Minetest logo

IRC log for #minetest-dev, 2016-04-22

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

All times shown according to UTC.

Time Nick Message
00:03 twoelk joined #minetest-dev
00:05 jin_xi joined #minetest-dev
00:41 ShadowNinja !tell est31 Oh, that's because I'm using ${BUILD_TYPE^} to capitalize the first character.  There's probably a shell-generic way to do that, but it'll be longer...
00:41 ShadowBot ShadowNinja: O.K.
00:54 twoelk|2 joined #minetest-dev
01:36 est31 joined #minetest-dev
01:37 est31 I'm testing the build now
01:37 est31 ShadowNinja, have you tested util/bump_version.sh ?
01:37 est31 otherwise I do it after the build finished
01:37 ShadowNinja est31: Nope.
01:39 est31 lol the curl download server now offers whopping 4,38KB/s bandwidth
01:40 est31 either it is doing some sort of backup or something, or they meter me because I downloaded their tarball the tenth time this day lol
01:45 est31 Thanks Warr1024 for testing
01:58 est31 Warr1024, can you check out the first failing commit 45db054 then attempt to build it, and after that failed, do "make openssl"? Can you put the output of that second make command to a pastebin or sth?
01:58 est31 thanks
02:00 est31 hrmmmm interesting
02:00 est31 now I get the openssl bug myself too
02:00 est31 that's good
02:00 est31 or bad dunno
02:04 ShadowNinja est31: https://github.com/minetest/minetes​t/pull/2966#issuecomment-213209485
02:05 ShadowNinja est31: The "text segment is not shareable" thing?
02:06 est31 ?
02:06 est31 dunno
02:06 est31 its just curl failing
02:06 ShadowNinja Oh, that thing.
02:06 est31 bc of openssl though
02:06 est31 no idea why that happens
02:06 ShadowNinja I had that at one point, but I don't remember what I did to fix it.
02:07 est31 I removed deps/openssl_timestamp and right now it seems to build fine
02:07 est31 but openssl is still not done ye
02:07 est31 t
02:07 ShadowNinja Might have just been an issue where a wasn't cleaning up enough things after switching to the x86 target.
02:08 ShadowNinja (x86 runs about 10x as fast in an emulator, so that target's much more useful to me)
02:08 est31 i see
02:08 est31 yeah i test it on a real device its arm
02:10 est31 ShadowNinja, next sh error
02:10 est31 /bin/sh: 12: Syntax error: redirection unexpected
02:10 est31 Makefile:755: recipe for target 'apk' failed
02:11 est31 i guess its the <<< it doesnt like
02:12 est31 trying the old fasioned way
02:12 est31 ShadowNinja, this seems to work for me:  BUILD_TYPE_C=$$(echo "$${BUILD_TYPE}" | sed 's/./\U&/') &&
02:13 ShadowNinja est31: K. Will fix.
02:19 ShadowNinja I hate this alignment in the Makefile.  I can't just use find and replace, because then I have to go back over everything fixing alignment.  And it uses a tabwidth of 4, so I can't just go by sight I have to actually count spaces.
02:22 est31 why do find and replace the first place :)
02:26 est31 aww
02:26 est31 bad
02:28 est31 you know I have an adb inside my $PATH that's from ubuntu
02:29 est31 and my smartphone has a weird reboot bug when I connect it to a pc via USB
02:29 est31 so I prefer to use adb over network
02:29 est31 so I have tried to connect the phone to adb via my $PATH adb
02:29 est31 then I did make install_debug
02:30 est31 but that failed bc it took the sdk provided adb
02:30 est31 and its of course a newer version
02:30 est31 well shrug there is nothing you do wrong here ShadowNinja
02:30 est31 its just my broked setup :p
02:35 est31 okay the version update script still works
02:52 sofar joined #minetest-dev
03:03 Warr1024 huh, seems I can't build mt android because the curl download is stalling every time.
03:06 Warr1024 what's the deal with mapgen v7?  It looks pretty awesome, but v6 is still the default...?
03:11 DI3HARD139 joined #minetest-dev
03:30 ssieb joined #minetest-dev
03:49 Player_2 joined #minetest-dev
03:59 sofar joined #minetest-dev
04:23 sofar joined #minetest-dev
04:31 sofar joined #minetest-dev
04:43 yang2003 joined #minetest-dev
04:44 yang2003 joined #minetest-dev
04:59 crazyR joined #minetest-dev
05:03 Hunterz joined #minetest-dev
05:06 paramat joined #minetest-dev
05:09 * paramat starts the euro shift
05:10 * sofar fights server upgrade gone wrong
05:16 paramat Warr1024 we considered making mgv7 the default but c55 thinks it unsuitable. no problem as i'm still working on the biome system
05:23 VanessaE paramat: until the shadow bugs are fixed, it's not suitable.
05:24 paramat the y = 63 bug is fixed
05:24 VanessaE I mean the moretrees-triggered bug.
05:25 Player_2 joined #minetest-dev
05:26 paramat ah
05:26 paramat that's present in mgv6 too though, it's a bug in l-system
05:27 VanessaE a2
05:27 VanessaE oops.
05:27 VanessaE um, no...
05:28 VanessaE I haven't seen those big black shadows in mgv6 in like 2 years.
05:28 paramat it's more subtle yes, but there are subtle shadows in water
05:28 VanessaE true.
05:29 VanessaE I'm only concerned with the black shadows, as they make a world unplayable.
05:29 paramat although they're subtle they're still unnacceptable, so that makes mgv6 as broken as mgv7
05:29 paramat :)
05:29 VanessaE well then both should be fixed before relesae.
05:29 VanessaE release*
05:30 VanessaE (or rather, the root cause thereog)
05:30 VanessaE thereof*
05:30 paramat i hope so
05:30 paramat i ownder if those subtle shadows in water were present when l-system was merged, maybe they were missed
05:31 VanessaE I don't remember one way or the other
05:34 paramat perhaps i'll try tackling this again during freeze
05:34 VanessaE well..
05:34 VanessaE I dunno, at this point I'd be inclined to continue to hold the freeze
05:35 VanessaE fix up as many *major* bugs as possible, then start the freeze, and work on minor bugs and bugs-caused-by-fixes.
05:36 VanessaE either that or start the freeze as scheduled and stay in that state until there's nothing left to fix (that's worth the effort)
05:36 paramat eheh
05:36 paramat release in 2019 then
05:36 VanessaE HAH!
05:37 Zeno` joined #minetest-dev
05:37 VanessaE maybe HE can do it! :)
05:37 * VanessaE points at Zeno`
05:38 Zeno` I am not the Zeno you're looking for
05:38 * Zeno` wiggles his fingers strangely
05:38 VanessaE you ARE the Zeno we're looking for.
05:38 VanessaE (and no, I'm not a toydarian)
05:38 paramat nah, it's time to release, in fact i'll probably reduce freeze to 1 week
05:38 Zeno` what have I done?
05:38 VanessaE paramat: no, that's a bad idea.
05:39 VanessaE Zeno`: nothing bad.  yet. :D
05:39 paramat well it can be discussed, 1 or 2 weeks
05:39 VanessaE at least 2.
05:39 jin_xi joined #minetest-dev
05:39 VanessaE I'm trying to account for how much time it takes to get things done here since people have less time these days.
05:43 sofar joined #minetest-dev
05:44 Zeno` 1 week?
05:45 VanessaE 1 week is nowhere near enough.
05:47 Zeno` hmmmm, have you re-reviewed #3948
05:47 ShadowBot https://github.com/minetest/minetest/issues/3948 -- Escape more strings: formspecs, item descriptions, infotexts... by Ekdohibs
05:47 Zeno` I don't think 1 week is enough either
05:47 hmmmm guys
05:47 VanessaE aw shit
05:47 VanessaE RUN
05:47 hmmmm i don't think our development model is a total mess
05:48 Zeno` I didn't say it was. Just that 2 weeks seems more reasonable for a feature freeze
05:48 VanessaE hmmmm: yeah... but?
05:48 hmmmm i'm referring to my own previous statements
05:48 VanessaE oh
05:48 Zeno` oh
05:48 * Zeno` glares at VanessaE
05:49 Zeno` :)
05:49 hmmmm i think this is the state of most software projects
05:49 * VanessaE glares harder at Zeno`  :D
05:49 hmmmm only a handful of 'elite' ones with all the right resources and circumstances have the solid software lifecycle that we want
05:49 * Zeno` has no idea what's being discussed... going to read logs
05:50 Krock joined #minetest-dev
05:50 hmmmm i'm not saying that having a feature freeze is a good thing, we should certainly always strive to improve the process
05:50 hmmmm but it's not horrifyingly bad and it seems to mostly work fine
05:52 VanessaE I wouldn't say our model is "bad".
05:52 VanessaE just that we need to allow a LOT more time for things like a freeze or whatever else.
05:58 paramat 2 weeks is fine by me
05:59 hmmmm https://github.com/minetest/minetest/commi​t/c350cfb50b3bf81c075a80307adb15393ea773d0  what the hell
05:59 hmmmm this does nothing to avoid the race condition
05:59 hmmmm it just makes it not as easily fatal
06:02 Zeno` the race condition is still there (the messages are still corrupted -- I brought this up after it was committed. Decided that it was better than crashing though... unfortunately the commit message is not accurate)
06:03 hmmmm i don't object to a lock here
06:03 hmmmm there is a huge misconception that locks are slow
06:03 Zeno` you don't think it will affect performance?
06:04 hmmmm that's false, a lock in contention is slow.  an uncontended lock takes around ~60ns to acquire on a modern processor
06:04 Zeno` that's what I'm asking :)
06:04 sofar joined #minetest-dev
06:04 Zeno` ok so it shouldn't affect performance
06:04 hmmmm and besides
06:04 hmmmm on windows, Mutex uses critical sections, right?
06:04 hmmmm and on everything else, pthread_mutex_lock
06:05 hmmmm those are both implemented with futexes
06:05 hmmmm they're all user mode
06:06 Zeno` a related question: why are two threads downloading the public server list?
06:07 Zeno` I started looking at that last night intending to continue the investigation today
06:07 hmmmm because sapier
06:07 hmmmm :)
06:07 Zeno` (and there are certainly two thread executing the exact same code)
06:08 paramat sofar possible issue http://irc.minetest.ru/min​etest/2016-04-21#i_4589556
06:09 Zeno` oh they're Sapier Threads
06:09 Zeno` didn't realise that was the concept :D
06:10 Zeno` std::vector<ServerListSpec> getOnline() in serverlist.cpp is called twice with the exact same paramaters in two different threads doing exactly the same thing heh
06:11 Zeno` it *seems* that there are two calls in the built-in Lua scripts that both get the server list
06:12 Zeno` (either that or mainmenu/init.lua is being executed twice)
06:12 Zeno` Thread 1: /home/crobbins/minetest/bin/../builtin/init.lua
06:13 Zeno` Thread 2: /home/crobbins/minetest/bin/../builtin/init.lua
06:20 VanessaE paramat: so getting back on topic...  feature freeze whenever, hold for 2 weeks, then release?  with lighting bugs being among the higher priorities? :)
06:21 paramat yes, freeze starts this weekend, hopefully
06:21 paramat sunday
06:22 paramat but the l-system lighting bug involves updateLighting which is very complex, i'm not confident
06:22 VanessaE it has to be fixed before mgv7 becomes the default, and honestly. 0.4.14 is a good place to make that change.
06:23 VanessaE s/y\./y,/
06:23 paramat updateLighting has issues elsewhere too, it's a big issue
06:23 Zeno` mgv7 is becoming default?
06:23 VanessaE no
06:23 paramat well mgv7 probably won't become default
06:23 VanessaE but it should be.
06:23 VanessaE it's a good mapgen
06:23 Zeno` i see. yeah
06:24 paramat c55 is not keen
06:24 Zeno` maybe in 0.5 then :)
06:25 jin_xi geological processes are known to take a while
06:25 VanessaE heh
06:25 paramat we discussed that possibly a new mapgen is needed as the default, that is smaller in scale
06:25 paramat and per-mapgen biome size is needed
06:26 yang2003 joined #minetest-dev
06:28 paramat i see mgv7 as an adventure, but some feel it's intimidatingly large scale
06:28 VanessaE nah
06:28 VanessaE it's perfect, in that regard
06:28 paramat damn minecrafters and their expectations
06:28 VanessaE when biomes are too small, they become too repetitive.
06:28 paramat yes
06:29 paramat i've noticed many players will freak out if there isn't a tree at spawn heh
06:29 Zeno` they will?
06:29 VanessaE heh
06:29 VanessaE so let 'em freak out
06:29 nrzkt joined #minetest-dev
06:29 VanessaE it's not like that tree would remain there in the long term anyway
06:29 paramat panicking because they assume mobs at nightfall
06:30 Zeno` it'd remain there for exactly 30 seconds
06:30 Zeno` kind of like Easter Island ("Yay! I cut down the last tree".. twiddles thumbs)
06:30 paramat well, some players, used to MC, it's funny to watch
06:31 VanessaE well this is MT, not MC.
06:31 VanessaE who cares what MC players are used to
06:31 paramat i agree
06:32 paramat still, i'm still working on the biome system so a delay is ok
06:32 Calinou there could be a mod to spawn a tree next to a player :P
06:32 paramat 'panic tree for MCers to cuddle'
06:33 VanessaE bahahahaha
06:54 Zeno` chop 'em down
06:56 sofar joined #minetest-dev
07:01 sofar joined #minetest-dev
07:19 nore #3948 updated, everything should be ok now
07:19 ShadowBot https://github.com/minetest/minetest/issues/3948 -- Escape more strings: formspecs, item descriptions, infotexts... by Ekdohibs
07:25 celeron55_ maybe the spawn code could be made to serach for a place that has trees
07:25 celeron55_ or the biome code to put a biome with trees at spawn
07:26 celeron55_ search*
07:30 jin_xi from a gameplay pov i would like some control about where to put biomes or a way to put exactly one foo into biome bar
07:32 Zeno` nore, I didn't end up working why the server list download is happening in two threads (which are identical) but I'm going to look again tonight
07:32 Zeno` working out why*
07:34 sofar joined #minetest-dev
07:35 nore Zeno`: ok, good luck :)
07:35 DI3HARD139 joined #minetest-dev
07:35 Zeno` umm thanks
07:35 Zeno` heh
07:36 Zeno` I don't find it easy debugging when things are jumping to/from Lua in different threads.
07:36 Zeno` But at least it's fun (sort of)
07:36 nore Oh, btw, I noticed a bit too late that my commit message implicitely said the race condition was fixed, when it wasn't, I guess there is no way to change it? :/
07:36 Zeno` There is not
07:37 Zeno` well, not a good way anyway
07:37 Zeno` (i.e. it's too late to change history now... and I think commit messages are included in that rule)
07:37 nore I think they are too
07:41 Zeno` Krock, is #4016 a WIP? It seems unclear because of the comment "Testing of conflicts with the texture modifiers is in progress."
07:41 ShadowBot https://github.com/minetest/minetest/issues/4016 -- tile.cpp: Automatically resize overlaying texture to base dimensions by SmallJoker
07:42 Krock Zeno`, no just [Testing]
07:42 Zeno` ok
07:44 Zeno` so it only resizes textures that are square?
07:47 paramat players can try a few seeds to find one with trees at spawn if they need that
07:49 paramat in most cases having trees nearby spawn is not essential, and it's a good excuse to explore
07:52 Krock Zeno`, yep
08:42 DFeniks joined #minetest-dev
08:58 paramat left #minetest-dev
09:12 Amaz joined #minetest-dev
09:20 nrzkt joined #minetest-dev
10:37 exoplanet hi is it possible to hide blocks so that they aren't displayed in craftguides? If so how?
10:50 Krock not_in_inventory = 1
10:50 Krock add that to the node def group
10:50 Krock ^ exoplanet
10:54 celeron55_ maybe, just maybe you could be more accurate so that the help could actually help someone
10:54 celeron55_ it's not_in_creative_inventory; not not_in_inventory
10:55 exoplanet thank Krock , thanks celeron55_  for the correction.
10:55 Krock oopsie
10:56 exoplanet btw: where can i find updates about minetest_game freeze? Or do the engine and the game share all the freeze dates?
11:14 iangp joined #minetest-dev
11:17 nore exoplanet: I think they share the freeze dates
11:28 Zeno` ugh
11:28 Zeno` this is impossible to track donw
11:40 Zeno` anybody wish to test (and probably add a lock): http://dpaste.com/1C3V6NA
11:41 Zeno` I can't exactly test (other than I know that it stops downloading the thing twice) because I cannot reach servers.minetest.net
11:41 misprint joined #minetest-dev
11:42 Zeno` (this won't be a PR... I'm just trying to figure things out before continuing further)
11:43 proller joined #minetest-dev
11:43 Zeno` note to self: check editor's whitespace settings
11:44 Zeno` I need to make sure that the public server list is actually downloaded with those changes (it should be) but can't test because of my connection issues
11:46 nore Zeno`: but if there are two threads downloading, one will get the online serverlist while the other will immediately get the local one, right?
11:46 nore is it what we want?
11:48 celeron55_ is that system thread safe to begin with?
11:48 Zeno` nore, I'm not sure which is why I want to see what happens with those changes
11:48 Zeno` celeron55_, it is not
11:48 Zeno` and I can't work out why two threads are doing the same task
11:49 Zeno` I've tried navigating the builtin/ lua maze heh
11:50 Zeno` the other option is to add a lock where the race condition happens
11:50 Zeno` but I'd really like to know why two threads are doing the same thing in the first place
11:54 damiel joined #minetest-dev
11:54 Zeno` nore, to be honest it shouldn't be an && condition, but as I said I'm just trying to work out WTF is actually going on
11:55 Zeno` it's not "real" code
11:55 Zeno` it's what I call research code lol
11:58 nore Zeno`: I've got the answer
11:59 Zeno` and the answer is...? :)
11:59 nore the multiplayer tab gets two "on_change" events
11:59 nore which means asyncOnlineFavorites gets called twice
11:59 Zeno` how did you track that down so fast?
11:59 Zeno` I've been looking for over an hour
12:00 nore so, I've modified it to make it not download the list if it already is
12:00 nore hm, using print? :D
12:00 * Zeno` grumbles
12:00 jin_xi yay print beats thinking every single time (in my case for sure)
12:01 Zeno` I added prints as well... they just indicated nothing useful lol
12:02 * Zeno` waits for nore to provide something to test
12:03 nore https://github.com/Ekdohibs/minetest/commi​t/1c6b6c5580e0eb98c8799f4ad2d27af7b1f2f1a7
12:03 nore Zeno`: ^
12:06 Zeno` it doesn't apply :P
12:06 nore ah
12:06 * nore rebases it
12:06 jin_xi but why does it get multiple identical events? that should be fixed also
12:06 jin_xi or is this how it works
12:09 nore Zeno`: https://github.com/Ekdohibs/minetest/commi​t/f1784d8a5de8819e0520779d4b42264880da6e69
12:09 nore jin_xi: well, on_change is a function to be called every time there is a change
12:10 nore so it is a function that can be called several times in a row without a different effect than calling it only once
12:10 nore (except that it will take more time, of course)
12:11 Zeno` that indeed stops std::vector<ServerListSpec> getOnline() being called by two different threads
12:12 Zeno` and stops the log corruption as well (obviously)
12:12 jin_xi i see, thanks for explaining, makes sense
12:13 * Zeno` is annoyed that nore tracked it down so fast
12:13 Zeno` (just kidding)
12:14 nore :D
12:14 Zeno` I just noticed that when the serverlist cannot be reached the text in the gui is now incorrect
12:14 millersman joined #minetest-dev
12:14 Zeno` "Try reenabling public server list and..."
12:15 Zeno` there is now no public server list checkbox
12:16 Zeno` tracing through C++ there everything seems as expected after your patch, nore
12:16 nore Oh, indeed
12:16 nore Hm, how should we fix that?
12:17 Zeno` fix what? The error message?
12:17 nore Yep
12:17 Hunterz joined #minetest-dev
12:17 Zeno` I suppose it needs translation which complicates things?
12:18 nore That's not the only problem
12:18 nore if serverlist didn't work
12:18 nore but then the player fixes their internet connection
12:18 nore then, they will have no way of re-downloading the serverlist
12:19 Zeno` oh I see
12:19 Zeno` hmm
12:20 nore there should be a "retry" button
12:21 nore Hm, leaving the tab and going back to it is enough
12:23 nore however, the error message can't be seen if you don't click on the "Loading..." button
12:25 nore #4018
12:25 ShadowBot https://github.com/minetest/minetest/issues/4018 -- Fix mainmenu code downloading the public serverlist twice. by Ekdohibs
12:26 nore Zeno`: ^
12:26 Zeno` looks good
12:26 Zeno` I'll leave it rest for an hour or so and merge
12:26 Zeno` doing some more tests as well (paranoia tests)
12:27 Zeno` although there is not much to test, lol
12:27 Zeno` the trace is as expected etc etc etc
12:28 Zeno` hmmm said he would have been ok with adding locks to log, but it's not necessary after this
12:29 Zeno` you know that this bug has been there for a long long time, right? :)
12:29 nore well, logging is still not thread-safe
12:29 Zeno` it just wasn't much of an issue beforehand
12:30 nore but it is doesn't crash now
12:30 nore it only produces corrupted messages
12:30 Zeno` no corrupted messages now
12:30 nore how long, exactly? :)
12:30 Zeno` since at least the last release
12:30 nore well, there will be if you try to log two things at the same time
12:30 Zeno` yes there will
12:31 Zeno` but this particular issue I meant
12:31 nore ah, yes, indeed
12:31 Zeno` I remember looking at it just before I took my break
12:31 Zeno` but then forgot because I never had "public server list" enabled heh
12:32 nore I wonder if it was what caused the "double free or corruption" crashes I got some time ago, randomly, when starting mt
12:32 nore (a race condition in logging, but not in public serverlist)
12:32 Zeno` I think it might have been
12:33 Zeno` because double free or corruption is exactly what I was getting before these changes
12:33 Zeno` not all the time but occassionally
12:33 Fixer joined #minetest-dev
12:33 nore same
12:34 Zeno` I'm somewhat scared of modifying Lua so thanks for helping :)
12:35 nore Zeno`: btw, I remember you made a huge work that abstracted Irrlicht away
12:35 Zeno` this might be the crash that has been intermittently been happening for the last two or more releases!
12:35 nore will you continue it?
12:35 Zeno` nore, which no longer builds
12:35 Zeno` it's 1 year behind in commits and I can't rebase
12:35 nore yeah, that's the usual problems :/ I have some huge prs that are in the same state
12:36 nore (meta_set_nodedef namely, which I rebased from celeron55's 1-year-old branch)
12:36 Zeno` so as I mentioned to someone else it would/will be easier to start from scratch than resolve the conflicts
12:36 nore you think it really would?
12:36 Zeno` yeah, it's impossible (or at least not reasonable) to rebase
12:36 Zeno` yeah I do... the main problem is that my changes touched 99% of the files
12:37 Zeno` it'd take me a month to rebase them all and probably less then a month to start again with a better strategy
12:37 nore I did the same thing with map generation limits
12:37 nore I wonder whether we should change positions back to a new type, v3pos
12:38 nore to make this change easier in the future
12:38 nore problem is, it affects a lot of the files
12:38 nore and that makes it impossible to maintain in a mergeable state for long
12:38 nore and it breaks all PRs :/
12:38 jin_xi lets throw cochinelle at mt
12:39 nore but larger maps have been a feature that has been asked for a long time
12:39 nore unfortunately, I think we would lose almost all compatibility (map format, protocol bump needed, etc)
12:40 Zeno` yeah I'm trying to come up with a method that will make merging easier
12:41 Zeno` i.e. something that can avoid conflicts when large amounts of files are changed
12:41 nore Zeno`: what is your opinion on changing that btw?
12:41 nore (making the maps larger)
12:41 Zeno` people really want larger maps?
12:42 nore I remember this was asked
12:42 Zeno` there *should* be some buffer at least to make it able to happen
12:42 Zeno` "buffer"
12:43 nore what do you mean, a buffer?
12:43 Zeno` i.e. not limited to s16 (or whatever it is)
12:43 nore yep, s16
12:43 Zeno` I mean abstracting away that limit
12:43 nore I did that once
12:43 Zeno` but you obviously had the same problem I did
12:44 nore problem was, if that limit was changed
12:44 nore it would break everything
12:44 nore that is: protocol, map saving format (blocks would be overwritten by others)
12:44 Zeno` I'm talking with a few other devs on how these things can be possibly be done without breaking existing PRs and being almost impossible to avoid conflicts
12:44 Zeno` but it's tricky
12:44 Zeno` hmm, yes it would
12:45 nore old clients wouldn't be able to handle bigger maps
12:45 nore so there would need to be a protocol bump
12:46 nore and server and client would have to agree on the size of positions
12:46 nore etc.
12:46 nore but that would mean, less efficient code
12:46 nore (since it would have to use larger data types)
12:47 Zeno` people really want larger maps though?
12:47 nore and, the client and server wouldn't be able to know at compile time the sizes of positions
12:47 Zeno` They must be insane
12:47 nore well, all sort of problems
12:47 Zeno` although some older servers have huge maps...
12:47 Zeno` I know on my server me and a few other players mined to the bottom of the world
12:48 nore yep, some servers are really *huge*
12:48 Zeno` and my server wasn't even that big heh
12:48 Zeno` (well it was for a while, but I mean explored terrain)
12:48 nore and since players often want space around them, surface space can run out
12:48 Zeno` yeah I know when I'm playing that I hate neighbours
12:49 nore see :) ?
12:49 proller joined #minetest-dev
12:54 paramat joined #minetest-dev
12:55 nore paramat: see my comment on game#1048
12:55 ShadowBot https://github.com/minetest​/minetest_game/issues/1048 -- Flowers: Fix waterlily node's duplication with stack of 1 by kilbith
12:56 paramat ok
12:57 nore according to me, the correct solution is to add node_placement_prediction = "" to the definition of the lily pad
12:57 nore but tell me your opinion about it
12:57 paramat hm. i tried changing wield to a different item straight after placing, no problems, however i am no expert on this stuff
12:57 paramat your idea sounds better
12:58 nore paramat: yes, now, changing wield causes no problem
12:58 nore the problem is if you change the lilypad to another item in inventory
12:58 paramat ah
12:58 nore (which could happen if the server is laggy)
12:58 paramat would your proposal stop the waterlily appearing underwater briefly? that looks sloppy
12:59 nore or, if the minetest.after doesn't run because the server is shut down
12:59 nore yes, it would stop that problem as well
12:59 nore could you try it and give me your opinion?
12:59 paramat i support your proposal then
12:59 paramat ok
12:59 paramat btw forget about larger worlds :}
13:00 paramat hmmmm feels the same way
13:00 paramat all my new mapgens will have multiple layers to create more space in a more interesting way
13:01 nore paramat: then, some kind of way to do that without having problems with sunlight would be good
13:02 paramat fixed that problem
13:02 paramat mapgen lighting calculation now has a bool to choose whether shadows above are propagated
13:02 Lunatrius` joined #minetest-dev
13:03 paramat now we can set that to false a few hundred nodes below an upper realm
13:05 nore ah, cool
13:05 nore but wait, only in mapgen?
13:05 Zeno` paramat, I will be merging #4018 kind of soon
13:05 ShadowBot https://github.com/minetest/minetest/issues/4018 -- Fix mainmenu code downloading the public serverlist twice. by Ekdohibs
13:05 paramat in lua mapgen too
13:05 nore or does it set a boolean in the blocks
13:06 paramat it's per-mapchunk
13:06 nore than tells them not to do that for later propagate_sunlight?
13:06 paramat it's effective only at mapgen time
13:07 paramat zeno ok that stuff is beyond me
13:07 nore paramat: hm... so there still are a few problems with it
13:07 Zeno` paramat, it's related to the race condition causing log corruption and crashes that was discussed last night
13:08 paramat yeah i read your discussions
13:09 paramat nore yes it still means the shadow is cutoff, but only in the way they already are when generating upwards
13:09 Zeno` oh people actually do that :D
13:10 Fixer I've noticed that new favourites needs additional polish :/
13:11 nore paramat: couldn't a boolean be added in the format of a stored mapblock, that tells later light propagation not to propagate shadows?
13:11 paramat ah i see, new shadows caused by builds
13:12 nore for example
13:12 nore so, in serialized mapblocks, you add one more boolean
13:12 paramat i'll investigate this in mgwatershed then, will add it soon after release
13:12 nore (that can be done in flags if we have one remaining)
13:12 Megal joined #minetest-dev
13:13 paramat (i mean, i'll investigate the potential problems)
13:15 paramat ok stuff to consider. i guess the workaraound is to not build in the space just below a realm, to avoid propagating the shadow
13:15 Fixer paramat, wait, you fixed y=63 and something new popped up?
13:16 Fixer paramat, what about that pain with vanessas trees mod shadow bug?
13:16 paramat no, this is an inevitable problem with stacked realms, nothing new
13:16 paramat the moretrees/mgv7 bug is a bug in l-system generation, not fixed yet
13:17 Fixer paramat, it is not in milestones, right?
13:20 nore paramat: just checked, in serialiazed mapblocks, only 4 bits out of 8 for flags are used
13:20 nore so you could add that flag in it
13:20 paramat interesting
13:21 paramat Fixer not in milestones
13:21 nore and when asking the mapgen not to propagate sunlight from above, just set this flag for all top blocks in the chunk
13:21 paramat very complex code, would be nice to fix during freeze though
13:22 nore it can wait 0.4.15
13:22 nore l-system trees should be fixed someday though
13:22 nore but same, it is complex code, that can wait
13:22 paramat yes, hopefully in 0.4.14dev
13:23 paramat updateLighting has other issues too
13:23 Fixer water reactivation *cough*
13:23 paramat that too
13:24 Fixer so much stuff to fix
13:25 paramat so i'll merge games 1029 and 1048 in 2 hours
13:25 paramat bbl
13:27 paramat left #minetest-dev
13:37 hmmmm joined #minetest-dev
13:53 proller joined #minetest-dev
14:01 davisonio joined #minetest-dev
14:05 asl97 joined #minetest-dev
14:26 iangp joined #minetest-dev
14:31 kaadmy joined #minetest-dev
14:35 paramat joined #minetest-dev
14:35 proller joined #minetest-dev
14:37 paramat nore is protection check needed in game#1048 ? i thought these checks already run and don't need to be coded
14:37 ShadowBot https://github.com/minetest​/minetest_game/issues/1048 -- Flowers: Fix waterlily node's duplication with stack of 1 by kilbith
14:37 nore paramat: I'm not sure that do
14:38 kilbith joined #minetest-dev
14:39 kilbith so i checked inside a protected area, and yes, needed
14:39 paramat ok
14:39 Fixer kilbith, hii, i've looked at new features feature, should it save them after restarts?
14:39 kilbith sorry ?
14:40 Fixer mistyped
14:40 Fixer kilbith, hii, i've looked at new favourites feature, should minetest save them after restarts?
14:40 kilbith they're already saved in local, nothing changed here
14:41 Fixer kilbith, i've noticed few problems: 1) sometimes favourites are gone 2) they are sorted in order of addition (maybe sort by player count?) 3) to activate del. favs button i need to recheck the server sometimes
14:43 Hunterz joined #minetest-dev
14:44 kilbith 1. no idea, never heard of that  2. could be possible with a small cost in performance  3. inherent formspec behavior when the table[] first appears
14:46 Fixer first issue was spontaneous, i have not reproduced it unfortunately
15:09 Krock joined #minetest-dev
15:11 MillersMan joined #minetest-dev
15:29 Warr1024 it seems that particlespawners do not remove themselves automatically from the server after their duration has ended...
15:30 Warr1024 I was creating new ones to replace old ones, for a persistent effect, and over the course of an hour or so, the server CPU slowly climbed to 100%
15:30 jin_xi yes this is a know bug and there is a pull but no one cares
15:30 Warr1024 ah, okay
15:30 Warr1024 is more +1 needed for it or something?
15:31 jin_xi https://github.com/minetest/minetest/pull/2602
15:32 Warr1024 wow, it's more than a year old
15:33 Warr1024 I can work around it easily enough in lua, just seems a shame to have to add a hack into my mod for that...
15:34 betterthanyou710 joined #minetest-dev
15:37 MillersMan Hi, I'm the developer of the new canyon mapgen. paramat mentioned something about the need for another C++ coder.
15:41 paramat hello
15:42 Fixer ohi
15:45 Zeno` hi hmmmm
15:45 Zeno` hi MillersMan
15:45 Krock Zeno`, would you be so nice and review https://github.com/minetest/minetest/pull/4016 please if you have got some time?
15:45 Krock ^ paramat too
15:45 Zeno` Krock I already have
15:45 Fixer it is very small commit, but useful, i tested it
15:46 Fixer must have for people with custom texture packs
15:46 Zeno` Krock, if there is no issue with existing mods etc then I'm fine with it
15:46 Fixer of different size than 16
15:46 Krock Zeno`, oh yes. Haven't seen the tags added by you
15:46 Krock There are several mod developers around that try to find a bug in it
15:47 Zeno` Krock, cool. Well give it a while for testing. Apart from that it seems fine
15:47 Krock Zeno`, I would like to see this pull merged for the next release because it exists for a long time
15:48 Zeno` paramat, thoughts?
15:48 Zeno` Krock, yeah it's something that annoyed me when I was writing mods for my server actually
15:49 Zeno` I'll add it to the release milestone unless paramat has a strong objection
15:50 Zeno` (or somebody else)
15:50 Zeno` I think it should be in the release at least
15:50 Krock Okay thanks :)
15:51 paramat it would be nice in release, but don't consider it essential
15:51 Zeno` it's not essential no, but it is nice
15:51 paramat add a milestone if you want though
15:51 Fixer i'm playing with custom textures on fancy servers, if i notice something wrong, i will post, looked at it since yesterday, no problems yet, moreblocks shapes are also fine
15:51 Fixer item frames also fine
15:51 Fixer columns etc
15:51 Zeno` ok added. If other devs disagree they can remove it anyway
15:52 paramat milestones are also used for things simple, nice to have in release, but not essential
15:52 Zeno` yup
15:52 Fixer reason is that sofar made excellent tool to convert MC texture packs, that i used, and only annoyance is this bug
15:52 Fixer which is fixed by this PR
15:54 Zeno` the "bug" that 4016 fixes?
15:54 Fixer yes
15:55 Zeno` ok thanks :)
15:55 Zeno` the re-scaling can be avoided if the texture is not square anyway
15:55 Fixer with it texture packs are ready to use
15:55 Zeno` maybe that can be documented somehow (even though it's probably an obscure use)
15:56 Zeno` Krock ^
15:57 Krock Yeah but how?
15:57 Krock and where?
15:57 Zeno` I have nfi, that's why I'm asking you
15:57 Zeno` lol
15:57 Krock ah
15:57 Krock lua_api.txt then
15:57 Zeno` it's not something critical... just an idea
15:57 Krock where the texture overlay is described
15:57 Zeno` quite possibly yes
15:58 paramat i'm no expert on the code in 4016, best someone else reviews it
15:58 Fixer also, it is not MC converted packs, it is MT packs too (of size different than 16px)
15:59 Fixer since many servers use custom ores and other stuff, any non16px texture pack will have resize problem on them
16:11 paramat game#1050
16:11 ShadowBot https://github.com/minetest​/minetest_game/issues/1050 -- Farming: Add quiet sounds to seeds by paramat
16:14 Krock LGTM
16:17 proller joined #minetest-dev
16:29 electrodude512 joined #minetest-dev
16:29 electrodude512 joined #minetest-dev
16:40 paramat left #minetest-dev
16:46 proller joined #minetest-dev
16:53 electrodude512 joined #minetest-dev
17:06 rubenwardy joined #minetest-dev
17:12 Void7_ joined #minetest-dev
17:13 turtleman joined #minetest-dev
17:15 iangp joined #minetest-dev
17:20 Robby_ joined #minetest-dev
17:21 betterthanyou710 joined #minetest-dev
17:28 Void7_ hmm.. is setting `generate_normalmaps=true` supposed to cause a segfault in git?
17:30 jin_xi idk but think its safe to say no to that
17:37 DI3HARD139 joined #minetest-dev
17:48 Void7_ backtrace: https://gist.github.com/anonymous​/3340aca612e2bef3457fb1b0e7432bbc
17:48 sfan5 looks like a driver bug
17:50 Void7_ works fine if i comment out the generate normalmaps part in the wielded shader
18:12 lag01 joined #minetest-dev
18:16 cheapie kilbith: I noticed that the server list got changed recently, and now not only does it show *all* servers all the time (not just ones I've connected to in the past) but also uses the names for them I'm not used to.
18:16 cheapie I could /probably/ get used to this over time, but it still seems a bit silly to me to remove a working feature (the ability to have the serverlist box unchecked) for no reason.
18:17 Fixer it is unified now
18:17 Fixer it is more simpler now, I guess
18:17 kaadmy Fixer: what happens if there's a server that you've joined that is currently down and you still want to see it?
18:17 cheapie Fixer: Yes, it is "unified" now, and that's exactly the problem.
18:18 Fixer kaadmy, don't understand
18:18 Fixer cheapie, but why names are different?
18:18 Fixer cheapie, it should mark it with star and go up in list and not repeat below
18:18 kaadmy ie. you have previously joined a server, and want to see it in the list, for example you want to check the ip address
18:19 kaadmy but if that server is down, will it still be shown in the list?
18:19 Fixer i think yes
18:19 kilbith cheapie, yes it's meant to be unified, many people like that
18:19 cheapie Fixer: I'm used to seeing things like "daconcepts.com:30000", not "VanessaE's Dreambuilder creative server"
18:19 kilbith kaadmy, yes
18:19 cheapie It's a lot easier to find the server I want by looking for, say, the ones starting with "daconcepts.com:" then for the port number I want.
18:20 Fixer cheapie, you can click once on it to see its ip address
18:20 cheapie Reading the whole "name" of the server takes a lot longer.
18:20 cheapie (or clicking on each one to see the address/port)
18:21 kaadmy maybe an option to enable showing the full ip address instead of names?
18:21 kaadmy doesn't exist yet afaik
18:22 cheapie I suppose I can understand showing "favorites" at the top when the server list is on, but IMO, the old favorites-only mode without the "friendly" names should still be there.
18:22 cheapie (even if it's buried in minetest.conf somewhere)
18:22 cheapie Setting the serverlist URL to be blank comes close to that, at least.
18:24 electrodude512 joined #minetest-dev
18:26 ssieb joined #minetest-dev
18:29 cheapie [13:28:44] <VanessaE> oh f--- that.
18:29 cheapie [13:29:06] <VanessaE> there should be a toggle for that, at most
18:32 kilbith well you were used with the old list for years, that's a matter of habit
18:34 kilbith i'm known for designing the fanciest formspecs around and i can ensure the old list was less user-friendly
18:39 * VanessaE git pulls...
18:39 VanessaE I haven't seen this rewritten menu yet.
18:41 VanessaE doesn't look too much different.
18:41 VanessaE but I agree with others, no "favorites" filter is bad.
18:42 VanessaE this is like always having to dig through your phone book (dead tree edition) to find entries that you normally keep in a personal directory ("little black book" or so).
18:43 VanessaE and it upsets the sorting that is provided by the master server list.
18:44 VanessaE (since it's putting my favorites ahead of the rest of the list)
18:45 VanessaE and in my case, entries seem to be randomly displayed as address:port for some, and descriptive name for others.
18:45 VanessaE (that's been the case in the "favorites" menu for a long time, though)
18:45 kilbith you can trust me it will ease your users to find their paths better
18:45 kilbith you can't tell me that a list of IP addresses is user-friendly
18:46 kilbith not even assorted with icon details or connected players
18:46 ElectronLibre joined #minetest-dev
18:46 VanessaE hence the suggestion that there be a toggle to show address:port
18:47 VanessaE many people know servers *only* by address:port.
18:47 VanessaE some of mine, I remember better by their names, others by address.
18:48 kilbith a minority of geeks that are accustomed to play on your servers only
18:48 kilbith not the our largest public audience
18:48 kilbith -the
18:48 VanessaE what does that matter?
18:48 VanessaE who cares whose servers the majority of the audience plays on?
18:49 kilbith i mean those who plays on your servers know your different servers by IP:port
18:49 kilbith *only those
18:50 kilbith and i'm sorry to say but it's not user-friendly
18:51 VanessaE you mean it's not "kilbith friendly" :)
18:51 kilbith anyways anything can't please to everyone, but i'm rather confident given the number of positive feedbacks i received so far
18:51 VanessaE to decide what is, or isn't "user friendly" you have to ask the users.
18:55 cheapie <kilbith> anyways anything can't please to everyone
18:55 cheapie ^^ That's why settings were invented.
18:55 VanessaE in any case, you can please more users by providing options.
18:55 VanessaE ninja'd.
18:56 cheapie (and the old version *had* them)
18:56 VanessaE put the damn checkbox back, and add one to show names vs. addresses.
18:56 VanessaE then everyone's happy.
18:57 kilbith ok i might thinker a toggler to a technical list with IP:port entries only
18:57 kilbith *tinker
18:57 VanessaE or another dev can..
18:58 Warr1024 dunno about IP:port, but address-as-I-entered-it:port would be good.
18:58 Warr1024 i.e. I would probably have put a DNS name in there instead of numerical IP
18:58 kilbith the damn checkbox looked out-of-place in that tab anyways
18:58 VanessaE foo.com:port  not 12.45.67.89:port.
18:58 Warr1024 unless IP is all you have
18:58 VanessaE of course.
18:59 Warr1024 the point is, don't try doing DNS lookups to display the list :-)
18:59 VanessaE or if a server has no domain.
19:00 VanessaE bbl
19:19 VanessaE back
19:24 electrodude512 joined #minetest-dev
19:35 davisonio joined #minetest-dev
19:36 davisonio joined #minetest-dev
19:39 nrzkt joined #minetest-dev
19:40 iangp joined #minetest-dev
20:20 Hunterz joined #minetest-dev
20:26 Fixer kilbith, lost favourites after game restart again, not sure how
20:29 kilbith your favoriteservers.txt file's still there ?
20:29 Fixer yes
20:30 kilbith it is filled or not ?
20:30 Fixer looks like this: http://pastebin.com/raw/iGmnYejP
20:30 kilbith then why employing the plural on "favourites" ?
20:31 kilbith no idea of your issue, 1st time it gets reported
20:33 Fixer before restart i got 4 favourite servers, after restart - none, then I joined Xanadu
20:33 kilbith then your file got cleared for sure
20:34 Fixer could be, not sure why and how
20:35 Fixer kilbith, try connecting to different servers, reconnecting, closing game, reopening etc
20:35 electrodude512 joined #minetest-dev
20:36 kilbith i did not touched the deleting mechanism, it's done elsewhere in the engine
20:37 Fixer ok, managed to cought it
20:37 Fixer joined Liberty Land, exited it, while server listed still not loaded I've exited, no file anymore
20:37 Fixer server list*
20:38 Fixer I'm using c350cfb version of minetest
20:40 blaze joined #minetest-dev
20:42 Fixer kilbith, it is lost on game exit, probably when getting a server list or in other way
20:44 kilbith well as i said i'm the wrong person to catch on that
20:45 Fixer i will post issue on github
20:56 Calinou https://github.com/minetest/minetest/issues/4022
20:56 Calinou important issue, please comment
20:57 davisonio joined #minetest-dev
20:59 Fixer Calinou, lol, that kind of bug is hard to analyze :(
20:59 Fixer Calinou, without more data
20:59 Calinou it's still problematic
20:59 Calinou when I get to profile mods, I'll do it
20:59 Calinou IMO this is almost a blocker, it's impossible to host a lag-free server like you can in 98% of games :(
20:59 Fixer Calinou, i have same problem on ESM server, every day
20:59 Calinou something is very likely to be broken in core
21:00 Calinou somebody "improved" the netcode
21:00 Fixer Calinou, make mod profile, post debug.txt with maximum debug info
21:00 Calinou IMPROVED
21:00 Calinou [spock]
21:00 Calinou Fixer: yeah I will try, if players don't complain...
21:00 Fixer Calinou, pfff slowpoke.png
21:02 Calinou updated the issue with a bit more information
21:03 Fixer Calinou, munin monitoring mod can help too i guess
21:04 Fixer Calinou, look into debug.txt do you have lines with
21:04 Fixer 'RE-SENDING timed-out RELIABLE' when lag appears?
21:05 Fixer Calinou, i was thinking "players will notice that other players stop moving for a second or two" was part of MT experience *cough*
21:06 Calinou see PM, for a monitor (netdata)
21:06 rubenwardy joined #minetest-dev
21:08 iangp joined #minetest-dev
21:08 Calinou the server does have a 5.8 GB log file... debug.txt
21:09 Calinou I deleted it, we'll see if it lags less
21:15 iangp joined #minetest-dev
21:16 nrzkt Calinou, is there any player timeout notified to chat after that ,
21:16 Calinou player timeouts are rare
21:18 asl97 Calinou: is there a cpu spike when it happen?
21:22 asl97 afaik, minetest only use 1 core for doing the main stuff, a cpu spike might not be so obvious depending on what you use to check it
21:23 asl97 my guess is the save interval,  as fast as the dummy database is, it probably take a few seconds to save once the whole world get deleted (mark as need saving), 5-10 seconds seem a little bit much though.
21:39 asl97 ^ignore above comment, i am probably wrong about the save interval. (i read part of the issue wrongly too)
21:49 Void7_ joined #minetest-dev
21:53 proller joined #minetest-dev
22:17 everamzah joined #minetest-dev
22:29 rubenwardy Calinou, What happens if you disable the ctf_events mod? That runs every 10 seconds
22:29 rubenwardy ~math c 19*3*6
22:34 damiel joined #minetest-dev
22:43 Void7_ joined #minetest-dev
22:56 Anchakor joined #minetest-dev
23:05 jhcole joined #minetest-dev
23:06 jhcole I'm getting ready to add 765 lbm's to my server; should I be worried?
23:08 proller joined #minetest-dev
23:16 sofar jhcole: yes
23:17 sofar jhcole: that's insane. try combining them or use nodetimers instead
23:22 jhcole OK, but I'm not sure how to combine them. I multiplied 17 colored node blocks into 765 varieties using stairsplus.register_all(), but now I want to change the node names and I need to keep the orientation of each node. ATM, I've written a loop to make a lbm for each type.
23:23 jhcole I considered aliases, but I liked the idea of fixing the nodes in the map instead of masking them.
23:28 yang2003 joined #minetest-dev
23:28 jhcole After some more thought, I think I can make it 1 lbm. Thanks sofar
23:51 rubenwardy joined #minetest-dev
23:53 turtleman joined #minetest-dev

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