Time  Nick         Message
09:23 sfan5        merging #12505 in 10m
09:23 ShadowBot    https://github.com/minetest/minetest/issues/12505 -- More corrections to lua_api.txt by sfan5
09:55 sfan5        I guess we'll declare feature freeze to start this evening?
10:51 nrz_         nice to hear ?
11:03 sfan5        we should really go through One approval PRs and decide whether we want them or not
11:03 sfan5        e.g. #12354
11:03 ShadowBot    https://github.com/minetest/minetest/issues/12354 -- Enable the gamebar on Android by rollerozxa
11:26 erle         sfan5 in #12505, why was “-- Selected item in inventory. 0 for no item selected.” removed? is the information already included elsewhere?
11:26 ShadowBot    https://github.com/minetest/minetest/issues/12505 -- More corrections to lua_api.txt by sfan5
11:44 sfan5        there's a different section for that yes
11:55 sfan5        now if someone reviewed the same PRs I did in the last hours we could merge about five at once ;)
12:19 Krock        I updated the changelog an hour ago. not final but will save some time in two weeks
12:25 Krock        > For the y- direction, it's rotated 180 degrees around the Z axis.
12:25 Krock        #11934 why would that matter?
12:25 ShadowBot    https://github.com/minetest/minetest/issues/11934 -- Briefly explain how facedir rotations work by aerkiaga
12:30 Krock        Y- can also be achieved by 0b001110 (X+) or 0b010010 (X-) but with the difference of the side rotations
13:06 sfan5        won't that mean the texture is flipped verticall?
13:07 sfan5        +y
13:07 sfan5        I admit I didn't double check it in-game
13:18 erle         luatic sfan5. i erroneously assumed that using zstd was about space savings. it is not – it is about having fast decompression time that is largely independent of the compression level. additionally, while it never decisively beats zlib on file size with default settings, it compresses about 2 to 3 times faster for the same filesize (once your payload gets over a few megabytes it seems) and never decompresses slower than zlib.
13:28 erle         unlike zlib though, better compression takes a *very* long time, the highest “normal” compression level of zstd (19) takes about 20 to 25 times as long as the default level, while only getting space savings on the order of 10 ± 3 %.
13:31 Krock        sfan5: yes, one would be flipped. it depends on what you want. I figured that the easiest explanation might be that param2=0 equals to a Y+ surface that points into Y+ and is then rotation according to those rules
13:31 Krock        s/rotation/rotated/
13:31 erle         i correct myself: in any case where zstd compresses better (i was able to observe this from payloads of >4MB), zlib absolutely is able to take over 100 times as much for marginal compression gains.
13:32 Krock        also didn't someone write a tool to manually specify param2 values? I think that would help
13:32 sfan5        in devtest maybe
13:32 erle         Krock i believe you can change it in devtest
13:32 erle         using the devtest thingy
13:32 erle         there is a tool for it
13:32 erle         could be that wuzzy made it (?)
13:33 erle         anyway, regarding zstd i think the API documentation should clearly point out that you are not getting better compression, you are getting faster compression/decompression.
13:33 erle         sfan5 i must concede that you were absolutely right about not fiddling with default values. it is very easy to make zstd take very long. this should also have an API warning IMO.
13:35 rubenwardy   Merging #12541 #12354 #12392 #12460 in 10
13:35 ShadowBot    https://github.com/minetest/minetest/issues/12541 -- Deprecate loading a world with unresolved dependencies by rubenwardy
13:35 ShadowBot    https://github.com/minetest/minetest/issues/12354 -- Enable the gamebar on Android by rollerozxa
13:35 ShadowBot    https://github.com/minetest/minetest/issues/12392 -- Fix automatic rotate for attached entities by appgurueu
13:35 ShadowBot    https://github.com/minetest/minetest/issues/12460 -- Remove workaround for normals not matching winding order by x2048
13:38 Krock        erle: didn't find any. do you remember the name by any chance?
13:39 sfan5        games/devtest/mods/testtools/init.lua
13:39 sfan5        6:minetest.register_tool("testtools:param2tool", {
13:41 Krock        oh thank you
13:42 sfan5        semi-related: I can recommend everyone who is still using grep on the command line to try ripgrep, it's great
13:42 erle         to illustrate the scale of the zstd issues: compressing 94kb of HTML i had laying around with zlib with highest compression setting (9) takes 0.01s on my machine. the default zstd compression level (3) compresses faster, but produces bigger output. the first zstd compression level that beats zlib (5) takes about the same time as zlib – 0.01s and uses about the same amount of memory.
13:45 erle         but zstd level 19 takes about 27 times as long (0.27s) and uses 13 times the amount of memory. zstd level 21 takes over 100 times as long and uses over 125 times the amount of memory as the defaullt that just barely loses out against zlib.
13:47 erle         sfan5, what feature about ripgrep do you like the most?
13:49 erle         anyways, does anyone else think that maybe, just maybe the zstd APi addition is not worth it, *unless* people regularly compress data in excess of several megabytes. luatic what do you think?
13:50 erle         (zstd authors obviously do not care about “small” payloads, excepd in the case of giving a dictionary for compression, which the proposed API addition is not going to do)
13:52 erle         it also strongly suggests that zstd is basically *never* worth it for compression of bitmaps (i.e. maps, mapblocks) or item meta, as these payloads are simply almost always *way* too small to beat zlib at reasonable compression speeds.
13:52 MTDiscord    <Jonathon> Imagine trying to block a feature for absolutely everyone else in the community because you don't like it.
13:52 erle         basicalyl, if your data is measured in kilobytes, use zlib. if it is measured in megabytes, use zstd. unless you can use a dictionary, then zstd is better.
13:53 erle         Jonathon, the problem is half-assing it. if the API gave the option to give a zstd dictionary, then it would be good for small payloads.
13:54 MTDiscord    <Jonathon> No, as usual your trying to block features because they don't fit your view point
13:54 erle         i said “add a warning”
13:55 erle         because its obviously a huge footgun
13:55 rubenwardy   way too much noise here
13:55 sfan5        I agree
13:55 erle         rubenwardy, should i have posted this to the issue or the PR instead?
13:55 sfan5        please paste your dissertation on zstd in #minetest
14:02 erle         i am done with spamming about the topic anyway. i assumed it was more like the webp thing (where no one benchmarked it and it turned out that you don't get meaningful gains over jpeg), but it seems it is more like minetest.encode_png() (where the API is half-assed, but obv most ppl think it is better than not doing it, as it is useful and can be improved later)
14:29 kilbith      x2048 sent me this: https://media.discordapp.net/attachments/979725367071883274/998234052429631498/Screenshot_from_2022-07-17_16-25-02.png?width=1117&height=628
14:30 kilbith      https://cdn.discordapp.com/attachments/979725367071883274/998226603106631804/Screenshot_from_2022-07-17_15-55-54.png?width=1117&height=628
14:31 MTDiscord    <Jonathon> I assume that's there work on top of the post processing pr
14:31 MTDiscord    <Jonathon> *?
14:31 kilbith      yep
14:31 rubenwardy   volumetric lighting
14:32 kilbith      https://media.discordapp.net/attachments/834178262133112893/998206422846287953/unknown.png?width=1117&height=628
14:33 kilbith      5.7 will be extremely promising I tell you
14:35 ROllerozxa   indeed
15:07 kilbith      MT will take over MC in 10 years the same way that OpenTTD massively took over TTD
15:07 kilbith      we are adding critical threatening weapons and MC is self-burying
15:17 erle         well, TTD also was not a moving goal and people wanted to clone it
15:17 erle         also, i think i am not alone saying that minetest is already better for some highly relevant purposes (but this is a boring topic and belongs to #minetest anyway)
15:19 erle         kilbith where can i get the branch with volumetric lighting? given the performance of shadows, i'd like to check out if there is any machine i own where it runs nicely.
15:19 kilbith      https://github.com/x2048/minetest/tree/volumetric_light
15:20 kilbith      a 10+ yo laptop will probably not run it tho
15:21 kilbith      anyone is able to spend 250-500€ to spend on a new laptop anyway
15:21 kilbith      -to spend
15:23 kilbith      I mean sometimes we have to give up on supporting GL 2.1
16:03 MisterE[m]   <kilbith> "5.7 will be extremely promising..." <- cant wait for 5.7 XD
16:04 MisterE[m]   I wonder... will there be server-sent shaders, or will there be a large library of preloaded shaders that the server can choose from.
16:16 rubenwardy   maybe in the future there will be server-sent shaders
16:16 rubenwardy   but it'll be the same as shadows now - it's a cient side option that the server can override/disable/enable
16:23 MisterE[m]   maybe there should be a simple format and accellerated PR review process for adding new shaders to the client, so that anyone who wants a new kind of shader for their specific game can get it merged fairly easily and quickly.
16:28 MisterE[m]   I built the volumetric lighting branch, got the following error:
16:28 * MisterE[m] uploaded an image: (21KiB) < https://libera.ems.host/_matrix/media/r0/download/matrix.org/jVPCtMDmMnGzPEMxoMzinhwI/image.png >
16:34 MisterE[m]   This post processing stuff wouldn't allow in-world colored volumetric lights and shadows would it?
16:38 MTDiscord    <GreenXenith> Colored lighting is something that needs to be implemented on the server side as well, so no
16:58 Noisytoot    kilbith: Not everyone can spend 250-500€ on a new laptop and I don't know of any x86 laptops with fully free boot firmware that support anything newer than OpenGL 2.1.
16:59 MTDiscord    <x2048> @GreenXenith You can place the color into node metadata, and the client can read it. For example.
17:00 rubenwardy   A big audience for minetest is schools as well, who tend to have underpowered devices
17:00 rubenwardy   features like these need to be progressive enhancements
17:00 kilbith      everyone is able to work one week in the summer to pay a new laptop with the money range I mentionned
17:01 sfan5        I don't know what the point of this discussion is, shadows aren't mandatory
17:02 rubenwardy   yeah, please cease this
17:02 sfan5        on that note we will definitely drop shader-less (OpenGL < 2.0) support sooner or later
17:02 rubenwardy   yeah
17:02 rubenwardy   there is a difference between underpowered and ancient
17:03 rubenwardy   2.0 is also 2004
17:03 MTDiscord    <x2048> MisterE[m]: try to enable tonemapping in the settings, it might be caused by that.
17:04 ROllerozxa   yeah that error log is caused when tonemapping is disabled, works when it's enabled
17:04 MTDiscord    <GreenXenith> @x2048 that is far from a good way to implement colored light
17:05 sfan5        PSA: core dev meeting in 2 hours
17:07 MTDiscord    <MisterE> It works now and is utterly amazing
17:07 sfan5        @x2048 since you're here: have you ever looked at the other vertex types in irrlicht? will we ever need anything other than S3DVertex?
17:22 MTDiscord    <x2048> sfan5: We will need 2TCoords for normal/bump mapping, but I don't see much use for tangents, since are moving towardd per-pixel lighting.
17:23 MTDiscord    <x2048> Or maybe even won't need 2TCoords if all normal maps are aligned with textures
17:23 sfan5        I see, thanks for that information
19:02 rubenwardy   meeting time
19:02 rubenwardy   1 sec
19:11 Krock        the meeting plan is pretty much like: 1. change the IRC channel title, 2. done?
19:12 MTDiscord    <GreenXenith> I'd like to mention #12549
19:12 ShadowBot    https://github.com/minetest/minetest/issues/12549 -- liquid flow leaves an air gap before reaching the end of another liquid flow
19:13 Krock        bisect would be very helpful
19:13 Krock        haven't seen liquid changes in quite a while
19:20 MTDiscord    <Niklp> Will #7629 still be a 5.6 feature?
19:20 ShadowBot    https://github.com/minetest/minetest/issues/7629 -- Run Minetest update checker on startup by SmallJoker
19:21 sfan5        I suppose we could start there, since review 5.6.0 Milestone is also on the list
19:22 sfan5        opinions? rubenwardy, Krock
19:22 sfan5        also HuguesRoss6, @x2048, nrz_ if they're here
19:23 Krock        I gave it a test yesterday and it works, so I guess if there's already the milestone, then we can consider to merge it if that's okay
19:24 rubenwardy   it's been a while since I've tested or reviewed it, but I was ready to approve minus the Android error
19:24 Krock        the Android error is now gone thanks to sfan5
19:24 sfan5        I didn't test yet but also I'm mainly worrying about the looks of the dialogue
19:25 Krock        yes, it pretty small. the main menu does suffer from the same problem though
19:25 sfan5        its also disabled on android now isnt it?
19:25 Krock        > -DENABLE_UPDATE_CHECKER=0       \
19:25 Krock        yes
19:27 Krock        testing #12131 in the meantime
19:27 ShadowBot    https://github.com/minetest/minetest/issues/12131 -- Add `minetest.settings` to CSM API and allow CSMs to provide `settingtypes.txt` by AFCMS
19:31 sfan5        I guess we can merge 7629 then, I'll review/approve later
19:35 sfan5        next: #11465
19:35 ShadowBot    https://github.com/minetest/minetest/issues/11465 -- New physics overrides by Wuzzy2
19:36 sfan5        nobody has re-tested recently and there were some bugs last time so -1 for forcing it into this release
19:37 Krock        plus I really wonder why the physics overrides aren't moved into a separate struct
19:37 Krock        for clean organization and serializing
19:37 Krock        this makes the mess even worse
19:37 sfan5        if we're postponing this PR anyway this should be done first
19:38 sfan5        (by one of us, not wuzzy)
19:38 Krock        yes, I'll add it to my TODOs
19:39 Krock        then there's a few bugfixes needed
19:39 Krock        #12048
19:39 ShadowBot    https://github.com/minetest/minetest/issues/12048 -- Autoforward no longer allows sideways movement
19:40 sfan5        dunno if this is so important, neutral from my side
19:41 Krock        neither it seems important to me. a bisect would be really helpful in this case too
19:41 rubenwardy   I conagreed on phys override
19:41 rubenwardy   I added that to the milestone as it's a regression, I'm not too bothered about it
19:42 rubenwardy   it's something that can be fixed in feature freeze as well
19:42 sfan5        true
19:43 MTDiscord    <luatic> if you're at physics already, what about #12353
19:43 ShadowBot    https://github.com/minetest/minetest/issues/12353 -- Fix acceleration by appgurueu
19:43 sfan5        no
19:43 Krock        is the "regression" statement about the "phys override" or is there a bugfix I missed in that PR?
19:43 rubenwardy   autoforward
19:44 Krock        okay then it makes sense
19:44 Krock        should be rather simple to fix
19:45 erle         sfan5 the problem with autoforward and sidewards movement is that riding boats or horses is really hard without that feature
19:45 Krock        as for #12536 - I assume this could be caused by the depth sorting PR
19:45 ShadowBot    https://github.com/minetest/minetest/issues/12536 -- Misplaced leaves with visual_scale > 1
19:45 sfan5        could be
19:45 erle         holding down a button actually hurts if you do it for several minutes
19:46 sfan5        @luatic the acceleration fix need thinking about later, definitely not in this feature freeze
19:46 MTDiscord    <luatic> aight
19:48 sfan5        the leaves thing I didn't check if that was already broken in 5.5
19:48 sfan5        but I think it shouldn't be too hard to fix so we should do it for 5.6 imo
19:48 erle         GreenXenith Krock #12549 has been a problem since at least 5.4.x, possibly longer, it makes some user-built structures in games that rely on liquids touching not work in some orientations – most notably cobble generators.
19:48 ShadowBot    https://github.com/minetest/minetest/issues/12549 -- liquid flow leaves an air gap before reaching the end of another liquid flow
19:49 Krock        erle: if it's that old, it's not important enough to be a blocker for 5.6.0
19:49 Krock        please bisect it for higher chances to have it fixed in ime
19:49 erle         yeah it's not a new thing, just a curiosity
19:49 Krock        s/ime/time/
19:50 x2048        Hi everyone, am I too late?
19:50 Krock        early enough to replace my spot ^^
19:51 sfan5        you can have a quick read through the log if there's anything you might want to comment on
19:51 x2048        I can take a look at #12536
19:51 ShadowBot    https://github.com/minetest/minetest/issues/12536 -- Misplaced leaves with visual_scale > 1
19:52 sfan5        that'd be nice
19:53 Krock        no hurries though, the freeze is 2 weeks unless that's changed
19:54 MTDiscord    <x2048> July 31-ish, right?
19:54 erle         regarding the autoforward, it is a big enough issue that some people (e.g. me) actually kept using 5.4.1 to play games where autoforwarding long distances is a thing (i.e. boats in every game with big oceans, walking on the nether roof in mcl, flying long distance on some servers)
19:55 Krock        x2048: yes
19:57 rubenwardy   The standard feature freeze length is 1 week, accordingto the dev wiki
19:57 rubenwardy   well, *default/minimum. It's usually extended
19:57 erle         Krock, i found bisection incredibly hard last time i tried due to the lack of submodules + tight irrlichtmt coupling and the buggy mess that is incremental builds (espcially when going backwards, you basically have to rebuild everything in a clean directory). do you have tips on how to do it?
19:58 Krock        you don't need to clean. bisect and checkout a specific IrrlichtMt version
19:59 Krock        granted, that version could be added to the Releases page of Minetest for convenience
20:01 sfan5        next ->
20:01 Krock        one last input: perhaps it would make sense to update the translations soon
20:01 sfan5        yes I'll do that next week
20:02 Krock        great
20:02 sfan5        do we want to have a look at this list? do we want to have a look at this
20:02 sfan5        uh
20:02 erle         Krock, i definitely need to have a clean build environment when i go back in time. i remember sfan5 explained once why cmake/make can't handle this, but i forgot what it was. anyways, if i figure it out, i'll tell later.,
20:02 sfan5        this list https://github.com/minetest/minetest/pulls?q=is%3Aopen+is%3Apr+label%3A%22One+approval%22
20:02 rubenwardy   update translations is usually done at the start of feature freeze, along with string freeze, to allow final updates
20:02 Krock        as long it's not features, they can still be merged later on. as for the feature PRs I think it's okay as-is with the milestone
20:02 rubenwardy   and then again at the end
20:02 rubenwardy   yeah that makes sense
20:03 sfan5        yes I'm not suggesting adding any to the milestone
20:03 sfan5        I guess we don't have anything to discuss about them then
20:03 rubenwardy   https://dev.minetest.net/Releasing_Minetest
20:03 Krock        the milestone takes care of that topic, and we had long time enough to add PRs to it
20:04 MTDiscord    <Jonathon> it would be nice if https://github.com/minetest/minetest/pull/11939 is considered, since that is something on the client, rather than delaying it another release
20:04 Krock        gtg, sorry. enjoy your evening.
20:05 MisterE[m]   yes, that would be nice
20:06 sfan5        I'll also tag a new IrrlichtMt release next week
20:08 MisterE[m]   oh, something to consider: adding a new movement physics group: frictionless. That would allow you to have blocks that act like arcade-game ice. When on the block, you can move and not slow down, whatever direction you start moving in, you immediately reach full speed in that direction. You keep moving in that direction until you either reach the end of the node(s) or you move in another direction.
20:08 MisterE[m]   actually, that should be in a feature request, not here
20:09 MTDiscord    <x2048> I'd like to add #12552 to 5.6.0.
20:09 ShadowBot    https://github.com/minetest/minetest/issues/12552 -- Interlaced stereo mode is broken
20:09 rubenwardy   you started the freeze but there's still 2 PRs on the list. I guess it's semi-started (:
20:09 rubenwardy   *feature PRs
20:09 sfan5        they are excluded obv
20:09 sfan5        x2048: sure
21:31 Zughy[m]     exactly what features is MT 5.6 gonna have?
21:35 rubenwardy   https://github.com/minetest/blog/issues/23
21:37 rubenwardy   https://dev.minetest.net/Changelog#5.5.0_.E2.86.92_5.6.0
21:37 erle         rubenwardy, 9-slice images are new? i thought i had seen some already somewhere
21:37 rubenwardy   new for images, not for backgrounds and buttons
21:45 erle         regarding position/speed sanitization … if a player is adjacent to the boundary, what's preventing them of firing an arrow right beyond the boundary or goading the game into doing whatever else past that position in some other way?
22:26 MTDiscord    <savilli> does anyone want to second approve #11885?
22:26 ShadowBot    https://github.com/minetest/minetest/issues/11885 -- Add unittests for item movement code by savilli
22:26 MTDiscord    <savilli> I don't want to rebase it again