01:14:58 -!- raskol_ is now known as raskol
02:10:16 <Gretell> Windows builds of master branch on crawl.develz.org updated to: 0.25-a0-766-g0ecfff5982
02:45:05 <Gretell> Monster database of master branch on crawl.develz.org updated to: 0.25-a0-651-g8c222c7a2b
03:24:13 <Rotatell> Unstable branch on crawl.beRotato.org updated to: 0.25-a0-766-g0ecfff5 (34)
03:53:50 <Kelbell> Fork (bcadrencrawl) on crawl.kelbi.org updated to: 0.22.1-2576-g2bcd4b577a
05:15:00 <Bcadren> Is it just oversight that timeout_terrain_changes and timeout_malign_gateways aren't called when returning to a floor after time away or is there a reason they shouldn't be timed out while the player is away?
08:57:19 <Cheibriados> 03ebering02 07* 0.25-a0-767-ga8ac3e9: Fix an arena crash 10(2 hours ago, 1 file, 3+ 0-) 13https://github.com/crawl/crawl/commit/a8ac3e9bb09e
08:57:19 <Cheibriados> 03ebering02 07* 0.25-a0-768-g6f2fc6e: Nameless horror tile (CanOfWorms) 10(34 seconds ago, 3 files, 2+ 1-) 13https://github.com/crawl/crawl/commit/6f2fc6eb8cf2
09:15:23 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.25-a0-768-g6f2fc6eb8c (34)
09:18:57 <Cheibriados> Arena in windows offline tiles doesn't work 13https://crawl.develz.org/mantis/view.php?id=12238 by Goratrix
10:23:25 <Cheibriados> 03Aidan Holm02 07* 0.25-a0-769-g34829ab: Fix another arena crash (#11894, #12238) 10(2 minutes ago, 1 file, 6+ 0-) 13https://github.com/crawl/crawl/commit/34829ab59be5
10:32:44 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.25-a0-769-g34829ab59b (34)
11:06:16 <travis-ci> The build is still failing. (master - 34829ab #13020 : Aidan Holm): https://travis-ci.org/crawl/crawl/builds/674433989
11:11:15 <aidanh> hmm, not sure what's happening there
11:15:23 <advil> looks like it's on arena `miscasts 5 pandemonium lord v 20 20-headed hydra delay:0 t:10`, it's probably replicable locally
11:16:32 <aidanh> oh, it is
11:18:05 <gammafunk> pan lords at it again
11:18:05 <advil> I'm not immediately sure why the alive check at the beginning of miscast_effect didn't catch this though
11:51:40 <Cheibriados> 03Aidan Holm02 07* 0.25-a0-770-g53dc336: Fix bugged acquirement menu (#12234) 10(2 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/53dc3366b14c
12:02:51 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.25-a0-770-g53dc3366b1 (34)
13:31:12 <Henzell> Unstable branch on crawl.akrasiac.org updated to: 0.25-a0-770-g53dc336 (34)
16:11:14 <Cheibriados> 03kate-02 07* 0.25-a0-771-ga3f3f81: Use BASELINE_DELAY for some enchantment durations 10(3 hours ago, 2 files, 3+ 2-) 13https://github.com/crawl/crawl/commit/a3f3f815865d
16:11:14 <Cheibriados> 03kate-02 07* 0.25-a0-772-gf7929f5: Give the plutonium sword a chance to polymorph monsters again 10(3 hours ago, 1 file, 8+ 3-) 13https://github.com/crawl/crawl/commit/f7929f599a5c
16:11:14 <Cheibriados> 03kate-02 07* 0.25-a0-773-g43196ba: Don't print auto item slot adjustments on turn 0 10(4 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/43196ba69759
16:18:50 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.25-a0-773-g43196ba697 (34)
16:26:37 <Cheibriados> 03kate-02 07* 0.25-a0-774-g40ae2d4: Add details to the "Beogh" status light description 10(2 minutes ago, 1 file, 1+ 0-) 13https://github.com/crawl/crawl/commit/40ae2d4f9023
16:33:04 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.25-a0-774-g40ae2d4f90 (34)
16:54:08 <travis-ci> The build is still failing. (master - 43196ba #13022 : Kate): https://travis-ci.org/crawl/crawl/builds/674573363
16:57:59 -!- Amnesiac_ is now known as Amnesiac
17:17:08 -!- jfcaron_ is now known as jfcaron
17:20:17 <travis-ci> The build is still failing. (master - 40ae2d4 #13023 : Kate): https://travis-ci.org/crawl/crawl/builds/674578563
17:54:09 <advil> well, this crash might be a deeper bug: mons_adjust_flavoured with BEAM_NEG calls drain_exp on the target, which applies random negative damage and can kill the monster (with cleanup). But mons_adjust_flavoured also returns a damage value for resist-adjusted damage, which is then used by most callers to apply some damage?
17:54:38 <advil> so the immediate bug is that it's wrong to assume that mons_adjust_flavoured can't kill the monster (I think this bug was latent, but the relevant beam types weren't used for miscasts before)
17:55:00 <advil> but it seems like in this particular case like mons_adjust_flavoured with BEAM_NEG does extra damage against monsters
17:57:49 <advil> so for exemple in bolt::affect_monster, there's also a mons_adjust_flavoured call followed by a hurt call (in this case correctly handling the dead monster case), which should do double duty for negative damage if I'm understanding this code right
18:12:30 <gammafunk> yeah, re mons_adjust_flavored, I've learned from e.g. god conducts that those need to be applied before calling that function
18:12:45 <gammafunk> drain xp was the culprit in those crashes as well
18:19:22 <Cheibriados> 03gammafunk02 07* 0.25-a0-775-gd9a95f1: Remove a no longer relevant Donald line 10(25 seconds ago, 1 file, 0+ 2-) 13https://github.com/crawl/crawl/commit/d9a95f1d30d0
18:22:52 <Lantell> Unstable branch on underhound.eu updated to: 0.25-a0-774-g40ae2d4f90 (34)
18:24:57 <Cheibriados> 03riverfiend02 {GitHub} 07https://github.com/crawl/crawl/pull/1352 * 0.25-a0-750-g7afe695: Improved map, replaced unnecessary tags 10(4 minutes ago, 1 file, 26+ 20-) 13https://github.com/crawl/crawl/commit/7afe69516451
18:32:50 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.25-a0-775-gd9a95f1d30 (34)
18:40:05 <advil> doesn't it seem incorrect that mons_adjust_flavoured should return a non-zero damage value if drain_exp has been called though?
18:42:06 <advil> or at least weird
18:43:57 <gammafunk> you mean specifically when it's killed the monster?
18:44:02 <advil> in either case
18:44:19 <gammafunk> oh, since no actual damage was done in addition?
18:44:31 <gammafunk> iirc the value returned is the adjusted damage
18:44:41 <advil> mons_adjust_flavoured does damage by way of calling drain_exp, and then the value returned is usually used for a hurt call
18:45:03 <advil> the value returned in that case is rolled independently of drain_exp, I think
18:46:12 <advil> `hurted = resist_adjust_damage(mons, pbolt.flavour, hurted);`, and then separately, `mons->drain_exp(pbolt.agent());`
18:46:40 <Cheibriados> 03riverfiend02 {GitHub} 07https://github.com/crawl/crawl/pull/1352 * 0.25-a0-751-g14eb6dd: removed spaces 10(4 minutes ago, 1 file, 16+ 16-) 13https://github.com/crawl/crawl/commit/14eb6ddf9f30
18:49:02 <gammafunk> yeah, the damage from drain_exp is just a small randomized damage value
18:49:34 <gammafunk> 2 + random2(3)
18:49:40 <advil> oh, I see
18:49:46 <gammafunk> this is meant to be applied in addition to any main damage
18:49:50 <advil> so it's basically a token value so some damage is done
18:50:06 <gammafunk> I suppose so, yeah
18:50:21 <gammafunk> or sort of like how acid damage does two damage components
18:50:25 <gammafunk> the main acid damage and the splash
18:50:49 <advil> yeah, I noticed that for acid damage too
18:51:43 <gammafunk> whatever the case, this way of managing monster damage is very error prone
18:51:52 <Cheibriados> 03advil02 07* 0.25-a0-776-gbf642f8: Fix a miscast crash 10(4 minutes ago, 1 file, 10+ 4-) 13https://github.com/crawl/crawl/commit/bf642f89ba69
18:52:17 <gammafunk> I tried to centralize monster damage from play in spl-damage.cc with _player_hurt_monster() for this reason
18:52:27 <gammafunk> but I wasn't able to move all logic fully into that
18:52:39 <gammafunk> partly because of the mons_adjust_flavored stuff
18:52:53 <gammafunk> s/from play/from player/
18:53:29 <gammafunk> but then there are just all these other cases of spells doing damage and having to manage that carefully outside of that file
18:56:15 <advil> for this case I suppose one simple way to reduce issues could be for mons_adjust_flavoured (and by extension stuff like drain_exp) to commit to never killing the monster
18:56:36 <gammafunk> possible, yeah
18:56:52 <gammafunk> just have to look at all the calls to it and refactor them (there are a lot)
18:56:56 <advil> I'm still not sure I understand why drain_exp needs to do 2+rnadom2(3) damage in the first place, I wonder if it was originally to trigger various other code
18:57:03 <advil> like annoy the monster
18:57:11 <gammafunk> heh, maybe yeah
18:57:26 <Cheibriados> 03riverfiend02 {GitHub} 07https://github.com/crawl/crawl/pull/1352 * 0.25-a0-752-g1c6c22b: Update overflow.des 10(5 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/1c6c22b9b1b7
18:57:37 <gammafunk> trying to think if anything can apply beam_neg without damage
18:57:51 <gammafunk> yred's drain life specifically uses beam_neg *without* the secondary effects
18:58:05 <gammafunk> but I can't think of anything that does drain only without the usual extra damage component
18:58:12 <gammafunk> hrm, phantom mirror?
18:58:16 <gammafunk> probably not
18:58:51 <advil> once upon a time that was hurt(agent, 2 + random2(pow), ...)
18:59:03 <advil> %git e250b60d2c
18:59:03 <Cheibriados> 07DracoOmega02 * 0.13-a0-2280-ge250b60: Revamp player xp draining 10(7 years ago, 13 files, 60+ 78-) 13https://github.com/crawl/crawl/commit/e250b60d2c6a
18:59:19 <advil> change to random2(3) there, but there's not really an explanation
18:59:43 <advil> "For the time-being, draining still works the same against monsters as it always has."
19:03:03 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.25-a0-776-gbf642f89ba (34)
19:03:42 <gammafunk> he lied to us....
19:04:02 <travis-ci> The build is still failing. (master - d9a95f1 #13024 : gammafunk): https://travis-ci.org/crawl/crawl/builds/674621231
19:05:27 <advil> well, it looks like before that pow had a default value 3 and that param wasn't used much
19:06:24 <advil> and in that commit he started using it for player draining
19:08:18 <Cheibriados> 03riverfiend02 {GitHub} 07https://github.com/crawl/crawl/pull/1352 * 0.25-a0-753-gd1a0de2: Update overflow.des 10(49 seconds ago, 1 file, 14+ 15-) 13https://github.com/crawl/crawl/commit/d1a0de22bb34
19:14:10 <gammafunk> wow, just noticed we're doing macOS builds in travis now
19:36:07 <alexjurkiewicz> and github actions!
19:42:25 <gammafunk> seems that macoS doesn't support the nproc command
19:42:55 <gammafunk> first line of the make -j$(nproc) recipe is it failing to find nproc
19:54:23 <travis-ci> The build was fixed. (master - bf642f8 #13027 : advil): https://travis-ci.org/crawl/crawl/builds/674630425
20:05:32 <alexjurkiewicz> sysctl -n hw.ncpu
20:08:14 <advil> mac doesn't provide gnu stuff; I have nproc via the macports coreutils package, and there's probably a similar homebrew package
20:09:05 <advil> the thing that alexjurkiewicz just pasted is how you answer the same question with builtin mac tools
20:12:44 <gammafunk> elliptic: watched the ttyrec of your CeTm, hope you enjoyed that worm tunnel vault in Swamp
20:13:29 <Cheibriados> New branch created: pull/1353 (1 commit) 13https://github.com/crawl/crawl/pull/1353
20:13:29 <Cheibriados> 03Alex Jurkiewicz02 07https://github.com/crawl/crawl/pull/1353 * 0.25-a0-748-g0884ee9: Fix CPU detection on macOS 10(6 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/0884ee954e83
20:13:40 <gammafunk> advil: yeah, I'll leave it to the travis experts to update the macOS recipe for that, if that's a thing we can do
20:13:40 <gammafunk> just happened to notice it when reading the build log
20:16:05 <gammafunk> another thing I found somewhat confusing was how the one action is labeled "Installing submodules (for crosscompilation)" yet it doesn't seem to actually be crosscompilation?
20:17:01 <gammafunk> maybe it is, but I don't see where it's specifying a crosscompiler (usual way is with a make arg)
20:17:44 <gammafunk> yeah, not sure what "runs-on: macos-latest" actually does
20:18:34 <advil> I bet that's just a copy & paste
20:18:40 <gammafunk> probably, yeah
20:19:21 <alexjurkiewicz> runs-on specifies the base os
20:19:27 <alexjurkiewicz> it's like `os: linux` in .travis.yml
20:19:40 <alexjurkiewicz> here's the reference for workflow files https://help.github.com/en/actions/reference/workflow-syntax-for-github-actions
20:20:42 <alexjurkiewicz> and yeah, that (for crosscompile) is a copy paste error
20:23:55 <alexjurkiewicz> and for the record, github actions and travis are unrelated. Unless you're trying to say "all CI looks the same to me"
20:24:20 <Cheibriados> 03Alex Jurkiewicz02 07https://github.com/crawl/crawl/pull/1353 * 0.25-a0-749-g6f30059: Correct a step name 10(2 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/6f300594afa3
20:25:34 <gammafunk> yeah I'm familiar with the general concept of github actions; we use it on the roguelikes Discord to print commits in channel as they are made
20:26:37 <gammafunk> actually I don't think I commented about github actions vs travis
20:27:31 <gammafunk> oh the "travis experts" thing; I assume that this is still considered "travis"
20:27:37 <gammafunk> as its the one reporting the build results etc
20:30:42 <alexjurkiewicz> ah -- the github actions results aren't reported by the bot in here at all
20:31:51 <gammafunk> right, I see now
20:31:58 <gammafunk> how did I actually find that macOS thing...
20:32:41 <gammafunk> ahh, the email
20:32:56 <alexjurkiewicz> if you look at the status checks of any build (click the green tick / yellow circle / red cross) in github UI, there are five million github actions checks and one travis check. One of the github ones will be the macos build
20:33:03 <alexjurkiewicz> or that!
20:33:21 <gammafunk> yeah makes more sense now, thanks
20:33:50 <alexjurkiewicz> FWIW, github actions has no integrated way to report to IRC. I started writing some code to add support, if you wanted to deprecate travis and migrate to github actions, I can finish it off and submit as a PR
20:34:22 <alexjurkiewicz> I believe the github actions checks are now a superset of what travis does, and lack of IRC reporting is the only missing feature
20:34:50 <alexjurkiewicz> but officially deprecating might be more a question of when youse are all comfortable with the github actions UX
20:39:02 <gammafunk> looks like we're doing crosscompile twice, once in travis and once in github actions
20:39:51 <gammafunk> ah I see
20:39:55 <gammafunk> one is zips, one is installer
20:42:31 <alexjurkiewicz> i figured they were equivalent. if you want, I can configure github actions to run both targets
20:44:38 <advil> in travis iirc it generates both zips and installer
20:44:45 <advil> in one target
20:45:12 <advil> eventually I do want to get all the relevant actions to produce products, so I will mess with that whenever I get to it
20:45:56 <Cheibriados> 03Alex Jurkiewicz02 {advil} 07* 0.25-a0-777-g713ffed: Fix CPU detection on macOS 10(39 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/713ffed48a15
20:45:56 <Cheibriados> 03Alex Jurkiewicz02 {advil} 07* 0.25-a0-778-g357ffb7: Correct a step name 10(23 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/357ffb74ea0d
20:51:11 <alexjurkiewicz> advil: produce those artifacts for every build, or just releases?
20:51:23 <alexjurkiewicz> I'll implement the latter for you now, if you'd like
20:52:12 <advil> probably every build? the windows one (if it's like travis, I haven't looked closely) may already be producing at least one of the relevant targets
20:52:12 <advil> so it's just a matter of exporting the product
20:52:19 <advil> sure...I think you may need to do something about the version issue we discussed though
20:52:31 <advil> so that the files can be named correctly
20:53:07 <alexjurkiewicz> yes. For every commit, you can upload images to artifact storage. Which is unlimited and free, but expires after 90 days
20:53:29 <alexjurkiewicz> for releases, you can upload builds as release attachments, which is also unlimited and free but never expires
20:53:39 <advil> sounds good
20:54:10 <advil> I looked into doing this in travis once upon a time but never got round to finalizing it, and travis doesn't have its own storage...but it looked really straightforward to set this up with actions
20:54:27 <advil> the windows targets in particular would be helpful in terms of moving things off of CDO
20:54:33 <alexjurkiewicz> is the goal here is to provide trunk builds for players to use?
20:54:52 <advil> yeah, that was my immediate goal...though if it automates release packaging too, that's gravy!
20:54:59 <advil> also trunk builds for mac
20:55:08 <alexjurkiewicz> i suppose it's good for branches to, because then you can easily share experimental builds
20:57:27 <Cheibriados> 03riverfiend02 {GitHub} 07https://github.com/crawl/crawl/pull/1352 * 0.25-a0-754-gf9d65e7: fixed spacing 10(5 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/f9d65e712f13
20:58:21 <gammafunk> github hosts the tournament
20:58:31 <gammafunk> scores calculated via github actions
20:59:20 <advil> maybe github actions can join the devteam
20:59:37 <alexjurkiewicz> you could do that... it has a "run action on cron schedule" task trigger
21:03:03 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.25-a0-778-g357ffb74ea (34)
21:06:54 <Cheibriados> 03Alex Jurkiewicz02 {advil} 07* 0.25-a0-779-g398c232: Example webtiles systemd unit file 10(26 hours ago, 1 file, 15+ 0-) 13https://github.com/crawl/crawl/commit/398c232eb95e
21:07:16 <alexjurkiewicz> i wonder if it's worth making linux trunk builds :D
21:07:44 <gammafunk> cdo does already do that
21:07:48 <alexjurkiewicz> it's nearly free so might as well. But would have to set up a makefile target to zip them up. So maybe not for now, until someone complains
21:07:48 <gammafunk> makes debs, of course
21:08:14 <gammafunk> you probably would make debs, not zips; this is linux, after all
21:08:36 <gammafunk> cdo just makes debs and puts them in our repo
21:09:43 <advil> maybe the `make package-source` result (which is needed for release anyways)
21:10:04 <alexjurkiewicz> wonder if that person asking about flatpack would contribute a makefile target for it...
21:12:13 <gammafunk> we still need debs for release (and currently rely on someone else to keep an outside yum repository of rpms)
21:12:39 <gammafunk> I'm sure some people are using github to make debs if they're doing all this other stuff
21:12:54 <gammafunk> but we'd probably need to have a fixed repository location(s)
21:14:19 <gammafunk> like we do now on CDO, I mean
21:16:59 <advil> huh, on a mac crawl ignores SIGQUIT
21:17:13 <advil> is that true on linux too?
21:17:35 <gammafunk> let's see
21:17:39 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.25-a0-779-g398c232eb9 (34)
21:18:27 <gammafunk> advil: does for me
21:18:34 <gammafunk> ignores it, that is
21:18:50 <advil> well, I guess that is one answer to my question to alexjurkiewicz of "why not SIGQUIT"
21:19:14 <advil> gammafunk while you're at it can you confirm that SIGABRT behaves like TERM?
21:19:18 <advil> on linux
21:19:43 <gammafunk> yeah, I get a crash in the usual SIGTERM way
21:19:43 <advil> it does on mac
21:20:04 <advil> ok...makes sense in that we do regularly see ABORT crashes in this channel (since libc uses that signal)
21:20:20 <gammafunk> right, it says "Aborted" instead of "Terminated"
21:21:26 <alexjurkiewicz> fwiw, I think ABRT fits better than QUIT anyway, because "abort program" sounds more abnormal than "quit program" (as per the signal meaning from signal(3))
21:22:05 <advil> well, I was originally inclined the other way because SIGQUIT is conventionally supposed to mean "crash and dump core" whereas SIGABRT is more "weirds glibc error to do with malloc"
21:22:11 <alexjurkiewicz> however, if you really prefer QUIT... once you can add it to the crash handler in crash.cc it should do what you want
21:22:28 <advil> *weird
21:22:41 <alexjurkiewicz> I think it would be good to update dgamelaunch-config to use QUIT rathe rhtan ABRT though, which then means needing to update both every webtiles server and every dgamelaunch server
21:23:30 <alexjurkiewicz> (imagine whoever hacked this together ten years ago discovering the path dependency leading to where we are today)
21:25:14 <advil> re the dgl-config precedent, I'm a bit confused as to why it's using abort and need to double check what webtiles server even does (if anything) on that signal
21:25:35 <advil> well, the script works to shut down webtiles
21:25:40 <advil> it might be crashing it though
21:29:07 <Cheibriados> 03Alex Jurkiewicz02 {advil} 07* 0.25-a0-780-g622e69b: Send DCSS processes SIGABRT rather than SIGTERM 10(26 hours ago, 1 file, 4+ 4-) 13https://github.com/crawl/crawl/commit/622e69b31f54
21:29:57 <alexjurkiewicz> webtiles server doesn't have special handling for abort. It looks like if you ABRT a python interpreter it crashes
21:30:07 <advil> yeah, that's what I was figuring
21:30:13 <advil> not sure what happens to the child processes
21:30:30 <advil> might vary somehow
21:30:45 <advil> I *think* that script should be sending TERM
21:30:52 <alexjurkiewicz> why are you interested in this? Does dgl-config send ABRT to webtiles server?
21:31:02 <alexjurkiewicz> I thought that script was purely killing crawl processes
21:31:11 <advil> that script you linked there sends ABRT to the webtiles server, not to crawl processes
21:32:05 <advil> it's an /etc/init.d script that is used to start and stop the server
21:32:42 <alexjurkiewicz> oh! heh
21:32:42 <Kelbell> Unstable branch on crawl.kelbi.org updated to: 0.25-a0-780-g622e69b31f (34)
21:33:07 <advil> I'm interested partly because I've been working on dgl-config, and noticed this during that
21:33:20 <advil> so effectively that script is an init.d counterpart to the systemd contrib thing you did
21:34:51 <alexjurkiewicz> hm, it's weird that the mac Makefile build process has one target to build tiles and one to build console, while windows has a single target to build both
21:35:52 <gammafunk> don't think it does have a target to do that?
21:35:54 <gammafunk> which target?
21:36:03 <gammafunk> unless you mean package-windows
21:36:12 <alexjurkiewicz> that one, and package-windows-zips
21:36:18 <gammafunk> what package-windows makes is the installer, so yeah it has to
21:36:31 <alexjurkiewicz> oh, there's one installer with both games inside?
21:36:32 <gammafunk> it's not just doing those builds, it's packaging them into an EXE installer
21:36:34 <gammafunk> right
21:36:42 <gammafunk> for zips they are standalone
21:36:47 <gammafunk> but also yeah building both, I guess
21:36:58 <advil> I'm not sure that's really necessary but the mac build process for actually doing releases historically has been somewhat finicky, and there are some parts that exist only in notes that I inherited that involve fairly carefully cleaning between those two targets
21:36:58 <gammafunk> this is just because we use this target for the release
21:37:16 <advil> it might have mattered more when someone was signing the releases
21:37:28 <gammafunk> yeah, I remember when geekosaur was going through all that
21:37:32 <gammafunk> was a true nightmare apparently
21:43:04 <alexjurkiewicz> yeah. i know with the mac targets you need to explicitly make clean between them
21:46:51 <gammafunk> for all I know the windows targets do that between builds as well
21:48:01 <gammafunk> seems not between builds, but they do have an extra make clean as an initial step
22:28:25 <alexjurkiewicz> advil: github artifact storage is limited to 500mb, once you reach that limit further artifact uploads will fail. I was thinking about three approaches here: 1) don't create trunk builds with GA yet 2) Create them and upload to external storage (eg S3) 3) Create them and have a manual script to trim all but the most recent 20 builds (or so)
22:28:28 <alexjurkiewicz> thoughts?
22:37:19 <elliptic> gammafunk: the first time I saw that vault it didn't have any monsters, it made a lot more sense with swamp worms...
22:41:58 <gammafunk> elliptic: oh no, I was outdone by spriggan druids and their convoking, I guess
22:47:52 <gammafunk> I wonder if we could just make those convoking things just not pull stuff from vaults in general
22:48:27 <|amethyst> gammafunk:  Would V:5 convokers even work then?
22:48:37 <|amethyst> since isn't everything a vault?
22:48:55 <gammafunk> |amethyst: presumably we'd have to distinguish vault from layout properly, and yeah that might not be trivial in all cases
22:49:22 <gammafunk> probably a sign more that convoking could use a bit of reworking as a concept, but I'm not sure how to fix it
22:50:19 <gammafunk> I wonder if more of a doom hound approach would be better
22:50:53 <|amethyst> hm, could the patrolling flag be used here?
22:50:53 <|amethyst> i.e. don't convoke things that are patrolling
22:50:55 <gammafunk> well, I suppose in the case of vaults that wouldn't influence AI in an adverse way, yeah
22:51:11 <gammafunk> since if you want them to not get sucked out of a vault, they're probably fine to patrol
22:51:29 <gammafunk> actually, now that I think about it
22:51:45 <gammafunk> it's not going to work so hot since patrolling will keep non-sealed monsters in the vault
22:52:04 <gammafunk> I guess this *could* be a solution for runed door/transporter vaults specifically
22:52:15 <gammafunk> we'd not want to put it on open vaults though
22:52:55 <|amethyst> well, if you want monsters to be able to leave the vault to come after the player, I would think having them convoked away from the vault to the player also wouldn't be that bad
22:52:56 <gammafunk> however that "doom hound" idea might also work; could limit convokers to one or two usages of their ability
22:53:10 <|amethyst> hm
22:53:14 <gammafunk> no, it wouldn't be bad for convoking, but it would be bad for non-convoking situations
22:53:27 <gammafunk> if we applied that flag to vaults liberally, I mean
22:53:41 <gammafunk> for sealed vaults it would indeed be ok
22:53:44 <|amethyst> ah, yeah, that's right
23:03:49 <alexjurkiewicz> similar to doom hound idea, convoking could summon a duplicate of a monster on the level
23:12:36 <gammafunk> right, another related issue with a fix like that would be zot trap effect, which convokes a few monsters; I was thinking durably summoned shadow creatures for that
23:12:51 <gammafunk> summoning a duplicate of a monster is kind of cute, but I wonder if there would be weird incentives for that somehow
23:13:42 <gammafunk> the epic mara convoke summon and then the real mara comes into view
23:22:17 <alexjurkiewicz> yeah, I guess you'd want to make the monsters durable summons to prevent some sort of hypothetical optimalness