so 60x20 = 20 minutes
14:35:39 <advil> maybe 20s with 3 retries is better?
14:36:55 <advil> ah
14:37:41 <advil> "The default is to retry 20 times, with the exception of fatal errors like "connection refused" or "not found" (404), which are not retried."
14:37:41 <advil> so probably the current cbr2 issues are tripping that, whereas past issues haven't
14:37:42 <advil> that would explain why I haven't much noticed the 20 retry things on cao
14:49:12 <advil> ok, I've restarted the tournament scripts with better timeout handling
14:49:47 <advil> it's possible that there will be order issues; the way this worked I think it may have been pulling + processing milestones from cao, but not any later servers, because cbr2 is second
14:57:58 <advil> I set it at 30s timeout with 3 retries
14:59:30 <advil> can someone remind me what the proper way to stop the t scripts is? I think I did it wrong
15:02:06 <Discord|> <12e​bering> iirc the correct way to stop them is to kill them
15:02:11 <Discord|> <12e​bering> sigterm is fine
15:03:50 <advil> that seemed to leave a lock file behind
15:04:11 <Discord|> <12e​bering> it does, I just cleaned that up manually in my local testing
15:04:30 <Discord|> <12e​bering> grepping in the code I think the correct way is a python repl importing the crawl_utils module and calling write_taildb_stop_request
15:06:45 <advil> haha deja vu https://github.com/crawl/scoring/commit/895c576b884da7e24
15:07:55 <advil> I guess I could port that over, the stop request file stuff seems very similar
15:16:04 -!- Vaevictu1 is now known as Vaevictus
15:21:54 <ebering> =nchoice VpVM
15:21:56 <Sequell> Defined keyword: nchoice => HaEn|FoSu|PaWn|GhDe|BaAE|MfDe|DEAr|PaGl|MuAM|VSCj|TeAK|VpVM
16:04:19 -!- cjm_ is now known as cjm
16:27:01 <Discord|> <03w​heals> @advil it seems like autofight_wait was broken? i press shift-tab with Pa charge quivered, and it says there's nothing in range, rather than waiting
16:27:06 <Discord|> <03w​heals> (with a monster in view but out of charge range)
16:32:02 <Cheibriados> Gozag gold detection reveals entire contents of stack 13https://crawl.develz.org/mantis/view.php?id=12438 by Siegurt
16:32:45 <advil> if you are using the default bindings, CMD_AUTOFIRE doesn't respect autofight_wait
16:32:47 <advil> I could make it do that, but didn't
16:33:14 <Discord|> <03w​heals> shift-tab is CMD_AUTOFIRE_NOMOVE though, right?
16:33:24 <advil> no longer, though you can rebind it to that
16:34:04 <advil> (CMD_AUTOFIGHT_NOMOVE)
16:34:20 <Discord|> <03w​heals> and that's unbound by default?
16:34:23 <advil> yeah
16:36:24 <advil> CMD_AUTOFIRE is also different in that it fires in melee range
17:26:11 <Cheibriados> 03advil02 07* 0.27-a0-75-g9f9dfd1: Fix a console hud issue at high turncounts (gressup) 10(4 minutes ago, 1 file, 10+ 11-) 13https://github.com/crawl/crawl/commit/9f9dfd16e913
17:30:02 <Cheibriados> 03advil02 07[stone_soup-0.26] * 0.26.0-8-g74fc34d: Fix a console hud issue at high turncounts (gressup) 10(8 minutes ago, 1 file, 10+ 11-) 13https://github.com/crawl/crawl/commit/74fc34ddfba3
17:32:33 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.27-a0-75-g9f9dfd16e9 (34)
18:25:18 <Lantell> Unstable branch on underhound.eu updated to: 0.27-a0-75-g9f9dfd16e9 (34)
18:28:04 -!- Tiobot is now known as Guest20996
18:56:48 -!- jfcaron_ is now known as jfcaron
20:48:25 <Henzell> buba (L18 OpFi) ASSERT(!invalid_monster(&mons)) in 'mon-death.cc' at line 1469 failed. (Elf:3)
20:49:15 <Henzell> buba (L18 OpFi) ASSERT(!invalid_monster(&mons)) in 'mon-death.cc' at line 1469 failed. (Elf:3)
20:53:47 <Lantell> Stable (0.26) branch on underhound.eu updated to: 0.26.0-8-g74fc34ddfb
22:33:07 -!- FutureShock is now known as Desertdew
23:10:04 <Cbrotell> dolemite99 (L20 MiFi) ASSERT(!invalid_monster(&mons)) in 'mon-death.cc' at line 1469 failed. (Elf:3)
23:10:36 <gammafunk> =nemelex HaEE
23:10:38 <Sequell> Defined keyword: nchoice => HaEn|FoSu|PaWn|GhDe|BaAE|MfDe|DEAr|PaGl|MuAM|VSCj|TeAK|VpVM|HaEE