00:10:56 when 0.25.0 gets tagged, please do it at least 24hrs before the tourney. That way I can make sure CPO is running the release version even though tourney starts early in the morning here 00:13:37 I don't think that will be a problem to do; we've done same-day-as-tourney releases before, but I think that was mostly due to people having free time 00:13:52 but there are a bunch of active people working on stuff now, so should be easy to guarantee 00:17:59 🙏 01:38:34 Unstable branch on crawl.develz.org updated to: 0.26-a0-37-g50178cf780 (34) 01:44:47 ebering: just realised the Okawaru Tier 3 banner requires winning a 15-rune game in 50k turns. Just checking it's deliberately that difficult 01:44:47 greensnark: !lg ... -graph doesn't work any more 01:49:05 alexjurkiewicz: yes. it's harder than the old conqueror t3 because it increases the rune count 02:11:01 Windows builds of master branch on crawl.develz.org updated to: 0.26-a0-37-g50178cf780 02:40:40 ebering: in the player_combo_score view, combos and won_combos are equal for all players in the tournament dataset you're using 02:40:49 think this might be a bug, but am not good at sql 02:58:29 You pick up Sif Muna's Almanac of Dhcmrlchtdj and begin reading... 02:58:38 has our name gen always made names like that? 03:29:44 no, they're usually more reasonable than that 03:32:02 Unstable branch on crawl.kelbi.org updated to: 0.26-a0-37-g50178cf780 (34) 03:32:58 Fork (bcadrencrawl) on crawl.kelbi.org updated to: 0.22.1-2757-gc346607f2f 03:48:15 Fork (gooncrawl) on crawl.kelbi.org updated to: 0.22-a0-996-ge17abb23d8 03:55:07 !git ae26e8c9cb 03:55:09 %git ae26e8c9cb 03:55:09 07advil02 * 0.26-a0-33-gae26e8c: Temporary check to bridge a change in client.js 10(34 hours ago, 1 file, 3+ 1-) 13https://github.com/crawl/crawl/commit/ae26e8c9cb12 03:55:27 advil: re that, which server was this for? 03:57:03 the usual cache-busting method here is for servers to disable etags, serve with a 1-year storage time, and add a version number to the CSS url 05:41:35 New branch created: pull/1436 (1 commit) 13https://github.com/crawl/crawl/pull/1436 05:41:35 03Gittourarmy02 {GitHub} 07https://github.com/crawl/crawl/pull/1436 * 0.26-a0-38-g9a63b16: Resurrect the dead orc with the full name. 10(3 minutes ago, 1 file, 1+ 1-) 13https://github.com/crawl/crawl/commit/9a63b16dc947 06:07:04 New branch created: refactor-describe (3 commits) 13https://github.com/crawl/crawl/tree/refactor-describe 06:07:04 03Aidan Holm02 07[refactor-describe] * 0.26-a0-38-g682f2c3: Make some tile picking code unconditional 10(2 hours ago, 3 files, 8+ 24-) 13https://github.com/crawl/crawl/commit/682f2c3f2f98 06:07:04 03Aidan Holm02 07[refactor-describe] * 0.26-a0-39-gae4997d: Refactor feature extra tiles handling 10(6 weeks ago, 1 file, 48+ 34-) 13https://github.com/crawl/crawl/commit/ae4997d50bc6 06:07:04 03Aidan Holm02 07[refactor-describe] * 0.26-a0-40-gfe7b5d5: Remove strange FLYING flag compat code 10(16 minutes ago, 3 files, 4+ 6-) 13https://github.com/crawl/crawl/commit/fe7b5d5a6f32 07:14:09 aidanh it's cao of course 07:14:28 yeah, i figured :) 07:15:26 I'm not actually sure why client.js and a few other things aren't served out of game_data, which would get versioning 07:15:43 client.js serves the main page as well 07:16:00 ah 07:16:20 which, yeah, not ideal 07:22:10 !tell gammafunk Dhcmrlchtdj is a borges reference, to The Library of Babel 07:22:11 advil: OK, I'll let gammafunk know. 07:22:24 oh sorry you're in channel 07:23:18 as are some other things that vault places 08:19:33 03Peter Gerlagh02 07https://github.com/crawl/crawl/pull/1421 * 0.25-a0-1138-g5b9b782: move ;; check from items.cc to main.cc, and remove global variable 10(6 minutes ago, 2 files, 6+ 15-) 13https://github.com/crawl/crawl/commit/5b9b782047f5 08:52:36 03Peter Gerlagh02 07https://github.com/crawl/crawl/pull/1421 * 0.25-a0-1139-g0d7788c: moved assignment of prev_cmd = cmd 10(7 minutes ago, 1 file, 6+ 6-) 13https://github.com/crawl/crawl/commit/0d7788cbf9e0 08:59:56 In the autopickup with ;; pull request (https://github.com/crawl/crawl/pull/1421) I stumbled upon the fact that prev_cmd = cmd used to happen before process_command(cmd) for some (but not all) values of cmd. 09:00:30 to implement a suggestion from a dev I had to move it so that it always happens after process_command(cmd) 09:04:43 This might have a lot of repercussions, because a -lot- of code is covered by process_command(). Would someone be willing to check whether this would cause any bugs, or would you argue that this change shouldn't be made? 09:09:21 the previous implementation of ; (CMD_INSPECT_FLOOR) would have been broken by this change, but on the other hand one of the code reviewers assumed prev_cmd = cmd would happen after process_command(cmd), because that's the natural way to do things. 09:14:04 hm, i don't think anyone could say with anywhere near certainty that that won't cause bugs 09:14:12 noone understands that code very well 09:20:59 New branch created: view-flash-tile (4 commits) 13https://github.com/crawl/crawl/tree/view-flash-tile 09:20:59 03Aidan Holm02 07[view-flash-tile] * 0.26-a0-38-g682f2c3: Make some tile picking code unconditional 10(5 hours ago, 3 files, 8+ 24-) 13https://github.com/crawl/crawl/commit/682f2c3f2f98 09:20:59 03Aidan Holm02 07[view-flash-tile] * 0.26-a0-39-gae4997d: Refactor feature extra tiles handling 10(6 weeks ago, 1 file, 48+ 34-) 13https://github.com/crawl/crawl/commit/ae4997d50bc6 09:20:59 03Aidan Holm02 07[view-flash-tile] * 0.26-a0-40-gfe7b5d5: Remove strange FLYING flag compat code 10(4 hours ago, 3 files, 4+ 6-) 13https://github.com/crawl/crawl/commit/fe7b5d5a6f32 09:20:59 03Aidan Holm02 07[view-flash-tile] * 0.26-a0-41-g6c8d4f1: Implement single-tile flashes on local tiles. 10(71 minutes ago, 6 files, 53+ 34-) 13https://github.com/crawl/crawl/commit/6c8d4f1df6e2 09:21:33 03Aidan Holm02 07* 0.26-a0-38-g682f2c3: Make some tile picking code unconditional 10(5 hours ago, 3 files, 8+ 24-) 13https://github.com/crawl/crawl/commit/682f2c3f2f98 09:21:33 03Aidan Holm02 07* 0.26-a0-39-gae4997d: Refactor feature extra tiles handling 10(6 weeks ago, 1 file, 48+ 34-) 13https://github.com/crawl/crawl/commit/ae4997d50bc6 09:21:33 03Aidan Holm02 07* 0.26-a0-40-gfe7b5d5: Remove strange FLYING flag compat code 10(4 hours ago, 3 files, 4+ 6-) 13https://github.com/crawl/crawl/commit/fe7b5d5a6f32 09:26:47 I suppose we can fix the bugs when they show up? 09:33:21 a better approach is to change process_cmd to take (command_type this_cmd, command_type prev_cmd) 09:34:45 that would be entirely safe, and it's better since you're not adding a new dependency on global state (something we're trying to eliminate) 09:50:09 process_cmd never uses prev_cmd 09:50:42 that is to say, it never uses the proper prev_cmd because prev_cmd is overwritten before process_command is called 09:52:00 right, so you should add a parameter that passes in the previous command, with the correct value (before it's overwritten) 09:52:51 I get it. That would work 09:53:16 but wouldn't that leave the code uglier, when we could make it nicer? 09:54:24 changing a dependance from global state to a named parameter is prettier 09:55:02 it makes the functional relationship clearer and easier to read 09:55:07 i also have a refactoring of this code in the works, and i'd rather not deal with merge conflicts, this code is hard enough to understand as it is :) 09:55:34 re the alternative of just swapping the code so prev_cmd is set after process_command() runs, that's quite likely to break things 09:55:37 alexjurkiewicz: thanks I'll take a look, but I think that's the property of the dataset. I'm stress-testing on the full 0.24 tournament to find slow queries still 09:55:42 so I have a larger dataset to check 09:56:47 that caught a couple issues with streak handling in the schema 10:02:58 ok, I'll make sure not to give you merge conflicts :) It will be a bit ugly because there'll be an extra (constant) variable which is assigned about 4 lines before it is used. Personally I think that a properly implemented global prev_cmd (and possibly current_cmd) would be prettier. 10:04:56 aidanh: Thanks for the feedback - The reason I put spaces in the #includes was because the other #includes in that test file I was working on were that way, so I was going for consistency 10:05:13 But I can drop all the superfluous spaces in those test file #includes 10:11:03 petercordia: no worries; I agree, but this code is going to take a bit of detangling before that can happen 10:11:49 elliotd123: ah, that's because there are four groups of headers, separated by line breaks, but no separation within each group 10:13:36 I just looked at the code. making process_command(current_cmd, prev_cmd) whilst crawl_state.prev_cmd isn't working properly is really messy. I don't think I can do it today (it's making me feel ill) and I'm not sure I can do it at all. 10:14:43 no worries, no rush 10:15:24 I'll take a look at it in a bit and see if I can add that 10:15:50 fwiw this is probably some of the nastiest code in crawl 10:20:52 The build passed. (view-flash-tile - 6c8d4f1 #13536 : Aidan Holm): https://travis-ci.org/crawl/crawl/builds/692874141 10:26:12 this query is the sole slow query culprit for updates that I'm trying to optimize 10:26:15 https://gist.github.com/ebering/f8cea7fae26330391b59464166cdc0ce 10:26:18 (see the update, second) 10:26:45 the select version is fast and uses ref/eq_ref for all of its joins except the primary (where it must be ALL since we're running all rows) 10:27:13 the update version decides to use the full index for both copies of the games table and I don't understand why 10:27:23 or what hints to whisper to mysql to get it to go fastly 10:30:55 ARGH 10:31:03 the manual says that updates are optimized the same way as selects 10:32:05 that is clearly false 10:35:12 what version of mysql is this on? 10:35:56 8.0.19 10:36:49 is this real mysql or mariadb? 10:36:52 aidanh: makes perfct sense, thanks. 10:36:59 real mysql 10:39:35 aha 10:39:55 if I enable the JOIN FIXED ORDER hint on the optimizer for the select I get the update's execution plan 11:01:28 ok, the query optimizer hints are dumb and specifyingn the good join order didn't help but refactoring the query did 11:11:39 hm but my refactor is wrong 11:16:49 ok, now I have a good factorization that is correct 11:17:06 and runs fast 12:11:24 actually it looks like at least some of the existing use of crawl_state.prev_cmd is bugged 12:11:30 drj (L6 VSIE) Crash caused by signal #6: Aborted (D:5) 12:11:42 !source main.cc:1737 12:11:44 https://github.com/crawl/crawl/blob/master/crawl-ref/source/main.cc#L1737 12:16:36 !crashlog 12:16:38 21975. Tripartite, XL2 SpAM, T:520 (milestone): https://crawl.kelbi.org/crawl/morgue/Tripartite/crash-Tripartite-20200516-050928.txt 12:43:21 New branch created: game_scale (1 commit) 13https://github.com/crawl/crawl/tree/game_scale 12:43:21 03advil02 07[game_scale] * 0.26-a0-41-g7356ca5: Add a global scaling option for local tiles 10(7 minutes ago, 6 files, 35+ 17-) 13https://github.com/crawl/crawl/commit/7356ca58d64c 12:44:33 !cmd !crashlog 12:44:33 Command: !crashlog => !lm ${*:-!ignoredcrashers} crash -log 12:53:25 New branch created: main-prev-cmd (1 commit) 13https://github.com/crawl/crawl/tree/main-prev-cmd 12:53:25 03Aidan Holm02 07[main-prev-cmd] * 0.26-a0-41-g0d7b898: Fix anti-reckless-fighting command 10(7 minutes ago, 2 files, 11+ 5-) 13https://github.com/crawl/crawl/commit/0d7b898a161a 12:54:13 advil: re your "why is this event handled both here and in ui.cc?", that's because both files contain two independent event loops, running at different times 12:54:40 is the one in the ui.cc file just not used during the regular game loop? 12:54:46 that's what it seemed like 12:55:08 ui.cc runs whenever there are any widgets, including in-game menus, main menu, etc; the other one is mostly just the main gameplay view 12:55:18 ah, I see 12:55:46 it's not, for hysterical raisins 12:57:27 i have a wip patch for that, but it's fairly invasive 13:00:24 03advil02 07[game_scale] * 0.26-a0-42-g06e9323: Replace a question with an explanation (aidanh) 10(19 seconds ago, 2 files, 11+ 1-) 13https://github.com/crawl/crawl/commit/06e93238f64e 13:02:16 I realize this branch might be termed a "feature" but it's local-tiles-specific and is a really often-requested thing, so it might be nice to merge before a release 13:02:23 I've only tested on mac 13:03:19 that sounds reasonable yeah 13:04:03 I can give it some testing tomorrow 13:04:09 yeah, linux testing would be good 13:04:46 since the biggest use case is linux with dodgy or non-existent os-level high dpi support 13:06:21 I guess windows testing wouldn't hurt either, if anyone can do that 13:06:30 Unstable branch on crawl.akrasiac.org updated to: 0.26-a0-40-gfe7b5d5 (34) 13:07:14 I mean, I don't have a hidpi monitor, but i imagine it'll still work 13:07:48 I really have to get my hands on some snazzy qhd monitors 13:08:39 yeah testing at game_scale=2 should work on pretty much any modern monitor, might even be valuable for people with poor vision 13:08:45 Build failed for main-prev-cmd @ 0d7b898a https://github.com/crawl/crawl/actions/runs/120009000 (failed: Catch2/codecov) 13:42:42 The build passed. (game_scale - 7356ca5 #13538 : advil): https://travis-ci.org/crawl/crawl/builds/692928072 14:00:32 this sort of scaling will result in blurry text, but that's something that can be fixed later 14:06:13 it doesn't for me 14:07:44 it's also not supposed to in that text should still be rendered using device pixels 14:08:49 though text seems to be glitchy at the more absurd values, not sure why 14:09:05 probably something to do with just using division 14:09:39 here's an example at scale 6; text isn't rendering right but it's also not blurry: https://www.dropbox.com/s/5a76ffu7wjai161/Screenshot%202020-05-30%2014.09.01.png?dl=0 14:10:44 alexjurkiewicz: you should (if you're running a large test dataset) see a speedup in the clan scoring section of quite a bit 14:10:53 also sorry I changed the schema in a big way 14:10:58 well, not that big 14:11:01 but big enough 14:11:39 aidanh though I guess blurriness is going to be affected by dpi 14:11:57 maybe 14:15:34 oh OK if text rendering is still using device pixels that should be OK then 14:16:24 aidanh: I've noticed that these tests are using opening braces on same line - Should I change that as well as I'm touching these test files? 14:16:41 alexjurkiewicz: I think the banner display in player.mako has a bug as well, since it does not match up with the GROUP_CONCAT banner tier listing shown in banner order 14:16:53 that clipping looks like each letter has a fixed height for the visible portion 14:18:02 oh 14:18:02 iirc glyphs are blitted into fixed size cells on a single texture, so that's probably why there's clipping 14:18:09 alexjurkiewicz: nevermind re bannners that's on me for not sending real data to the template! 14:19:55 elliotd123: that'd probably be good yeah, thanks 14:20:57 hm I guess it'd also be nice if scaling images didn't antialias 14:21:38 yeah, they look bad at super high values 14:22:01 it's possible I should cap this at 4, I don't know if anyone would use anything higher for non-testing purposes 14:23:34 oh yeah the clip value for caps definitely looks like it's using the lowercase values 14:23:51 there must be a calculation somewhere that is wrong only if you do something insane like this 14:24:14 or even just some hard max on the glyph size 14:25:28 i think if you measure the height of each glyph in that image, you'll find its the same 14:25:37 probably a power of two 14:26:15 I suspect it's a hard max that starts from the top of the glyph 14:26:24 right, in the texture 14:26:41 based on this: https://www.dropbox.com/s/fszdw7waetwzf7x/Screenshot%202020-05-30%2014.26.37.png?dl=0 14:26:45 that's 8x 14:27:30 iirc glyphs are rendered on 64x64 squares on a single 1024x1024 texture 14:28:26 so if you have a glyph that's taller (or wider) the rendered quad will show part of the adjacent squares 14:30:04 this is all in fontwrapper-ft.cc 14:30:21 what we need is someone with an 8K monitor 14:31:20 or this: https://www.digitaltrends.com/home-theater/sonys-massive-new-microled-display-stands-17-feet-tall-and-packs-16k-resolution/ 14:33:26 wow that's pretty ridiculous 14:50:06 advil: I have a 4k monitor 14:53:04 bh: which OS? 14:53:12 I do have a 4k tv so I could probably test on that 14:56:42 MacOS 14:57:02 my regular screen is natively high dpi 15:01:56 03Elliot Dronebarger02 07https://github.com/crawl/crawl/pull/1427 * 0.26-a0-42-gebebe5a: Small refactoring and cleaning up some things 10(4 minutes ago, 4 files, 70+ 58-) 13https://github.com/crawl/crawl/commit/ebebe5a46adf 15:10:03 bh: the thing I'm wondering is what pixel multiplier would make local tiles look good on a 4k screen that is not natively high dpi; I suspect 4 is still around the highest that would be usable 15:11:16 if you wanted to test this there's a branch (game_scale) that has the patch to do the scaling (though slightly glitchy still), but I can definitely test it myself if you don't want to go to the trouble of building that branch 15:15:02 desktop tiles or webtiles? 15:19:30 hey guys the spacex launch is happening in 3 minutes in case you want to watch, first launch with a crew from US soil in 10 years 15:22:22 hope it goes better than that static fire test 15:22:29 bh: desktop tiles 15:26:02 floraline thanks for mentioning that, would have totally missed it otherwise 15:30:26 no problem lol, it's not every day we get to see history being made 15:51:47 -!- Shroomator is now known as Tiobot 15:54:19 did it launch 16:07:06 cebolla: it did 16:14:54 03advil02 07[game_scale] * 0.26-a0-43-g088b66a: Fix font glitches at very high device density 10(5 minutes ago, 1 file, 12+ 10-) 13https://github.com/crawl/crawl/commit/088b66a4113d 16:19:55 sry tournament team, I keep working on the game instead of tourney scripts 16:20:56 btw if it is ready we could easily put up index.html as a static file on CDO so people could link to it 16:53:44 New branch created: pull/1437 (1 commit) 13https://github.com/crawl/crawl/pull/1437 16:53:44 03Anders Papitto02 07https://github.com/crawl/crawl/pull/1437 * 0.26-a0-41-g4c9f5af: Reword hailstorm description to be less hand-wavy 10(11 minutes ago, 1 file, 4+ 5-) 13https://github.com/crawl/crawl/commit/4c9f5aff7f6d 16:54:36 03advil02 07[game_scale] * 0.26-a0-44-gf0b4f8d: Let tile_filter_scaling=false work on mipmapped textures 10(4 minutes ago, 1 file, 3+ 1-) 13https://github.com/crawl/crawl/commit/f0b4f8db1ee6 16:57:08 so that was a fun find 16:57:27 I wonder if there are other obvious opengl tweaks to make scaled pixel art look better 17:43:41 i think the rules page is ready to go up 18:03:46 -!- amalloy_ is now known as amalloy 18:11:04 yeah, it would be good to get that up soon, as we're less than a week away 18:11:06 gammafunk: You have 1 message. Use !messages to read it. 18:24:02 Unstable branch on underhound.eu updated to: 0.26-a0-40-gfe7b5d5a6f (34) 18:30:16 I did actually put the silly cao countdown thing on cdo is a temp placeholder: https://crawl.develz.org/tournament/0.25/ 18:30:20 *as a 18:31:05 nice 18:31:25 I can do a local test of the index page and then a CDO test 18:32:00 advil: what about adding said countdown to the index page in new_scoring? 18:32:38 I can try to do that myself, since this doesn't look like it's difficult to adapt 18:33:00 yeah, it would be easy to do 18:40:28 gammafunk btw if you contact server admins, can you point them to this issue: https://github.com/crawl/crawl/issues/1315 19:02:28 advil: will do 19:46:53 -!- Tuxedo[Qyou] is now known as jonge1 19:47:08 -!- jonge1 is now known as Tux[Qyou] 20:01:17 ebering: how is the total points for a player calculated? I thought it was the sum of points for each category, but this seems wrong