00:30:31 Unstable branch on CRAWL.XTAHUA.COM updated to: 0.22-a0-351-g29664b4 (34) 02:40:06 -!- aditya is now known as adibis 02:40:19 -!- adibis is now known as aditya 02:53:52 -!- aditya is now known as adibis 03:15:25 -!- amalloy is now known as amalloy_ 03:22:50 Unstable branch on crawl.beRotato.org updated to: 0.22-a0-351-g29664b4 (34) 07:18:12 Quaffed potion of experience worth 120`000+ XP potions (without actually quaffing it!) 13https://crawl.develz.org/mantis/view.php?id=11446 by Ololoev 08:11:35 !tell advil do you have any idea what opengl availability is like on macs? I'm considering bumping the required version up, probably to 2.0 -- 3.2 08:11:36 aidanh: OK, I'll let advil know. 08:13:46 aredel (L27 NaFi) Crash caused by signal #11: Segmentation fault (Coc:1) 09:31:09 According to the x-v monster description, animals can be recited to. 13https://crawl.develz.org/mantis/view.php?id=11447 by NormalPerson7 09:36:29 Statues do not have "It cannot regenerate" in their description 13https://crawl.develz.org/mantis/view.php?id=11448 by NormalPerson7 09:50:16 aidanh: as long as it's not the "legacy opengl extensions" (I don't know what that is) 3.2 works for most things people could have from 10.7 onwards, see https://developer.apple.com/opengl/OpenGL-Capabilities-Tables.pdf p. 24 09:50:17 advil: You have 2 messages. Use !messages to read them. 09:50:29 it probably does put a bound on older computers 09:51:35 is this comment worth responding to? https://github.com/crawl/crawl/commit/910c0f972a9f9e6bb966cfd934611e5287c5668a#commitcomment-28544031 09:51:48 probably not, but I don't even like accepting the presuppositions of it 09:52:06 I guess alex already gave his thumbs down, maybe that's enough 09:54:46 I'd say that's enough; not worth responding to 09:55:33 thanks for that compat table, that looks great 09:56:41 I'm not sure how important compoat for old versions would be, but it could potentially be possible to keep the current implementation with a compile time switch 09:56:58 s/old versions/old computers/ 09:58:08 I might have an ancient macbook that still works and is old enough to be impacted, it would be from about 2010; it is pre 10.7 and iirc has a graphics card that isn't on that list 09:58:29 the main motivation for this is that there's some weird malloc/free/mutex contention stuff that's blowing the 16ms/frame budget 09:59:03 I wonder if it would magically fix the occasional weird windows white tile problem 09:59:50 I have my own SDL2+opengl 3.2 game engine, and could port the major pieces over 10:01:44 that'd be a bonus; I have a hunch the white tile issue is caused by the player.png being >1024px tall 10:04:06 well, that would be easy to test, but why would that do it? 10:05:42 it'd be easy to test if we could reproduce the issue (can we? I was under the impression we couldn't reliably do so) 10:06:17 it reliably happens in a VM 10:06:26 graphics cards tend to like square power-of-two textures, and some cards will have bugs when you have weird sizes 10:06:50 my VM timed out so I'd have to reconstruct it though 10:06:52 particularly, 1024w x <1024h will probably be rounded up to 1024w x 1024h internally 10:08:25 if it reliably happens in a VM, then that guess is probably completely wrong 10:08:47 well, if you s/card/driver then it could be 10:08:59 in past reports it's often gone away when updating drivers 10:09:26 yeah, I meant drivers 10:09:36 and VMs have specialized graphics drivers 10:09:57 but VMs often have some sort of passthrough to the host graphics system 10:10:19 yes, but there's a graphics driver layer in the vm OS that's supplied by the VM 10:10:30 I don't actually know if it reliably happens in *any* VM, but there are multiple reports of it in virtualbox 10:11:01 https://crawl.develz.org/mantis/view.php?id=11378 10:13:13 well, "multiple" = me and one other person 10:13:24 worth checking then 10:14:22 I'm not sure how you'd reduce player.png; just crop it and ignore the borked rendering for the bits on the lower region? 10:14:47 for testing that seems plausible 10:15:08 well, I suppose if you just cropped it it'd throw all the uv coords off; you'd get distorted smears instead of white 10:15:43 hm 10:16:08 actually, I think by your logic if you crop it in a *working* system, shouldn't it produce the same effect? 10:16:53 it should 10:17:20 it's not really important what effect it produces, as long as it's not a segfault and it's not pure white 10:17:44 also, your hypothesis probably makes predictions about which tiles show up and which go white 10:18:34 I based it off the imgur links in https://crawl.develz.org/mantis/view.php?id=11004 10:19:48 one possible fix would be to blit all the individual textures onto a single 2048x2048 texture, and then use that for everything 10:19:50 hm wall.png is very slightly over, but it's definitely stuff towards the beginning that isn't showing 10:22:32 floor.png is 1024 x 960, but from googling it actually sounds like any texture should ideally be a power of 2 square? 10:22:57 ideally, yeah, for both compatibility and performance reasons 10:23:50 if it is a driver issue, the bad size would probably bork all pixels on the texture, not just the "cropped" ones at the bottom 10:25:37 ah, feat is 1024 x 512, and it does seem that tiles from that are the ones that consistently do show up in these screenshots 10:32:34 it seems like this could be done somewhat seamlessly by modifying rltiles/tool/tile_page.cc 10:32:46 especially if the first try doesn't enforce squares 10:33:25 you mean round up the output textures? 10:33:34 just make sure that m_height is the next power of 2 before writing 10:33:36 yeah 10:33:54 should be easy, and it should add very little extra size to the png files 10:34:59 I can't remember what the support for 2048x2048 textures is like though 10:35:16 well, I was thinking that since 1024x512 seems to work (at least for the VM case) we could first try just ensuring that the height is a power of 2 10:35:42 right now the width in that code is fixed at 1024, so changing that might be a bit of effort (haven't looked closely) 10:36:20 only the height varies, and it just picks the exact height for the relevant # of rows 10:37:11 I think that 1024x2048 would likely still be buggy, but of course you'd need to test to find out for sure 10:38:13 cropping the tall pngs to a height of 1024px only adversely affected rendering of the cropped region, so I'd say that's the quickest way to test 10:38:59 well, it would take me a while to rebuild my VM, the dev images timed out and I need to redownload them; I won't be able to do that before next week 10:39:38 fair enough, no rush really 10:40:18 but this seems like a very solid lead, given that feat.png works 10:41:25 yup; separately 2048x2048 should be fine for all hardware 11:13:28 -!- Guest64792 is now known as rofl 12:43:56 i'm looking at trying to fix bug number 11448 12:44:44 -!- amalloy_ is now known as amalloy 12:44:52 in mon-data.h, where all the monster data is stored, zombies have M_NO_REGEN flag 12:44:58 but statues don't 12:45:02 why is that? 12:45:36 are their forms of having no regen fundamentally different, or something? 12:47:56 my advice is to work backwards from git grep 12:48:22 using M_NO_REGEN, since you know that flag will lead to the codepath where regen happens and it checks for this flag 12:48:33 right 12:48:41 that flag my be checked directly or via helper function, but find where this check is done in regen code 12:48:49 you will almost certainly find the block where statues get regen disabled 12:52:47 !vault minmay_statue_bubble 12:52:48 1/1. https://github.com/crawl/crawl/blob/master/crawl-ref/source/dat/des/variable/mini_monsters.des#L2536 12:53:14 whoever replaced silver statues with obsidian statues failed to account for a couple of things 12:53:18 1. this vault can have statues completely closing it off, requiring lrd/dig to get inside 12:53:22 2. obsidian statues have mesmerise 12:53:54 ok, I can take a look at that 12:54:41 you can break LOS with the walls so it's not a huge deal but 12:54:55 (p.s. remove this vault) 12:55:23 minmay is one of our top devs 12:55:27 how could we remove it 12:59:57 NSUBST: ? = 1:. / *:? 13:00:40 or i guess since it's symmetrical and will rotate randomly anyway, you could just change one of the glyphs in the map to a . 13:11:18 yeah, the original idea (which wasn't a GOOD idea) was to have a variable number of openings, possibly 0 13:11:28 that doesn't work very well when the statue has mesmerise 13:13:58 right, i was just suggesting a way to keep the variable number of openings but make 0 impossible 13:25:25 Unstable branch on crawl.akrasiac.org updated to: 0.22-a0-351-g29664b4 (34) 13:52:28 ok it would appear the difference between statue no-regen and zombie no-regen is that zombie no-regen can be affected by the regen status whereas statue no-regen cannot 13:52:47 in that zombies with the regen status will heal 13:53:19 i don't actually see a reason why statues can't be given the M_NO_REGEN flag though, and that would fix the description issue 13:55:42 now i just need to figure out how to compile crawl... 14:07:19 NormalPerson7: On Unix it's really pretty straightforward 14:08:13 I'm back with today's stupid question: https://github.com/elliptic/dcss_scoring last commit 5 years ago - is this still approximately what CAO runs or is there some newer repository, please? 14:16:04 so how do you do it with unix pinkbeast? 14:16:20 ./configure ; make 14:16:38 apparently this file existed 14:16:40 https://github.com/crawl/crawl/blob/master/crawl-ref/INSTALL.txt 14:16:44 haven't read it yet 15:09:10 ...and after all of that i got an error 15:09:39 just classic 15:14:31 -!- amalloy is now known as amalloy_ 15:22:12 getting what appear to be syntax errors from windows files doesn't seem like a good thing... 15:23:06 syntax errors like expecting semicolons in extra places 15:47:42 right, i found an error in my PR 15:47:48 great 15:48:52 heteroy (L16 DsFi) ASSERT(!invalid_monster(&mons)) in 'mon-death.cc' at line 1904 failed. (Elf:3) 15:55:06 uh oh 16:03:07 ok so i wasted five hours of my life getting crawl to compile 16:03:12 that's depressing 16:03:16 but it finally worked 16:03:18 at long last 16:04:35 03NormalPerson702 07https://github.com/crawl/crawl/pull/732 * 0.22-a0-358-g2e7b0c8: A couple of bugfixes 10(4 minutes ago, 1 file, 2+ 2-) 13https://github.com/crawl/crawl/commit/2e7b0c802d26 16:17:24 canofworms do you have to re-compile crawl after you make every change? 16:17:28 cos it takes forever 16:17:45 depends on your change 16:17:53 some aspects of crawl depend on external data after compilation 16:18:09 like vaults and descriptions, I believe 16:18:19 but most source code, yes? 16:18:23 if you're testing a vault you just need to re-launch crawl 16:18:25 yes 16:18:27 ok 16:18:44 if you've already compiled once, re-compilation time can be shorter depending on what the change is 16:18:52 changed 1 line of source code and now it's taking 10 mins to compile, that's all 16:19:05 ok 16:19:07 right 16:19:22 thx 16:19:26 basically when you recompile the compiler only recompiles files that need to be changed 16:19:33 ok 16:19:39 so don't delete crawl.exe 16:19:43 the change that tends to force a complete recompile is when you change enums because those are referenced in every file 16:19:52 kk 16:20:43 the files my compiler looks at are the .o files, if those don't need to be changed my compiler skips compiling those 16:22:08 right 16:22:17 ye same for me 16:41:46 New branch created: pull/734 (4 commits) 13https://github.com/crawl/crawl/pull/734 16:41:47 03NormalPerson702 07https://github.com/crawl/crawl/pull/734 * 0.22-a0-260-g98a31b2: All commits up to 9th April 10(3 days ago, 261 files, 1333+ 1142-) 13https://github.com/crawl/crawl/commit/98a31b2f0f5a 16:41:47 03NormalPerson702 07https://github.com/crawl/crawl/pull/734 * 0.22-a0-261-g1ff23eb: gammafunk's Hellbinder wizlab nerf 10(3 days ago, 1 file, 4+ 4-) 13https://github.com/crawl/crawl/commit/1ff23eb071b2 16:41:47 03NormalPerson702 07https://github.com/crawl/crawl/pull/734 * 0.22-a0-351-gc641d5d: Merge the two branches, to try to start again 10(2 days ago, 0 files, 0+ 0-) 13https://github.com/crawl/crawl/commit/c641d5d7817c 16:41:47 03NormalPerson702 07https://github.com/crawl/crawl/pull/734 * 0.22-a0-352-g6c86498: Make statue descriptions accurate about no-regen (#11448) 10(11 minutes ago, 1 file, 5+ 5-) 13https://github.com/crawl/crawl/commit/6c86498a091f 16:43:05 mission accomplished, at last 16:43:10 :horray: 16:46:23 I see you got it to build. 16:48:18 ye it took like 3 hours 16:48:23 tiles just won't work 16:48:31 but i got it to compile on console 16:49:05 tiles needs a bunch of extra stuff, and if you're on windows or os x you need to make sure you pulled all the submodules too 16:54:54 right 16:55:16 well console is basically sufficient for anything non-tiles related 18:18:48 Unstable branch on underhound.eu updated to: 0.22-a0-351-g29664b41f0 (34) 18:47:29 -!- amalloy_ is now known as amalloy 20:09:50 yo. Does anyone have time to look at my species datafication PR <3 20:15:36 -!- adibis is now known as aditya 20:43:25 -!- amalloy is now known as amalloy_ 21:09:48 -!- amalloy_ is now known as amalloy 23:28:55 -!- MadCoyote is now known as FunkyBomb