00:00:10 <|amethyst> elsewhere when we show 'extra' digits like that (I'm thinking HP and MP regen) there really is a hidden accumulator
00:00:31 <|amethyst> jfcaron:  the problem is that everything starts with base delay at an exact number of turns
00:00:38 <|amethyst> jfcaron:  and the maximum XL, 27, is odd
00:01:09 <|amethyst> so if something doesn't hit the 'statutory' mindelay of 0.7 before level 27, it's not going to hit an exact integer AUT delay
00:01:38 <|amethyst> err, maximum skill, not XL
00:02:07 <jfcaron> Does anyone really think displaying the weapon delay at 28 skill is a reasonable solution this dillema?
00:02:45 <|amethyst> I do admit it's kind of weird
00:03:21 <jfcaron> As a player it looks like a bug.
00:03:30 <|amethyst> and is slightly misleading in its own way because a player might think it means skill 28 is somehow reachable, with Ashenzari or something
00:03:36 <jfcaron> mhmm
00:04:15 <jfcaron> Would a solution be to make Sniper reach mindelay at 26 skill instead of 27?
00:05:20 <|amethyst> making level 27 not give a delay reduction would be an annoying special case, as would hardcoding its mindelay
00:05:31 <|amethyst> OTOH, we already have a special case for rapiers
00:06:31 <|amethyst> OT3H, rapiers are an entire weapon class and not two artefacts
00:06:37 <|amethyst> (this also affects Dark Maul)
00:07:09 <|amethyst> I would rather special-case the display logic than the actual delay calculation logic
00:07:20 <jfcaron> Srsly, showing 28 skill is ridiculous.
00:07:30 <|amethyst> maybe "This weapon's minimum attack delay (1.3 to 1.4) is reached at skill level 27
00:07:33 <|amethyst> "
00:07:48 <jfcaron> Yeah that'd be fine.
00:08:53 <jfcaron> Sorry for being so offended by the 28 skill thing.
00:08:53 <|amethyst> hm
00:08:53 <jfcaron> It looks like several systems got coded into a corner here.
00:09:19 <jfcaron> But it should be possible to modify the Sniper and Dark Maul to end up at a round number?
00:09:40 <|amethyst> jfcaron:  not without making their progression different from every other weapon
00:09:44 <jfcaron> Like not a special case at 27 but just adjust their base delay or the skill minimum.
00:10:03 <jfcaron> I mean how do regular triple crossbows avoid having this problem?
00:10:21 <|amethyst> jfcaron:  then the base delay would be not a round number of AUTs, which is even worse I think
00:10:54 <|amethyst> regular weapons reach mindelay at 26 or earlier
00:11:08 <jfcaron> Ok, so why not make them also reach mindelay at 26?
00:11:50 <|amethyst> because that would require making it quite a bit faster
00:12:28 <|amethyst> hmm
00:12:40 <jfcaron> I don't think I understand the constraints here.
00:12:57 <|amethyst> you could add a "mindelay" field to unrands, but there isn't one right now
00:13:24 <|amethyst> hm
00:14:25 <jfcaron> I need to go to bed.
00:14:25 <jfcaron> Sorry for ranting.
00:16:11 <|amethyst> I guess if we made the division by two in weapon_min_delay "... and unless it would take more than skill 27" round down instead of up, that would make those two weapons stop getting faster at XL 26
00:16:15 <|amethyst> err, skill 26
00:49:43 <spicycat> couldn't the discord bridge take those pipes and convert them into line breaks, maybe wrap the whole thing in ```
00:50:02 <spicycat> whoops, i was stuck scrolled up
01:22:12 <Lantell> Experimental (bcrawl) branch on underhound.eu updated to: 0.23-a0-3545-g7c84a495e7
01:49:18 <floraline> !tell gammafunk is beem still open source? and/or is the old gpl release still posted anywhere?
01:49:18 <Sequell> floraline: OK, I'll let gammafunk know.
03:12:00 <Rotatell> Unstable branch on crawl.beRotato.org updated to: 0.26-a0-598-g61e6210 (34)
03:30:55 <Kelbell> Fork (bcrawl) on crawl.kelbi.org updated to: 0.23-a0-3545-g7c84a495e7
05:18:10 -!- amalloy is now known as amalloy_
07:01:07 <alexjurkiewicz> when I introduced that description line, I felt like dark maul saying it reached min delay at skill level 38 (or whatever) was very funny and good
07:01:30 <alexjurkiewicz> but it's been complained about at least twice in here since
07:02:21 <Pinkbeast> Sniper ditto; I certainly find it confusing.
09:27:06 <advil> "This weapon's minimum attack delay (1.6) is impossible to reach."
09:27:52 <Pinkbeast> Suggest "impossible to reach; it would require XX skill. At 27 skill, the attack delay is n.m."
09:28:22 <Pinkbeast> rationale; distinguish Sniper (you can nearly get there) from Dark Maul; the user who is replete with XP actually cares what n.m is.
09:28:43 <advil> dark maul appears to be identical to sniper?
09:29:21 <advil> I think |amethyst's point is that the skill level 27 attack delay for both is a distribution over two aut values though
09:35:05 <Pinkbeast> Dark Maul> oh, is it? I think I believed "skill level 38" above.
09:35:16 <Pinkbeast> I last got it in 2017 or so.
09:35:53 <Pinkbeast> Even "n.m - n.m+1" would be better than not knowing
09:42:17 <advil> I take it you're in the pro-numbers contingent
09:42:48 <twelwe> pro numbers guy here. preference for even numbers
09:42:56 <Discord|> <08w​ormsofcan> "your attack delay will still be greater than 1"
09:54:39 <Pinkbeast> advil: Not as an ironclad rule, but you can find out mindelay for every other weapon.
09:55:28 <advil> well, right now you can find out mindelay for this one too!
09:55:28 <advil> you just can't get there
09:55:31 <advil> (I actually don't really see why people are annoyed by the status quo)
09:55:52 <Pinkbeast> OK; you can find out the minimum achievable attack delay for any other weapon.
09:56:54 <advil> maybe it's because I spend a lot of time dealing with modal logic
09:57:05 <advil> who cares what the actual world is like
09:57:12 <Pinkbeast> I dunno I can say much to that besides reiterating what I don't like about the status quo, which doesn't seem helpful because I already iterated it :-/
09:59:49 <advil> another way to put it is that I think it's pleasantly flavorful if there are a couple weapons that are so big that you can't know their exact achievable mindelay
09:59:50 <Discord|> <03w​heals> The perils of too many mathematicians on the devteam
10:00:24 <Pinkbeast> Ah, see, I'm a failed mathematician, that's why I don't agree.
10:01:00 <advil> if it were a common issue I'd basically agree that not showing it is bad
10:01:00 <advil> it might even be better if it were just one (dark maul, probably)
10:02:56 <Pinkbeast> How about "This weapon's minimum attack delay cannot be reached; it would require skill level 28."
10:03:14 <Pinkbeast> I think I remove my desire for "n.m at 27" now I know it's 28 for mindelay not 38.
10:03:47 <Pinkbeast> Er, "This weapon's minimum attack delay (1.6) cannot be reached; it would require skill level 28."
10:49:25 <Discord|> <09b​h> At least cross training is no longer incoherent
11:22:06 <twelwe> FR: noise level on evocables
11:46:11 <Naruni> twelwe: in the description of how loud they are or actually give them noise?
11:53:32 <advil> they have noise, so I'm guessing something comparable to the info you get about spell noise
12:08:07 <Cheibriados> New branch created: pull/1538 (1 commit) 13https://github.com/crawl/crawl/pull/1538
12:08:07 <Cheibriados> 03amcnicky02 07https://github.com/crawl/crawl/pull/1538 * 0.26-a0-599-gaada003: Add 3 new altar vaults for second round review 10(27 minutes ago, 1 file, 100+ 0-) 13https://github.com/crawl/crawl/commit/aada0030587e
12:19:23 <Discord|> <06a​dvil> I wonder why the discord github linking is adding in so many linebreaks
12:22:05 <Discord|> <06a​dvil> oh, maybe it's just that there are linebreaks in the PR comment which breaks text flowing
12:24:56 <Naruni> !source ability_menu
12:24:56 <Sequell> 1/1. https://github.com/crawl/crawl/blob/master/crawl-ref/source/options.h#L407
12:24:56 <Naruni> no, not ability, umm evo menu
12:25:32 <Naruni> !source prompt_invent_item
12:25:32 <Sequell> 1/1. https://github.com/crawl/crawl/blob/master/crawl-ref/source/invent.cc#L1868
12:26:42 <Yermak> > Those under Zin's wrath will find [...] their bellies emptied
12:57:12 <Yermak> Weren't food vaults removed? I just found infamous hornet autoexplore trap.
13:01:43 <advil> some of them I converted to non-food vaults, most bee-related things got food more or less swapped to ambrosia
13:02:30 <advil> not sure to what degree Real Vault Devs followed up on any of those changes
13:06:28 <Henzell> Unstable branch on crawl.akrasiac.org updated to: 0.26-a0-598-g61e6210 (34)
13:31:36 <Yermak> Pillars of salt got dying plants animation, this is probably not intended.
13:32:50 <Discord|> <10P​leasingFungus> yep, bug!
13:32:52 <Discord|> <10P​leasingFungus> oops
13:33:17 <Discord|> <10P​leasingFungus> i wonder if ball lightning does too
13:37:26 <Discord|> <12e​bering> Probably, maybe also foxfire and ice blocks
14:00:39 <Discord|> <10P​leasingFungus> i was thinking about those when i wrote the code, but i guess i messed up...?
14:08:52 <Discord|> <12e​bering> i'm not able to test atm
14:14:27 <Discord|> <10P​leasingFungus> samesies. maybe tonight
14:14:41 <Yermak> You can't wield non-weapons now. Is this a change I missed or some bug>
14:14:42 <Yermak> ?
14:40:44 <Discord|> <06a​dvil> afaik that's not something that's happened on purpose
14:43:16 <Discord|> <06a​dvil> I was just able to successfully wield a potion on cao trunk
14:44:09 <Discord|> <06a​dvil> there's a restriction against wielding armour/jewellery that I don't think is very new
14:45:01 <advil> !source player::could_wield
14:45:02 <Sequell> 1/1. https://github.com/crawl/crawl/blob/master/crawl-ref/source/player-act.cc#L372
14:46:30 <Yermak> advil, at some point in the game I couldn't wield anything, when I pressed * to choose a non-weapon, showed inventory had no letters.
14:47:21 <Yermak> Now I can't reproduce it.
14:49:45 <advil> were there forms involved?
14:55:16 <Yermak> no, I'm with Zin, most form-like action was trying to drink unknown !ligni which Zin protected me from.
15:46:01 <Naruni> !source memorize
15:46:02 <Sequell> Can't find memorize.
15:46:11 <Naruni> !source learn
15:46:11 <Sequell> Can't find learn.
15:55:46 <floraline> has anyone done the upgrade from tornado 2.4 on a standard dgl setup yet?
15:55:46 <Sequell> floraline: You have 1 message. Use !messages to read it.
15:56:05 <floraline> !messages
15:56:05 <Sequell> (1/1) gammafunk said (7h 8m 51s ago): beem and my other bots are closed source for now. You can search github for forks of beem from some point in the relatively recent past, and you'll also need to do the same for my 'webtiles' package.
15:57:09 <floraline> gammafunk: ok, i found some forks on github but that copy doesn't seem to run out of the box with python 3.8
15:57:25 <Discord|> <09g​ammafunk> oh, it definitely wouldn't yeah
15:57:40 <Discord|> <09g​ammafunk> you will need an older python, maybe 3.6, maybe a bit earlier
15:59:54 <Discord|> <09g​ammafunk> however I recall that the fixes to get that old source working with later are not too bad, maybe use of the (now in 3.6+) keyword async conflicting with a function having that name
16:03:54 <floraline> i resolved that one, but it seems to be misbehaving in other ways during runtime. it connects to the websocket but won't join any games
16:04:33 <Discord|> <09g​ammafunk> maybe you don't have autowatch enabled?
16:04:46 <Discord|> <09g​ammafunk> not sure if you have a pre-existing user db
16:05:09 <floraline> i've been tinkering with the config, i hope it's not something silly like that
16:05:13 <Discord|> <09g​ammafunk> sounds like you should verify that it has your user db or that it's actually both got autowatch working and has a reason to autowatch
16:05:34 <floraline> it does have my user in the db with subscribe flag on
16:06:17 <floraline> it's really weird though, the config loading happens first and verifies whether the db is correct. but it hasn't actually created the db yet so it always fails
16:06:17 <floraline> i had to disable some checks to get it to work
16:06:23 <Discord|> <09g​ammafunk> yeah, that's now how it was ever programmed to behave; it created db on fly if didn't exist
16:07:16 <floraline> i'm assuming there are some types with the watch_player vs watch_username?
16:07:51 <floraline> typos*, i think i fixed that but single user mode doesn't seem to work, and autowatch won't join my game
16:09:50 <floraline> i'll keep messing with it, it looks like it's running fine except that it just doesn't decide to join any games
16:10:03 <Discord|> <09g​ammafunk> yeah, I assume you actually have the proper number of spectators
16:10:43 <floraline> yeah i put in 3 spectators
16:10:52 <Discord|> <09g​ammafunk> and could be, re: watch user typos, as that feature hasn't been used much in recent years, since my setup doesn't require that so much for testing anymore
16:13:04 <floraline> webtiles is running on tornado 6.0.4, could that be a problem?
16:14:03 <floraline> if i have the subscribe flag in the db enabled, it should join my games without any spectators right?
16:18:57 <advil> floraline I wouldn't call cao standard but it's not really too far off, I moved it over to Tornado 4 before the 0.25 release (it can't go any higher without upgrading python)
16:20:21 <floraline> gammafunk: well, i reverted all my local changes and now it works fine right out of the box. i guess i'm just dumb
16:21:01 <floraline> advil: thanks, i'm trying to stay ahead of this looming deprecation of 2.4 so maybe i'll do the same
16:21:30 <advil> :-)
16:22:54 <advil> in principle it should be painless (easy to say I know)
16:23:07 <advil> since I used cao to work out a lot of the weirdnesses
16:25:40 <advil> one thing I would definitely suggest is using the --live-debug option on a different port to test things out beforehand, which will be easiest if you have a manual install of whatever tornado version you move to
16:26:16 <advil> if you are sticking with python 2 your options are either tornado 4 or 5, you could go to python 3 + tornado 6 if you are feeling brave
16:27:24 <advil> I would say that I suspect that tornado 5 is actually the slowest option, at that point most of their attention was on moving everything to asyncio rather than optimization
16:27:34 <advil> but they all should be more robust than hacked 2.4
16:29:06 <advil> hm, I guess I should incorporate some of these notes in this issue into the webserver docs
16:29:14 <advil> definitely before closing the issue
16:40:41 <floraline> i wonder if i should get to buster before dealing with that
17:03:15 <Naruni> advil: I've been trying to get beem into my spectators on CAO and haven't succeeded
17:04:25 <advil> @gammafunk will probably be able to help more than me
17:04:30 <advil> what's your cao username?
17:04:49 <Naruni> Naruni
17:05:42 <Naruni> also I've been reading menus trying to decide how to put noise in evocables menu and ugh it's a mess
17:06:13 <advil> I would consider putting in the item description to start with
17:07:10 <advil> @gammafunk also someone pointed out in one place or another (can't remember where or who) that the beem command guide is 404 -- I'm guessing maybe it linked directly to a github md?
17:07:10 <advil> maybe worth putting that up somewhere else
17:07:29 <advil> Naruni you're not currently playing on cao now, right?
17:07:42 <Discord|> <09g​ammafunk> yeah, I'm not fully sure what I want to do with that guide, but I may put a general bot guide on wordpress or something
17:07:52 <Naruni> advil: I have a game in progress but I am not logged into it.  I can do so if that would help
17:08:03 <Discord|> <09g​ammafunk> you'd need to log in for me to see a problem, yeah
17:08:21 <Naruni> gammafunk: ok I'm logged into a .25 game
17:08:24 <advil> if you've unsubscribed at some point in the past, I think you'll just need to run !subscribe somewhere that you can find beem while logged in
17:08:45 <advil> there it is
17:08:46 <gammafunk> beem is in your chat
17:09:16 <gammafunk> ah, nice
17:09:21 <gammafunk> found a but in the subscribe command
17:09:23 <gammafunk> I'll fix that
17:09:32 <Naruni> groovy, thanks
17:11:37 <advil> @gammafunk for a quick fix you could probably just put it up for a while as a gist, that supports md
17:15:16 <Naruni> advil: re: putting noise in description of evokables, going down that path I'm assuming eventually we get to a place where evocables will have noise and power similar to the spell menus?
17:15:46 <Naruni> *displayed to the player
17:17:25 <Henzell> deenic (L2 FeAE) Crash caused by signal #6: Aborted (D:1)
17:17:25 <gammafunk> Naruni: can you try !subscribe again?
17:17:35 <Henzell> dukeofspoot (L9 MiGl) Crash caused by signal #6: Aborted (Temple)
17:17:36 <Henzell> KShinobi (L14 MiBe) Crash caused by signal #6: Aborted (Orc:1)
17:18:15 <Henzell> luke0000015 (L3 FeEn) Crash caused by signal #6: Aborted (D:2)
17:18:15 <Henzell> adibis (L5 MfVM) Crash caused by signal #6: Aborted (D:2)
17:18:15 <Henzell> corvin (L8 DsBe) Crash caused by signal #6: Aborted (D:6)
17:18:15 <gammafunk> hrm
17:18:15 <Naruni> he left
17:18:15 <gammafunk> cao seems to be having issues
17:18:15 <Henzell> bipbip (L18 OgCj) Crash caused by signal #6: Aborted (Swamp:2)
17:18:15 <Henzell> adibis (L5 MfVM) Crash caused by signal #6: Aborted (D:2)
17:18:19 <Henzell> luke0000015 (L2 FeEn) Crash caused by signal #6: Aborted (D:2)
17:18:20 <Henzell> ir0nHorse (L12 FoMo) Crash caused by signal #6: Aborted (Lair:1)
17:18:21 <gammafunk> advil: anything wrong with cao that you can see?
17:18:31 <gammafunk> maybe a restart is happening?
17:18:33 <Naruni> looks like a buncha games just shit the bed too :)
17:18:33 <Henzell> Mulky (L5 VpBr) Crash caused by signal #6: Aborted (Sewer)
17:18:55 <Henzell> adibis (L5 MfVM) Crash caused by signal #6: Aborted (D:2)
17:19:05 <Henzell> bipbip (L18 OgCj) Crash caused by signal #6: Aborted (Swamp:2)
17:19:14 <Henzell> deenic (L2 FeAE) Crash caused by signal #6: Aborted (D:1)
17:19:30 <advil> looks like scoring is freaking out
17:19:32 <Henzell> luke0000015 (L2 FeEn) Crash caused by signal #6: Aborted (D:2)
17:19:35 <Henzell> ryryur (L7 SpAr) Crash caused by signal #6: Aborted (D:4)
17:20:12 <Henzell> Mulky (L5 VpBr) Crash caused by signal #6: Aborted (Sewer)
17:20:18 <Henzell> bipbip (L18 OgCj) Crash caused by signal #6: Aborted (Swamp:2)
17:20:18 <Henzell> tess (L11 GrEE) Crash caused by signal #6: Aborted (D:10)
17:20:21 <Henzell> deenic (L2 FeAE) Crash caused by signal #6: Aborted (D:1)
17:20:47 <Henzell> Euph0ria (L9 DrCj) Crash caused by signal #6: Aborted (Temple)
17:20:48 <gammafunk> Naruni: well if you can run that command when you see beem actually spectating, it might work
17:21:00 <Henzell> adibis (L5 MfVM) Crash caused by signal #6: Aborted (D:2)
17:21:01 <gammafunk> it keeps popping back into your chat
17:21:11 <Henzell> nmf (L15 DsVM) Crash caused by signal #15: Terminated (Lair:1)
17:21:17 <Henzell> jboogie (L13 TrMo) Crash caused by signal #6: Aborted (Lair:3)
17:21:18 <Henzell> TomBombadil (L10 MuNe) Crash caused by signal #6: Aborted (D:9)
17:21:21 <advil> hm, I don't think it's disk space but the chroot is at 99%
17:21:29 <Henzell> SecAgentSteve (L9 GrFi) Crash caused by signal #6: Aborted (D:8)
17:21:32 <Henzell> DekuNut (L27 DsMo) Crash caused by signal #6: Aborted (Geh:7)
17:21:39 <advil> I may need to restart the server here
17:21:41 <Henzell> TheOldMan (L12 PaFi) Crash caused by signal #6: Aborted (D:10)
17:21:48 <Henzell> 9to (L17 KoEE) Crash caused by signal #6: Aborted (Shoals:1)
17:21:54 <advil> there's zombie processes everywhere
17:22:27 <Henzell> Webtiles server stopped.
17:23:46 <Henzell> Webtiles server started.
17:24:46 <gammafunk> Naruni: not sure if cao is fully back, but you can try that again wrt beem
17:25:00 <Discord|> <06a​dvil> it should be back
17:25:06 <Discord|> <06a​dvil> I don't know exactly what happened though
17:26:31 <Naruni> gammafunk: ok it worked, he is watching now. thank you
17:26:49 <gammafunk> cool
17:28:15 <Discord|> <06a​dvil> I guess even at 99% there's still 19G free because this disk is gigantic
17:29:18 <Discord|> <06a​dvil> but that's up from 96% last time I worked on it a few months ago, so it's getting not entirely ignorable
17:43:33 <floraline> i'm assuming we are all done with kobold-nightstalker experimental?
17:44:23 <Discord|> <06a​dvil> yeah probably
17:44:38 <Discord|> <06a​dvil> ok back to 97% (14G) from a round of binary stripping
17:44:57 <Discord|> <06a​dvil> probably crashlogs from cao will be messed up until a rebuild happens, I just did every crawl binary that was there
17:45:55 <Discord|> <06a​dvil> 14G = just stripping binaries cleared up 14G
18:25:22 <Lantell> Unstable branch on underhound.eu updated to: 0.26-a0-598-g61e6210cf4 (34)
18:36:46 <alexjurkiewicz> wow
18:36:57 <alexjurkiewicz> consider upxing old binaries if there are that many
18:37:30 <alexjurkiewicz> you get about 20% compression, not as much as the order of magnitude from stripping
18:37:38 <alexjurkiewicz> that is, 1/5th size
18:46:49 <advil> that sounds like a fancy thing that won't work on cao
18:47:23 <advil> possibly at some point I just need to figure out how to force-update saves in dgl
18:48:10 <advil> like, we don't *actually* need to be keeping 241 different trunk versions around
18:49:14 <floraline> there are scripts to do that in dgl-config
18:49:25 <floraline> i have a cron job run it hourly
18:49:45 <advil> how do you decide what versions to force update?
18:50:34 <floraline> it keeps the 40 newest versions
18:50:42 <floraline> anyone else gets their save transferred and that version deleted
18:50:55 <advil> huh, that's actually pretty aggressive given cao's rebuild schedule
18:50:57 <advil> er
18:50:59 <advil> cko's
18:51:15 <advil> you could actually go through 40 versions in a couple of days
18:51:20 <advil> I guess that's why you need to do it though
18:52:25 <floraline> yeah i guess it is. i'm assuming i was okay with that at the time i did it
18:53:05 <advil> well, it's not unreasonable if that's a known policy, but some players don't necessarily want to upgrade at all once they start playing, if they are playing actively
18:53:26 <advil> for games that haven't been touched since Jan 2019 I will have fewer qualms
18:53:31 <floraline> yeah that's a good point
18:53:38 <floraline> i'm pretty sure 40 versions goes at least a few months
18:53:52 <floraline> keep in mind it's not the last 40 builds but only the ones where players have actually started playing
18:53:57 <advil> ah that's probably plenty for someone playing normally
18:53:58 <advil> yeah
18:54:12 <advil> makes sense, I guess you are running clean-trunks on a cron job also?
18:58:18 <alexjurkiewicz> advil: upx will definitely work. It's been around for decades
18:59:16 <alexjurkiewicz> `upx path/to/binary` will repack in-place. Don't need any other flags, except maybe `-3` or similar to be quicker
19:00:41 <floraline> i've uploaded my cron scripts here https://github.com/floraline/dgl-cron-scripts
19:02:34 <floraline> i use remove-trunks yes
19:05:40 <floraline> it looks like my most recent 40 builds only goes back to august 12 ... not that long actually
19:05:44 <floraline> maybe i should change that
19:06:45 <floraline> alexjurkiewicz: does it make sense to use this upx in conjunction with strip?
19:07:00 <floraline> or do they do the same thing?
19:07:23 <Henzell> nmf (L16 DsVM) Crash caused by signal #15: Terminated (Elf:2)
19:10:52 <advil> !crashlog
19:10:54 <Sequell> 22556. nmf, XL16 DsVM, T:38187 (milestone): http://crawl.akrasiac.org/rawdata/nmf/crash-nmf-20200827-230721.txt
19:11:28 <advil> wonder why that's term and not abort
19:14:28 <advil> alexjurkiewicz: `-bash: upx: command not found`
19:14:45 <floraline> ok i've changed cko policy to keep the 80 most recent trunk games, which should be at least a month
19:14:53 <floraline> i will also not strip symbols from the most recent 20 trunk builds instead of 10
19:15:05 <advil> sorry, didn't mean to make you worry, if no one's complained it might be totally fine
19:15:11 <floraline> nah it was a good point
19:15:24 <floraline> i don't know if i thought about it enough before, two weeks before forcing players to upgrade really sounds too soon
19:15:42 <advil> was it ~ two weeks in practice?
19:15:54 <floraline> not sure about historically, but right now 40 versions is two weeks ago
19:16:05 <floraline> a lot less than i thought
19:17:16 <advil> makes sense, on cao 40 is Jul 18
19:17:31 <advil> with daily upgrades (plus rebuilds)
19:18:47 <advil> 241 goes back to Jan 1, 2019 :D
19:19:24 <floraline> lol wow, i think it's safe to say that's long enough
19:19:28 <floraline> you must have an awful lot of disk space over there
19:20:01 <floraline> is cao on bare metal?
19:28:29 <advil> I am really not sure what the story is with cao but it's a 1TB disk
19:38:53 <Naruni> twelwe: when you use a wand do you usually use the 'V' menu for evoke?
19:59:54 <alexjurkiewicz> floraline: i think so. It might make crash logs even less useful though. It might make the speed and ram size of dcss a little larger, but for the old essentially-never-used binaries that seems fine. I wouldn't upx binaries before you are ready to strip them though
20:00:14 <alexjurkiewicz> advil: `sudo apt install upx` -- but I think you knew that? So not sure what the message means
20:00:56 <alexjurkiewicz> I didn't mean to imply upx is part of gnu coreutils and pre-installed on every linux box, sorry if that's what you understood
21:02:27 <floraline> are morgue text files ever updated once created?
21:16:29 <alexjurkiewicz> no
21:17:12 <alexjurkiewicz> some of the other files in that directory are though. Like the .lst and .map files are updated every time the game is saved
22:00:32 <twelwe> naruni yes, but mostly through a macro. why do you ask?
22:21:14 <floraline> alexjurkiewicz: only the files that are username.lst though right? the ones beginning with morgue- and the timestamp in the filename are a one time thing, yes?
22:25:03 <floraline> i am planning on having cko compress all morgue files with gzip now; does anyone see any problems with this plan?
22:28:44 <floraline> looking for feedback before i kick off this script to process everyone's very important files
22:30:14 <Discord|> <12e​bering> Will your webserver still serve them in a way that they appear plaintext when I click a morgue link
22:30:58 <Discord|> <12e​bering> All the old tournament and csdc score links will break otherwise, which would be a bit of a pain
22:31:27 <Discord|> <12e​bering> Oh, and sequel will break
22:31:52 <Discord|> <12e​bering> (If the gzipping isn’t link transparent)
22:35:53 <floraline> if you click in a modern browser, yes you should see the contents
22:36:01 <floraline> i have no idea if bots will see it correctly
22:38:15 <floraline> they will be accessible both by their current url and also with .gz extension
22:38:45 <floraline> either way you'll get a gzipped file with headers for browsers to see correctly
22:50:47 -!- amalloy_ is now known as amalloy
23:02:36 <advil> alexjurkiewicz I'm impressed you think it is possible to install things on cao
23:02:48 <Naruni> twelwe: because I found something in source where there is a prompt of "Zap which item?" which I don't ever recall seeing in game.  I've always used 'V' to evoke a wand which asks something akin to "Evoke which item?"
23:03:25 <advil> floraline I set up precompression on cao using gz, that might be the way to go if you can do it
23:04:05 <advil> using mod_deflate in apache IIRC
23:04:25 <Naruni> twelwe: if there is to be a feature of seeing noise associated with evocables there will be a lot of things to do.  I was just trying to do some homework
23:04:32 <advil> so it will serve .txt.gz files as .txt
23:04:41 <advil> I have to say that it didn't really lead to the gains I hoped for
23:04:46 <advil> though I didn't really do it in bulk
23:05:19 <advil> there were some noticeable savings for a few really high-use accounts, but for everything else it was the wrong order of magnitude at least relative to the cao disk
23:05:25 <twelwe> i wasn't envisioning a noise prompt through V, but through inspection - like inspecting a spell
23:05:26 <advil> maybe it'll help more on a smaller disk?
23:06:12 <advil> or rather, be proportionately more noticeable
23:11:14 <Naruni> twelwe: putting it in inspection I might be able to pull that off quicker than an entire new menu of evoke
23:12:29 <twelwe> you can do it, this thing
23:20:18 <Naruni> !source describe_item_popup
23:20:18 <Sequell> Can't find describe_item_popup.
23:25:43 <floraline> advil: my files get moved off disk to s3 storage, so i can't rely on nginx alone
23:26:17 <floraline> i have a cgi script that lists the files from s3 and adds the necessary gzip headers
23:26:48 <floraline> from what little i tested it looks like gzip is compressing to smaller than 50% compared to what i have now which is zero compression
23:28:13 <floraline> right now i have the txt morgue files on disk, taking up about 10GB of my 60GB disk
23:29:06 <floraline> if i can compress and move them offsite that should save a lot of space and stop the problem from growing
23:30:00 <floraline> after this, the only looming threat is the log files which grow forever and can't be moved
23:30:16 <Naruni> ok I think I found it.  In describe.cc, around line 2189 I can add to the descriptor of wands.  Might need a case statement for which spell the wand casts but hopefully I can just plugin spell_noise_string(spell)
23:42:54 <alexjurkiewicz> advil: rip
23:45:06 <advil> floraline: well the log files can be moved easily once they are no longer being added to. It also actually shouldn't be a problem to break logfiles/milestones into chunks either as far as I can tell, as long as it's done cleanly
23:45:52 <advil> for whatever reason not too many server owners have done it intentionally, but it's happened as a repair mechanism in some cases and it works fine
23:46:05 <advil> cjr data for example
23:46:19 <floraline> i'm really just thinking of trunk which never really changes
23:46:27 <floraline> that would be nice to split it off and move offsite
23:46:46 <advil> well, I guess it would take some thinking about how exactly to do it without rereading a lot of data
23:47:00 <advil> since in the cjr case it definitely did need rereading
23:47:27 <floraline> hmm yeah
23:50:41 <floraline> does cao store all ttyrecs and morgues on disk? i don't understand how it's not out of space yet
23:51:15 <advil> remember, 1TB
23:51:22 <floraline> i'm seeing a rate of maybe 75gb/yr, i'm already over halfway full with a little over 300GB total space
23:51:45 <floraline> and cao has at least twice as many players
23:52:07 <advil> there's also a separate disk for the chroot, everything else is on a different 200G disk
23:53:01 <floraline> how long has that system been in service? it's gotta be getting close to the limit
23:53:16 <advil> 1000 / 75 is 13.3 which is not too far off
23:53:29 <advil> earliest file dates I see are 2009
23:54:11 <floraline> it's gotta be on bare metal with that kind of storage
23:54:17 <advil> probably
23:55:06 <floraline> well anyway. cko will keep chugging along for the time being, i have enough space for about 2 more years before i have to start making decisions
23:55:25 <advil> I do kinda wonder what the lifespan on this cao 1TB disk is going to be
23:55:26 <floraline> next on my plate is buster upgrade and figure out what to do about tornado
23:55:35 <floraline> ah yeah that's a good point
23:55:43 <Naruni> I need a bit of reference / pointer help here.  the function I'm working with is called with const item_def &item.  in that function, there is another function called with _append_weapon_stats(item).  _append_weapon_stats requires const item_def &item.  the function I want to call requires (wand_type wand).
23:56:06 <Naruni> can I assume item_def is != wand_type
23:56:13 <advil> definitely
23:56:47 <advil> item_def is a pretty heavy struct, wand_type is probably an enum
23:56:47 <Naruni> thank you
23:57:58 <advil> yes, wand_type is an enum, if you know you have a wand (which seems a bit unlikely relative to the other function names tbh) the item's sub_type is an int which should work if cast to wand_type
23:58:19 <advil> well, it's a uint8_t
23:59:18 <Naruni> so I tried calling with (wand_type item->wand_type) and (wand_type item.wand_type) but errors.
23:59:34 <advil> that is, if `item.base_type == OBJ_WANDS`, then `item.sub_type` will be interpretable as a wand_type