Minetest logo

IRC log for #minetest, 2021-03-11

| Channels | #minetest index | Today | | Google Search | Plaintext

All times shown according to UTC.

Time Nick Message
00:08 Fractalis joined #minetest
00:31 DS-minetest I've found another reason to wish µ$oft windoze wouldn't exist, even though I don't need to use it: it still affects me https://github.com/Desour/minetest/runs/2081924088?check_suite_focus=true
00:34 sfan5 why not use UUIDs instead of inventing your own
00:34 sfan5 beside the point though
00:38 DS-minetest 128 might be too small
00:38 DS-minetest s/128/128 bit/
00:39 sfan5 for what
00:40 sfan5 are you concerned with collisions?
00:40 DS-minetest yeah
00:40 DS-minetest my plan:
00:40 DS-minetest 1. gain world dominion
00:40 sfan5 idea: base part of the uuid on the current game time
00:41 DS-minetest 2. force everybody to make new objects in lua with a guid
00:41 DS-minetest 3. ...
00:41 DS-minetest 4. profit
00:41 sfan5 brilliant plan
00:42 DS-minetest the game time is just another counter
00:42 sfan5 but it's a monotonic counter
00:43 DS-minetest anyway, it doesn't really matter how you generate the numbers, does it?
00:45 DS-minetest so anything is valid, and having no collisions even in extreme situations is a thing that is better to have and not need than to not have and need
00:45 DivideByZero Vanessa: I've been getting this error on your dreambuilder_hotbar since you updated about 5 hours ago:  https://0bin.net/paste/ENjY9n2N#FArqf0H5Ojfk3XLZLtyaZxABQbwlumNgCQ8+7HUpSjU
00:52 SwissalpS joined #minetest
00:58 VanessaE damn it.
00:58 VanessaE I'll fix it asap
00:58 VanessaE must be a copypaste error while working on dreambuilder game stuff
00:59 VanessaE can anyone tell me where the ESC/Pause menu is defined?  or rather, at exactly what point in the game are its colors picked up?
01:00 DS-minetest the colors come from the formspec prepend
01:01 VanessaE DivideByZero: https://pastebin.com/GaRtriRU  replace init.lua with this and tell me if it works
01:01 VanessaE DS-minetest: ok, thanks
01:04 DivideByZero Ok, one moment Vanessa, and I'll report back.
01:05 cheapie Heh, negative formspec versions are... interesting :P
01:06 VanessaE interesting?
01:07 cheapie (same results as with anything that's not an integer - but not 0 - 0 acts like 1)
01:07 cheapie VanessaE: Default size, all elements crammed into the upper left corner, with a size of 0,0 for any elements that use a size.
01:08 DS-minetest -1 is probably casted to 0xffff, as m_formspec_version is a u16
01:09 cheapie Version 4 (how this formspec is meant to look): https://cheapiesystems.com/media/images/screenshot_20210310_190838.png
01:09 cheapie What most garbage versions (-1, 3.5, "potato", "", etc.) give: https://cheapiesystems.com/media/images/screenshot_20210310_190849.png
01:10 luk3yx VanessaE: https://github.com/minetest/minetest/blob/75eb28b95994781d52eb2c09303b1cd04e32b6c5/src/client/game.cpp#L4097-L4232
01:11 cheapie DS-minetest: Yep, sure enough, formspec_version[65535] causes it to render like 4, but 65536 makes it act like 0/1 again.
01:11 luk3yx Oh your question was already answered
01:11 cheapie (and 65540 gives correct rendering again, and so on)
01:12 DS-minetest yep, normal arithmetic modulo 2^16
01:13 cheapie That does however suggest that -1 isn't turning into 65535 though, as 65535 gets treated like 4.
01:15 VanessaE DivideByZero: fixed in git
01:16 DS-minetest weird
01:20 numzero cheapie: 65535 remains 65535, but the comparison is like `version >= 4` in most places so there is no difference between 4, 5, ..., 65535
01:20 DivideByZero Vanessa: That seems to have fixed it. Thanks for quick fix.  I'll update from git now and make sure it works for you also.
01:21 VanessaE yw
01:21 cheapie numzero: Yeah, that part makes sense - I'm mostly just curious why the negative/garbage versions act the way they do.
01:24 DS-minetest ah, there's a is_number check.
01:24 DivideByZero My server is slow to startup. I noticed skinsdb takes 13 seconds to load through all the skins.
01:24 DS-minetest so, this is why it doesn't parse the negative numbers
01:24 DS-minetest in formspec_version
01:28 DivideByZero Vanessa: git version looks good now too.
01:28 VanessaE 👍
02:10 hlqkj1_ joined #minetest
02:31 delta23 joined #minetest
02:32 BuckarooBanzai4 joined #minetest
02:32 Elouin2 joined #minetest
02:36 Hawk777 joined #minetest
02:51 Conradish006 joined #minetest
03:58 moi_ joined #minetest
03:59 moi_ joined #minetest
04:02 Biggvs_Richardvs Hi all, got a bug report for aliveai:
04:02 Biggvs_Richardvs https://justpaste.it/4tcdh
04:02 craigevil joined #minetest
04:04 Biggvs_Richardvs Running MT 5.3.0, not sure if upgrade will help
04:04 MTDiscord <Jonathon> make a issue on the repo https://github.com/AiTechEye/aliveai
04:08 Biggvs_Richardvs I did.
04:09 Biggvs_Richardvs But I posted on IRC about it in the dev channel and they asked me to post it here instead :)
04:09 MTDiscord <Jonathon> yeah, i see that sorry
04:09 Biggvs_Richardvs np
04:09 MTDiscord <Jonathon> this guy......does not use git
04:09 MTDiscord <Jonathon> well, in the usual sense
04:09 MTDiscord <Jonathon> so i cant even look at difs
04:11 MTDiscord <Jonathon> *correction, i can, its just a step longer
04:12 Biggvs_Richardvs neato. It's a cool mod, would like to have it working. Willing to offer my server up as guinea pig
04:13 Biggvs_Richardvs I probably should have mentioned that I turned off every other mod, like all of them, and ran nothing but aliveAI and all its associated mods (aliveai_aliens etc) and it still crashed
04:14 MTDiscord <Jonathon> do you have a way of replicating the error?
04:14 MTDiscord <Jonathon> >compatible with minetest 0.4.16 and newer (tested in 5.4-dev) this.......is odd
04:16 Biggvs_Richardvs Ya basically whenever we get attacked by a mob, there are a bunch of "rapid fire" hits on us, we die, and then the server crashes. Like every time
04:17 Biggvs_Richardvs It's like machine gun rapid fire  - go from full health to dead in about 3 seconds
04:17 Biggvs_Richardvs That may be by design - there's some nasty mobs in there :)
04:19 MTDiscord <Jonathon> you could test 5.4, not sure if that would actually do anything
04:20 Biggvs_Richardvs Ya that's what I'm working on
04:20 Biggvs_Richardvs Server is an older (18) version of Mint, so working on upgrading to 5.4 now
04:21 MTDiscord <Jonathon> seems to be 2 ways of fixing it, nil check(then do what?) or figure out why its getting nil
04:28 craigevil joined #minetest
04:29 Biggvs_Richardvs Ok I'm now upgraded to 5.4.0 and still got exactly the same crash
04:35 MTDiscord <Jonathon> not suprised
04:37 Biggvs_Richardvs The version of the mod I'm using is the latest from git
04:39 MTDiscord <Jonathon> do you actually want a fix, or just a dont crash my server till the maintainer patches it?
04:40 Biggvs_Richardvs yes :)
04:40 Biggvs_Richardvs I'll take either
04:42 MTDiscord <Jonathon> block comment out aliveai.searchobjects=function(self) (the whole function) and try adding below it  aliveai.searchobjects=function(self)      return end
04:43 MTDiscord <Jonathon> since it looks like at the start it checks a bunch of stuff, and if those conditions are meet it doesnt execute anything, so all your doing is making sure it never executes
04:43 MTDiscord <Jonathon> (im not at a computer where i can test that)
04:43 MTDiscord <Jonathon> not sure if that will break things further on in the mod or not
04:44 Biggvs_Richardvs ok. Worth trying. Just a sec
04:45 Biggvs_Richardvs Is that in aliveai_threats?
04:46 MTDiscord <Jonathon> . /home/minetest/.minetest/mods/aliveai/aliveai/event.lua:600 its literally in your error message lol
04:50 kamdard joined #minetest
04:51 Biggvs_Richardvs hah.  I thin kthere's a few files mentioned in there - just wanted to make sure
05:00 MTDiscord joined #minetest
05:06 Biggvs_Richardvs So I commented out the whole function and put this in:
05:06 Biggvs_Richardvs aliveai.searchobjects=function(self)      return end
05:07 Biggvs_Richardvs end is on another line
05:08 Biggvs_Richardvs getting  /home/minetest/.minetest/mods/aliveai/aliveai/event.lua:590: unexpected symbol near '#'
05:08 Biggvs_Richardvs I'm a noob
05:08 Biggvs_Richardvs Something stupid simple I know
05:11 Biggvs_Richardvs WAit. I was using #'s for comments.   Bash habits
05:12 MTDiscord <Jonathon> oof
05:12 MTDiscord <Jonathon> is it "working" now?
05:12 Biggvs_Richardvs lol What do I use for comments in lua?
05:13 Biggvs_Richardvs --[[ got it
05:14 Conradish006 joined #minetest
05:16 MTDiscord <Jonathon> and close with ]]--
05:18 Biggvs_Richardvs Ok it's "working" but now I don't seem to have any mobs
05:18 Biggvs_Richardvs But it's not crashing with aliveai running now, so thank you
05:19 MTDiscord <Jonathon> "working"
05:19 MTDiscord <Jonathon> does it have spawn eggs or something?
05:19 MTDiscord <Jonathon> or can you call /spawnentity mod:entityname
05:20 Biggvs_Richardvs Not sure
05:21 Biggvs_Richardvs How could I tell?
05:21 MTDiscord <Jonathon> look in inventory?
05:23 MTDiscord <Jonathon> ofc aliveai cant register from a entity def itself
05:23 MTDiscord <Jonathon> 🤦‍♂️
05:23 MTDiscord <Jonathon> that would make it to easy
05:24 olliy joined #minetest
05:25 MTDiscord <Jonathon> try /spawnentity aliveai_ants:ant
05:25 MTDiscord <Jonathon> seems to register them as modname:mobname
05:37 zceejkr joined #minetest
05:58 Conradish006 joined #minetest
06:06 olliy_ joined #minetest
06:11 Flabb joined #minetest
06:55 Conradish006 joined #minetest
06:55 craigevil joined #minetest
07:08 Conradish006 joined #minetest
07:30 emacsomancer joined #minetest
07:44 craigevil joined #minetest
08:00 ShadowNinja joined #minetest
08:39 FeXoR joined #minetest
08:53 calcul0n_ joined #minetest
08:58 Conradish006 joined #minetest
09:03 YuGiOhJCJ joined #minetest
09:06 TomTom joined #minetest
09:22 emacsomancer joined #minetest
10:00 Conradish006 joined #minetest
10:01 Talkless joined #minetest
10:03 hlqkj joined #minetest
11:00 YuGiOhJCJ joined #minetest
11:11 moi_ joined #minetest
11:44 tech_exorcist joined #minetest
11:47 proller joined #minetest
11:58 tech_exorcist joined #minetest
11:58 z812 joined #minetest
12:07 calcul0n__ joined #minetest
12:08 Conradish006 joined #minetest
12:37 qj joined #minetest
12:43 bingfengzs joined #minetest
12:46 olliy joined #minetest
12:55 Verticen joined #minetest
13:03 Extexo1 joined #minetest
13:05 celeron55_ joined #minetest
13:05 Rafi59_ joined #minetest
13:06 Taserface joined #minetest
13:06 TC01 joined #minetest
13:06 basxto2 joined #minetest
13:07 sknebel_ joined #minetest
13:09 mransom joined #minetest
13:11 hlqkj1 joined #minetest
13:23 SwissalpS joined #minetest
13:25 celeron55 joined #minetest
13:27 sknebel joined #minetest
14:04 Talkless joined #minetest
14:08 kamdard joined #minetest
14:10 hlqkj1_ joined #minetest
14:28 Wuzzy joined #minetest
14:36 qj joined #minetest
14:49 Conradish006 joined #minetest
14:56 Conradish006 joined #minetest
15:01 z812_ joined #minetest
15:02 galex-713 joined #minetest
15:06 FreeFull joined #minetest
15:09 kamdard joined #minetest
15:27 VanessaE although there appears to be no present need for it, is there a way to tell the client "here's a piece of text, cache it client-side and let me call upon it with some handle later"?  like node metadata but without being tied to a node
15:29 VanessaE I've been giving some thought to my revised UI code, since it did increase the formspec size a bit (not by much compared to before). I wouldn't mind shoring that up, juuuuust in case some dinosaur is still using dial-up :)
15:29 sfan5 no
15:29 sfan5 that sounds so specific that it could be solved by CSM
15:29 sfan5 (once it's implement in an unspecified timeframe)
15:29 sfan5 implemented*
15:29 VanessaE well if SSCSM becomes a thing I suppose that would do the job
15:30 VanessaE if I run into a dial-up user, I'll just tell them to get out of the 1990's ;)
15:31 MTDiscord <Warr1024> If you're worried about bandwidth, I did some experiments and looked into per-packet compression using stuff like zlib, lzo, etc.  It would probably help a lot with formspecs ... but I found it only had like a 2% to 5% impact overall because block transfer dominated bandwidth.
15:31 MTDiscord <Warr1024> For all the trouble it'd entail, 5% doesn't seem worth it.
15:32 VanessaE Warr": well it's just that I have this fleeting worry that the extra 2k of formspec data might negatively impact like, one person somewhere :P
15:32 VanessaE though I suppose UI was already generating like 10k of it :P
15:32 MTDiscord <Warr1024> I worry about network issues too, and there are people who play on cellular or satellite, which is basically the modern equivalent of dial-up (only somehow worse in some cases).
15:32 VanessaE haha
15:34 Conradish006 What is satellite? :P
15:34 MTDiscord <Warr1024> Some users are so bandwidth-constrained that I'd considered tunneling UDP over TCP so I could get cross-packet compression, but unfortunately those networks are also the kind that have the worst packet loss so TCP meltdown would be a problem.
15:35 ircSparky joined #minetest
15:35 VanessaE well I guess since it's not really a problem (more a "could I do better?" thing) then I won't worry about it
15:35 Conradish006 I one time stayed at a place that had satellite internet, and the whole week it was stormy.
15:35 MTDiscord <Warr1024> Satellite internet is a punishment for living too far from a city, where they charge you $75 a month for what 256k ISDN would have been like if the ISDN line had 2000ms latency, and with a 10GB monthly cap.
15:35 Conradish006 I tried to play MT once, but it wasn't happening.
15:36 VanessaE Warr: oh G*d.  no thanks :P
15:36 Conradish006 Even rural places here have fiber
15:36 VanessaE I'm out in the sticks and still get 200 Mbps down (12 or something up) on cable.
15:37 MTDiscord <Warr1024> The only reason HughesNet is legal in the USA is because the 8th amendment only applies to action by the government, not a private business.
15:37 Conradish006 We *have* fiber to our house, but we only have cable going inside.
15:38 VanessaE supposedly my ISP offers nearly[*] gigabit in my general region.  dunno if that's fibre, or if it's available at my address
15:38 VanessaE [*] nearly being like 950 Mbps or something
15:38 MTDiscord <Warr1024> I live in low-density-suburbia/rural, about 30 minutes from the city, and I've got fiber with at least gigabit speed options.  I have family that lives another 40 minutes away, and they recently switched from satellite to cellular and get the same crappy service, but for like $15 cheaper, and those are their only 2 options.
15:38 Conradish006 Yeah, they never actually push 1000 Mbps
15:39 VanessaE but at the price they charge for it, no thanks
15:39 MTDiscord <Warr1024> tbh I didn't even consider the cellular option at first because I assumed that no "fixed-location service" could be worse or more expensive than cellular.  I was wrong.
15:39 Conradish006 We pay $90 a month for 150 down, and we're switching to a plan that is $90 a month for gigabit.
15:39 Conradish006 Have to buy a new modem tho
15:40 VanessaE 'scuse me, "Gigabit Internet - up to 900 Mbps downloads"  for $150/mo
15:40 Conradish006 DOCSIS 3.1 modems are really expensive though
15:40 VanessaE (my present pipe is $50/mo)\
15:40 VanessaE
15:41 MTDiscord <Warr1024> I pay $75 I think and get about 130 up/down, at least on a good day (can't tell how much is constrained by my link, my router, or speed test servers)
15:41 VanessaE (had to satisfy that escaped newline ;) )
15:41 Conradish006 It's weird how we pay for 150 down but get over 200 down
15:41 VanessaE that's how it was for me for a while there, Conradish006
15:42 VanessaE I was paying for 100/3, then one day I noticed my speeds had gone up, discovered that they gave everyone an upgrade
15:42 Conradish006 huh, the ping has gone down
15:42 Conradish006 I finally replaced all the cat 5 cables with cat 6, and the ping went down
15:43 Conradish006 Went from 30-35 to 15-20
15:43 VanessaE 9ms here
15:44 VanessaE my ISP may be a little pricey but I can't fault their performance
15:44 Conradish006 same
15:44 Conradish006 They go above and beyond what is promised
15:45 MTDiscord <y5nw> At least you all use wired Internet
15:45 erlehmann joined #minetest
15:46 numzero joined #minetest
15:47 celeron55 it seem starlink is the solution to rural USA, once you can get it without begging them
15:47 celeron55 seems*
15:48 VanessaE I still have my reservations about that one
15:48 rom1504 is there any physical difficulty to installing fiber in the US or is it only political issues ?
15:48 celeron55 it's satellite, but they fly much closer so they don't have the latency
15:48 Conradish006 I just ran a speedtest, I got 230 Mbps down, 12 up. And that was about 20 feet from the router behind a closed door.
15:48 VanessaE but, just being relatively new tech and all
15:48 Conradish006 This ASUS router is amazing
15:48 celeron55 and modern so it's high bandwidth
15:48 VanessaE celeron55: exactly.  it's irrational caution on my part :)
15:51 rom1504 why is 1Gb/s fiber at 35$/month in europe and 4 time the price in US (if available at all) ?
15:51 VanessaE rom1504: why else?
15:51 VanessaE greed.
15:51 rubenwardy europe is also denser
15:51 VanessaE people like to claim it's population density but that wouldn't explain it being expensive inside a m ajor metropolis
15:51 rubenwardy although, that would only effect rural areas
15:51 rubenwardy yeah
15:51 MTDiscord <y5nw> rom1504: Until you realize that cellular costs a lot
15:52 MTDiscord <y5nw> rom1504: I pay 80€/month for 50GB of volume from Vodafone
15:52 rom1504 I pay 20€/month for 4G and unlimited
15:53 rom1504 yeah ok so basically nobody challenged the big players and the gov let it happen ?
15:54 MTDiscord <y5nw> I hope I could get by own wired Internet connection, except I can't
15:54 rom1504 I don't see any reason for starlink not to make people pay >100$/month
15:55 rom1504 apparently other companies are doing it, so why wouldn't they increase their margin
15:56 celeron55 starlink will ask a bit less to get every customer from every company to switch to them
15:56 celeron55 but probably not much less
15:56 rom1504 yeah
15:57 rom1504 in france what happened is one provider decided to do /2 on the usual price 10 years ago
15:57 rom1504 then all providers had to do the same
15:57 rom1504 if that kind of things doesn't happen, the price will stay high
15:58 MTDiscord <y5nw> Here only the mobile network from Vodafone and (afaik) Telekom work at all. I think the case is similar for wired
15:58 perrier joined #minetest
15:59 Fixer joined #minetest
15:59 MTDiscord <y5nw> Here = the town I live in right now
16:02 MTDiscord <y5nw> I don't think the telecom operators here want to reduce the price here
16:02 delta23 joined #minetest
16:11 Fractalis joined #minetest
16:20 Hawk777 joined #minetest
17:01 galex-713 joined #minetest
17:06 DS-minetest joined #minetest
17:08 ircSparky_ joined #minetest
17:22 MTDiscord joined #minetest
17:22 galex-713 joined #minetest
17:34 DS-minetest how can I build minetest with an explicitly set luajit version? I have both luajit2.0 and luajit2.1 installed and want to build with luajit2.0. I'm calling cmake with `cmake <path to minetest source> -DRUN_IN_PLACE=1 -DLUA_INCLUDE_DIR=/usr/include/luajit-2.0/ -DLUA_LIBRARY=/usr/lib/libluajit-5.1.so.2.0.5`. But when I print jit.version in minetest it still says LuaJIT 2.1.0-beta3.
17:35 galex-713 joined #minetest
17:36 DS-minetest oh, and cmake says: -- Found LuaJIT: /usr/lib/libluajit-5.1.so.2.0.5 (found version "")
17:36 DS-minetest -- Using LuaJIT provided by system.
17:37 sfan5 are you sure it's loading the right luajit at runtime?
17:37 sfan5 check with ldd
17:37 sfan5 also: you can see the lua version with ./bin/minetest --version, no need to start an in-game session
17:38 DS-minetest ldd says libluajit-5.1.so.2 => /usr/lib/libluajit-5.1.so.2
17:39 DS-minetest minetest_dev --version says Using LuaJIT 2.0.5
17:41 craigevil joined #minetest
17:44 TechDude joined #minetest
17:45 DS-minetest hm, there might be something wrong with my luajit installations
17:45 galex-713_ joined #minetest
17:50 sfan5 libluajit-5.1.so.2 will pick whatever libluajit-5.1.so.2.* it finds first
17:50 DS-minetest yes, I see that now too
17:50 DS-minetest how does that work?
17:50 galex-713 joined #minetest
17:51 sfan5 https://en.wikipedia.org/wiki/Soname
17:51 DS-minetest I didn't know such symlinks exist
17:51 FeXoR joined #minetest
17:52 DS-minetest ok, but then why does minetest pick libluajit-5.1.so.2 and not libluajit-5.1.so.2.0.5 ?
17:53 sfan5 % objdump -p /usr/lib/libluajit-5.1.so.2.0.5 | grep SONAME
17:53 sfan5 SONAME               libluajit-5.1.so.2
18:04 olliy joined #minetest
18:06 DS-minetest ok, I've made a new libluajit-5.1.so.2.3.0 that is a link to libluajit-5.1.so.2.0.5
18:06 DS-minetest ugly, but works
18:07 DS-minetest though I get a src/script/cpp_api/s_base.cpp:8c: static int ScriptApiBase::luaPanic(lua_State*): A fatal error occurred: LUA PANIC: unprotected error in call to Lua API (not enough memory) when using much memory in lua
18:07 sfan5 built luajit with gc64?
18:08 DS-minetest idk, tbh
18:09 DS-minetest the 2.0 luajit is the one from the arch repos and the 2.1 is from the AUR with some modifications
18:11 DS-minetest :/ now I can't use luajit2.1 anymore
18:13 sfan5 "some modifications"?
18:14 DS-minetest no modifications in the build settings, just some lines to avoid conflicts between both versions being installed simultaniously
18:14 sfan5 I could just have said this: if you do not remember explicitly enabling gc64 then it's not enabled
18:15 sfan5 replacing lj205 with lj210 system-wide is unlikely to break anything btw
18:15 DS-minetest yeah, but I'd like to have both for testing
18:17 DS-minetest thanks for your help! TIL something new
18:37 proller joined #minetest
18:48 MTDiscord <y5nw> Apparently LuaJIT does have ARM64 support, but not for MacOS
18:48 MTDiscord <y5nw> (Or is it specific to Debian?)
18:54 eugenefil joined #minetest
19:05 proller joined #minetest
19:08 majochup joined #minetest
19:09 galex-713_ joined #minetest
19:12 sagax joined #minetest
19:16 galex-713 joined #minetest
19:22 galex-713 joined #minetest
19:30 galex-713 joined #minetest
19:38 homthack joined #minetest
19:47 ryzokuken joined #minetest
19:49 Unit193 joined #minetest
19:57 galex-713_ joined #minetest
20:04 TechDude joined #minetest
20:04 Talkless joined #minetest
20:11 galex-713 joined #minetest
20:19 galex-713 joined #minetest
20:25 sfan5 @y5nw no, not specific to debian; luajit works fine on Linux/ARM64
20:30 galex-713 joined #minetest
20:31 duckie joined #minetest
20:35 galex-713 joined #minetest
20:46 eugenefil hi, can anyone please help, trying to build minetest from master w/ forked irrlicht, getting "libGL error: failed to create drawable" and then "X Error of failed request:  GLXBadContext"
20:46 eugenefil full logs at https://pastebin.com/i9rbuNqy
20:46 eugenefil minetest 5.4.0 from distro package works fine
20:46 eugenefil am I doing smth wrong? (I'm building minetest for the first time)
20:47 sfan5 when you run ldd ./bin/minetest, is the path Irrlicht is loaded from correct?
20:48 eugenefil libIrrlicht.so.1.9 => /home/fil/w/mt/irrlicht/lib/Linux/libIrrlicht.so.1.9
20:48 eugenefil seems so
20:52 sfan5 if you happen to have the system package around, does it also link to /usr/lib/libOpenGL.so? or a different one (perhaps named glvnd)?
20:53 Krock eugenefil: are you sure that's specific to new builds, and old ones still work? That sometimes break when it's doing driver updates behind the scenes
20:56 eugenefil sfan5: they both link to libGL.so.1 => /usr/lib/libGL.so.1 (0x00007fc73e63d000)
20:56 sfan5 why does cmake say it found /usr/lib/libOpenGL.so then? 🤔
20:57 eugenefil Krock: i'm not sure, building minetest for the first time ever
20:59 eugenefil sfan5: in CMakeCache.txt i see lines both for libGL.so and libOpenGL.so if that helps
20:59 eugenefil OPENGL_gl_LIBRARY:FILEPATH=/usr/lib/libGL.so
20:59 eugenefil OPENGL_opengl_LIBRARY:FILEPATH=/usr/lib/libOpenGL.so
21:00 eugenefil but libOpenGL.so is not in ldd output
21:00 sfan5 that's not a problem
21:03 sfan5 this is a shot in the dark but please try reconfiguring and recompiling irrlicht with -DOPENGL_GL_PREFERENCE=LEGACY
21:03 Krock ^ that should be default I think
21:04 sfan5 I put it in Irrlicht's cmake file too but I think the command might be incorrect
21:04 sfan5 because it doesn't show in my CMakeCache.txt
21:04 proller joined #minetest
21:05 eugenefil in my it shows OPENGL_GL_PREFERENCE:STRING=LEGACY
21:05 Krock had another issue with GLVND a while back, different rig. https://github.com/minetest/minetest/pull/7655#issuecomment-414150525
21:06 Krock .. but there Irrlicht failed to initialize, and that's after the Xorg stuff IIRC
21:07 Krock eugenefil: does glxgears work?
21:08 eugenefil yes
21:08 Krock hmm
21:09 eugenefil sfan5: should I rebuild then if OPENGL_GL_PREFERENCE:STRING=LEGACY already in my CMakeCache.txt?
21:09 sfan5 you're looking at irrlicht's CMakeCache.txt right?
21:11 eugenefil oh, please sorry
21:11 eugenefil was looking at minetest's
21:11 sfan5 don't worry it's easy to get it mixed up
21:14 eugenefil started build
21:14 eugenefil -- Found OpenGL: /usr/lib/libOpenGL.so
21:17 Fractalis joined #minetest
21:26 eugenefil sfan5: rebuilt, but no luck, getting same error :(
21:26 sfan5 aw
21:26 eugenefil btw, is 'make clean && rm CMakeCache.txt' enough to clean up before reconfiguring and recompiling?
21:26 sfan5 yes
21:26 eugenefil ok
21:26 sfan5 there must be something wrong with how irrlicht 1.9 creates the glx context
21:27 eugenefil maybe i should try to build say 5.4, maybe it's smth wrong w/ my setup
21:32 eugenefil started building 5.4.0, see if it'll throw the same error
21:39 sfan5 hm
21:39 sfan5 could you pastebin the full glxinfo somewhere?
21:40 eugenefil nope, just built 5.4.0, runs just fine
21:40 eugenefil ok
21:41 galex-713 joined #minetest
21:42 eugenefil here it is https://pastebin.com/tUNSHk57
21:43 sfan5 thanks
21:43 sfan5 unfortunately you have the same GLX version as me so this doesn't really help me understand the bug
21:44 eugenefil got it, anything else i can try?
21:46 sfan5 if you run minetest master with --trace, does it output anything that looks interesting?
21:48 eugenefil no, nothing special
21:49 eugenefil irrlich seems to start fine
21:49 eugenefil 2021-03-11 23:45:28: INFO[Main]: Irrlicht: Irrlicht Engine version 1.9.0mt0
21:49 eugenefil 2021-03-11 23:45:28: INFO[Main]: Irrlicht: Linux 5.11.2-arch1-1 #1 SMP PREEMPT Fri, 26 Feb 2021 18:26:41 +0000 x86_64
21:49 ShadowBot https://github.com/minetest/minetest/issues/1 -- GlowStone code by anonymousAwesome
21:50 eugenefil and after that goes the error
21:50 eugenefil libGL error: failed to create drawable
21:50 sfan5 :/
21:53 eugenefil yeah
22:01 sfan5 one more idea: revert c4503eaf17664bdf501ee96ba86d3a24636b83bc inside irrlicht, rebuild and see if it fails with a different error
22:02 sfan5 it won't make it work but it'll give me something I can look for
22:05 eugenefil ok, i'll do that
22:09 eugenefil also i found there is a minetest-git package for my distro (arch linux) maintained by y5nw that builds from master but somewhat differently, trying that at the moment
22:23 submariner joined #minetest
22:25 Verticen joined #minetest
22:28 eugenefil sfan5: btw, should i rebuild w/ -DOPENGL_GL_PREFERENCE=LEGACY?
22:28 sfan5 makes no difference
22:28 eugenefil ok
22:42 VanessaE is there no way to change the border color of a textarea[] element?
22:43 VanessaE cheapie just discovered a bug where using style_type[] to turn the border off causes the scroll bar (if visible) to stop working altogether
22:47 cheapie I'm on 75eb28b95 if that matters.
22:54 eugenefil sfan5: so I reverted and rebuilt and got another error
22:54 eugenefil X Error of failed request:  BadWindow (invalid Window parameter)
22:55 eugenefil Major opcode of failed request:  18 (X_ChangeProperty)
22:55 sfan5 that's good
22:55 eugenefil really? :)))
22:56 eugenefil great news
22:56 sfan5 https://github.com/minetest/minetest/blob/master/src/client/renderingengine.cpp#L227
22:56 sfan5 add "return false" here, recompile and the game will run
22:56 sfan5 "return false;" actually
22:56 eugenefil sounds like magic, ok
22:58 eugenefil should I keep that commit reverted in irrlicht?
22:59 sfan5 for now yes
22:59 eugenefil ok
22:59 [Pokey] Just a general modding question, is it normal to provide two versions of a mod if the mod requires platform specific libraries?
23:01 sfan5 if that's necessary, sure
23:01 [Pokey] Ah okay, thanks
23:01 eugenefil yep, the game really ran
23:03 eugenefil so what was the culprit?
23:03 sfan5 that change in irrlicht even though it apparently works fine for everyone else violated some assumption
23:04 eugenefil is the problem more w/ irrlicht fork then or w/ my setup?
23:05 sfan5 former
23:06 Elouin3 joined #minetest
23:07 eugenefil got it, will there follow some github issue/fix?
23:08 sfan5 if I figure out why it's broken, sure
23:08 eugenefil ok, i'm just curios
23:09 eugenefil so i keep those changes for now
23:09 eugenefil anyway, sfan5, thank you for you work and time
23:15 craigevil joined #minetest

| Channels | #minetest index | Today | | Google Search | Plaintext