Minetest logo

IRC log for #minetest-dev, 2015-08-04

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

All times shown according to UTC.

Time Nick Message
00:06 paramat phew complete
00:56 paramat left #minetest-dev
01:15 kaeza joined #minetest-dev
01:54 sloantothebone joined #minetest-dev
03:11 AnxiousInfusion joined #minetest-dev
03:57 sloantothebone_ joined #minetest-dev
04:56 swaaws joined #minetest-dev
06:03 Calinou joined #minetest-dev
06:16 Hunterz joined #minetest-dev
06:28 nrzkt joined #minetest-dev
07:03 Darcidride joined #minetest-dev
07:06 Darcidride_ joined #minetest-dev
07:13 leat joined #minetest-dev
07:24 rubenwardy joined #minetest-dev
08:00 Yepoleb_ joined #minetest-dev
08:02 alket_ joined #minetest-dev
08:17 GunshipPenguin joined #minetest-dev
08:26 Amaz joined #minetest-dev
08:31 bluegreen joined #minetest-dev
08:37 Ardonel joined #minetest-dev
08:37 kaeza joined #minetest-dev
08:37 blaise joined #minetest-dev
08:37 selat joined #minetest-dev
08:37 ShadowBot joined #minetest-dev
08:37 Kray joined #minetest-dev
08:47 TBC_x man... single-threaded emerging takes forever
08:48 bluegreen joined #minetest-dev
09:01 deltib joined #minetest-dev
09:12 sfan5 joined #minetest-dev
09:43 Darcidride joined #minetest-dev
09:55 leat joined #minetest-dev
10:14 blaze joined #minetest-dev
10:15 twoelk joined #minetest-dev
10:19 lag01 joined #minetest-dev
10:22 TBC_x TSan warnings printed for server running for over 12 hours http://sprunge.us/FGFP
10:23 est31 good old connection.cpp
10:29 OldCoder joined #minetest-dev
10:33 FR^2 joined #minetest-dev
10:38 twoelk just found out that a more recent client seems to have scrambled my account password on a server. after connecting into the old account with a new client I can not connect with an old one, roughly preminimap, anymore
10:39 nrzkt twoelk: SRP related
10:41 twoelk whatever - just reporting something that might be considered a sort of compatabilty breackage and therefore might be of importance when regarding the naming of the next stable version
10:46 celeron55 is there no proper error message for that situation?
10:46 celeron55 there should be
10:47 twoelk it seems to be once you have decided to use a newer client you cannot go back and use an older because you cannot use the same nick-password set anymore
10:49 twoelk this may be of concern when you use different devices to connect to the same account at different times
10:49 VanessaE no offense to est31 but SRP needs to be completely disabled (both client and server) by default until stuff like ^^^^ that, and irc_commands compatibility works properly.
10:57 sfan5 yeah
10:58 sfan5 SRP should be disabled until all problems and issues with it are solved
11:00 nrzkt celeron55, no you only have a login failed because a password mismatch
11:00 celeron55 that's pretty bad
11:01 TBC_x anyone looked into GNU/Hurd bug? https://lists.debian.org/debian-hurd/2014/10/msg00020.html
11:02 TBC_x hmm... when I look at the code it looks fixed
11:03 VanessaE Hurd still exists? :)
11:06 RealBadAngel im about to fix all the fsaa/non tileable textures issues
11:07 RealBadAngel question: have we switched to protocol 26 already?
11:07 VanessaE a wild RBA appears!
11:07 TBC_x probably, yes
11:07 TBC_x git log says so
11:08 RealBadAngel i was waitin for it
11:08 VanessaE RealBadAngel: note that we're in feature freeze now
11:08 RealBadAngel thats a bunch of bugfixes
11:08 RealBadAngel not a feature
11:08 VanessaE I know.
11:09 VanessaE just saying you'll have to be extra careful :)
11:09 RealBadAngel i know i know, valgrind has to show no more errors after my commits ;)
11:10 VanessaE TBC_x: including some of the harder stuff like address and thread sanitizing?\ :)
11:10 TBC_x at least address sanitizing
11:10 TBC_x thread sanitizing makes the code running single-core
11:10 VanessaE otherwise hmmmm will kick your ass ;)
11:11 TBC_x and disables OpenGL
11:11 VanessaE ok so forget that part :P
11:11 TBC_x I pointed out that we should disable minimap radar mode when it is `broken`
11:12 TBC_x unless you want to fix it before release
11:12 RealBadAngel TBC_x, i will fix it
11:12 TBC_x you got six days
11:12 RealBadAngel they have deleted too much
11:12 RealBadAngel in wild hunt for leaks
11:13 VanessaE I suggesting pushing what you have to your own fork now so that others have time to review
11:13 RealBadAngel scanner is receiving too few data thx to deleting whatever they thought is not necesary
11:13 VanessaE there's really no time for your usual "I'm fixing/polishing and will have it out in some days"
11:14 RealBadAngel minimap is broken thx to sanitizing it
11:14 RealBadAngel i wont allow minimap in its current state be a part of a release
11:15 RealBadAngel there should be a rule: dont understand the code, dont touch
11:15 RealBadAngel ask first for christ sake
11:15 TBC_x that's bullshit
11:15 RealBadAngel no its not
11:16 RealBadAngel breaking a funcionality and fixing leaks are two different things
11:16 TBC_x introducing `broken` functionality is another
11:16 TBC_x and I don't like ho mapper utilizes CPU
11:16 TBC_x s/ho/how/
11:16 RealBadAngel for the leaks im sorry. for the functionality im not to blame
11:17 RealBadAngel TBC_x, so you would expect some bunch of magic dwarves doing the scans instead of cpu?
11:17 Puma_rc joined #minetest-dev
11:17 VanessaE strange, I don't see anything wrong with the minimap in HEAD.
11:18 RealBadAngel radar mode on surface is broken
11:18 TBC_x at first, mapper shouldn't generate minimap blocks for mapblocks that are not in range
11:18 RealBadAngel it should look like night vision googles effect
11:18 VanessaE yeah I see now.
11:18 RealBadAngel TBC_x, mapper doesnt genereate any blocks
11:18 TBC_x well... pixels
11:18 RealBadAngel no
11:19 VanessaE underground there is another issue - visible mapblock borders
11:19 VanessaE or rather, chunk borders I think
11:19 RealBadAngel whatever is loaded is in the cache
11:19 TBC_x yes
11:19 RealBadAngel thats not going to change
11:19 TBC_x even if it never gets to the screen at all
11:20 RealBadAngel you seems to not understand one fundamental rule
11:20 TBC_x i listen
11:20 RealBadAngel block is minimapped on load or on change only
11:20 RealBadAngel so basically once
11:21 RealBadAngel factor of calculations needed is reduced by 16
11:21 RealBadAngel its 8,5kk nodes divided by 16
11:21 RealBadAngel to get full scan
11:22 TBC_x you don't have to scan the entire universe when you're making a minimap
11:22 RealBadAngel ONLY because of that we are able to make it run realtime
11:22 RealBadAngel i dont
11:22 RealBadAngel i scan the cache selecting visible blocks
11:23 TBC_x I don't think that realtime minimap is better than fast minimap
11:23 TBC_x well... I want to see your results when you're done with your modifications
11:23 RealBadAngel realtime means it have to be fast
11:23 RealBadAngel i want to make it even faster
11:24 VanessaE RealBadAngel: you need to make your modifications available NOW.
11:24 RealBadAngel but not the way you think
11:24 VanessaE in your own repo or something
11:24 RealBadAngel no
11:24 RealBadAngel its a long term goal
11:24 VanessaE no
11:24 RealBadAngel i need to rewrite whole core
11:24 TBC_x too poor someone pushed your minimap to master
11:24 VanessaE we don't have time for "long term" here.  feature freeze --> release in 6 days.
11:24 RealBadAngel TBC_x, its not poor. it is fast already
11:25 RealBadAngel on the other hand theres no code that is perfect from the begining
11:25 TBC_x you want me to make comparisons?
11:25 RealBadAngel moreover it will never be propably
11:25 RealBadAngel i just have an idea how to make things faster
11:25 TBC_x minecraft minimaps decrease performance just by around 10%
11:26 RealBadAngel but thats have nothing to do with general algorithm for scanning
11:26 RealBadAngel TBC_x, you must be joking
11:26 RealBadAngel i was playing mc
11:26 TBC_x well, ok 15%
11:26 RealBadAngel on my box minimap caused like 50% fps drop
11:27 TBC_x on my box, your minimap caused like 75% fps drop
11:27 RealBadAngel on mine theres no visible drop :P
11:27 TBC_x too poor you're not developing on an average supported box
11:28 RealBadAngel only thing that slows down minimap is not scanning but preparation of textures to be shown
11:28 OldCoder joined #minetest-dev
11:28 TBC_x and why all textures are so slow anyway?
11:28 RealBadAngel TBC_x, im developing on a box thats worth circa 250$
11:28 RealBadAngel dare to call it high end one?
11:29 RealBadAngel my new tablet cost me way more than that :P
11:29 TBC_x I'm on a box worth of $400
11:29 TBC_x eight years ago
11:30 RealBadAngel have you found it while lookin for some burried treasures?
11:30 VanessaE RealBadAngel: I'm on a box that cost me $700 (a few years ago) and which is objectively faster than yours, yet as you well know you get better performance than me...  so value/cost != speed.
11:30 VanessaE (not in minetest anyway)
11:31 TBC_x then you should know what I had before this one
11:31 RealBadAngel 8yrs old computer is worth only throwing away
11:31 TBC_x it was Toshiba with 486, 32MB RAM and 4GB HDD
11:31 RealBadAngel its way before dinosaurs :P
11:31 VanessaE 8yrs-old computer is perfectly functional and is what minetest is supposed to be targeted at.
11:31 RealBadAngel lol
11:32 RealBadAngel sorry but most of what im coding is targeting nowadays effects
11:32 Calinou the game can be played without minimap anyway
11:32 RealBadAngel ofc
11:32 Calinou if performance is too low, just disable it
11:32 VanessaE RealBadAngel: that's why you make sure that you have adequate fallbacks
11:32 RealBadAngel just dont enable it if box is too slow
11:32 RealBadAngel fallback for minimap is to disable it
11:32 Calinou also, why not update the minimap less often?
11:33 RealBadAngel maybe there should be a setting for it in main menu
11:33 TBC_x not acceptable for other people
11:33 Calinou wouldn't that reduce resource usage by a lot?
11:33 VanessaE I mean in general.  there's nothing even remotely "nowadays" about a minimap
11:33 TBC_x that's what I was proposing
11:33 VanessaE I've seen games on the C64 do that.
11:33 TBC_x probably in a wrong way
11:33 Calinou Doom had a minimap, VanessaE :P
11:33 Calinou very basic one, but then source ports made it textured and such
11:33 VanessaE Calinou: case in point.
11:34 TBC_x doom minimap was wireframe
11:34 RealBadAngel doom didnt have to check 8,500,000 nodes to get single frame :P
11:34 TBC_x mt minimap does not either
11:34 RealBadAngel well it does 8,500,000 / 16
11:34 TBC_x In what order are you processing the mapblocks?
11:34 RealBadAngel thx to the mapblock_mesh trick
11:35 RealBadAngel in a column you need to get highest non-air node
11:36 RealBadAngel you check cached blocks to get that node
11:36 RealBadAngel for each block stored in cache you have relative height and node stored
11:36 TBC_x too bad the server don't send mapblocks in such order
11:36 RealBadAngel TBC_x, what sending has to do with it?
11:37 RealBadAngel its only client side thing
11:37 RealBadAngel server is not aware of existence of minimap
11:37 TBC_x I know
11:37 RealBadAngel so i can respond after you
11:37 RealBadAngel bullshit
11:38 TBC_x but minimap probably doesn't check whether the opaque mapblocks sits beneath unloaded mapblock
11:38 RealBadAngel you havent spent any time to get the logic behind it and yet comment it
11:38 RealBadAngel ....propably....
11:38 TBC_x I'm just describing what I see is happening
11:39 RealBadAngel see the code, then comment it
11:39 RealBadAngel you dont get how it works at all, sorry
11:39 RealBadAngel if something is not loaded, its not in cache
11:40 TBC_x ...
11:40 TBC_x Just finish your minimap already
11:41 TBC_x at least into release-able state
11:41 VanessaE TBC_x: what do you mean "opaque-below-unloaded"?  you mean like how minetest mapper sometimes shows a lower, stone mapblock where you'd expect grass?
11:41 sfan5 <TBC_x> anyone looked into GNU/Hurd bug? https://lists.debian.org/debian-hurd/2014/10/msg00020.html
11:41 Amaz joined #minetest-dev
11:41 sfan5 hm
11:41 RealBadAngel basically i need to move scanning into propably completely new thread
11:41 sfan5 wouldn't s/__MACH__/__APPLE__/ make more sense there?
11:41 TBC_x VanessaE: yes
11:41 RealBadAngel main problem with mapblock_mesh updates is that it is triggered way to often
11:42 RealBadAngel client is flooded with mesh updates
11:42 VanessaE TBC_x: well honestly I don't see a problem there, since the client can't force the missing block to be loaded.
11:42 TBC_x then you shouldn't depend on that RBA
11:42 VanessaE TBC_x: are you suggesting that the entire column below the not-yet-loaded part just not be added to the minimap image?
11:42 RealBadAngel TBC_x, frankly that was c55 idea
11:43 RealBadAngel i didnt realized how much often is mesh update thread called
11:43 TBC_x well, now you see how it runs in practice
11:43 RealBadAngel on modern pcs, good enough
11:44 RealBadAngel having 60fps usually with all the gfx settings on, and minimap is quite good
11:44 TBC_x VanessaE: partly, yes
11:45 RealBadAngel but anyway, i will make that code faster
11:45 VanessaE RealBadAngel: again, there's NO time for long-term stuff.  you have to fix and push your code to your repo right away or you will miss the release
11:45 RealBadAngel but please dont push me to do it in a few days. i had a break lately, because of my health
11:46 RealBadAngel just need some time to finish it
11:46 TBC_x I thought that in master are allowed only finished features
11:46 RealBadAngel it is finished
11:47 VanessaE TBC_x: at this point yes, but generally apparently not :)
11:47 RealBadAngel but it doesnt mean it cannot be made faster
11:47 RealBadAngel but its not polishing a piece of code
11:47 RealBadAngel its rewritting some core parts of it
11:47 TBC_x RBA: Go search trash cans for a piece of hardware that you can test the minimap on
11:47 RealBadAngel lol
11:48 RealBadAngel go play that mc again and check of different mappers how much they cost
11:48 RealBadAngel before complaining mine is too slow
11:49 RealBadAngel our is capable of runnin the same frame rate as whole world :P
11:49 VanessaE stop comparing to MC
11:49 TBC_x I'm comparing minimaps
11:49 VanessaE just because A is slower than B doesn't mean B isn't also slow.
11:50 RealBadAngel so just disable the fuckin minimap for the release :P
11:50 TBC_x mapblock emerging could also use a culling algorithm
11:50 RealBadAngel i simply have not enough time before release to make it faster
11:51 RealBadAngel and get the sticker "486x ready" for 0.4.13 :P
11:51 VanessaE forget about making it faster!
11:51 VanessaE just fix whatever bugs it has
11:51 TBC_x support for 8086 would suffice
11:51 RealBadAngel meanwhile im fixing the bugs
11:52 VanessaE faster == feature.
11:52 RealBadAngel fsaa broken many things
11:52 TBC_x fix bugs, prepare for release, then make it god damn faster
11:52 VanessaE TBC_x: echo...echo...echo... ;)
11:52 RealBadAngel idk yet it will be faster. i just believe it will be
11:52 TBC_x just setting up RBA's TODO list
11:53 RealBadAngel what i know theres shitload of things to change
11:53 RealBadAngel mostly not related to minimap
11:53 RealBadAngel but bounded to FPS
11:54 RealBadAngel point is all those changes are effecting how the game overall feels
11:55 RealBadAngel "luckily" im coding VISIBLE changes, so im the boy to kick his ass when fps drops
11:55 TBC_x I think I already mentioned what milestones for the next release should be
11:55 Puma_rc joined #minetest-dev
11:55 TBC_x if it is either 0.5.0 or 0.4.14
11:56 RealBadAngel i can understand kicks when i screw something, like leaks. but not when im adding functionality that requires cpu or gpu power to work
11:56 RealBadAngel got trashcan pc? just turn them off for christ sake
11:56 RealBadAngel or buy yourself new box
11:57 TBC_x If I do that I can't afford food
11:57 TBC_x hmm
11:57 TBC_x screw food
11:57 RealBadAngel keeping compability with 9 yrs old standards is pain in the ass
11:57 RealBadAngel see opengl 2.1
11:57 TBC_x opengl 2.1 is ok
11:58 RealBadAngel no its not
11:58 RealBadAngel its a museum
11:58 TBC_x opengl 1.x is not
11:58 RealBadAngel nothing more than lighting and waving will be supported for <3.0
11:58 RealBadAngel instructions pipeline is too short in 2.1
11:58 TBC_x I don't care about reflections and stuff
11:59 RealBadAngel even parallax wont work there
11:59 RealBadAngel reflections? forget bout it
11:59 TBC_x tesselation on 16x16 is creepy
12:00 RealBadAngel such old gpus are simply not capable to have more than 512 instructions per shader
12:00 RealBadAngel and loops are unrolled, if you would like to know
12:01 RealBadAngel same applies to android devices
12:01 RealBadAngel im not sure yet about android watches and fridges :P
12:01 TBC_x well If people can't run minetest on their android fridges, we're totally screwed
12:02 TBC_x you definitely must support android watches and fridges
12:03 RealBadAngel :)
12:03 RealBadAngel i will try to do my best ;)
12:03 TBC_x well then... stop talking and go work!
12:04 TBC_x ;)
12:04 TBC_x I have at least three data races to fix
12:08 blaise joined #minetest-dev
12:12 TBC_x what is the purporse of RemoteClient::SetBlockNotSent?
12:15 TBC_x and why are two threads touching it?
12:20 Gael-de-Sailly joined #minetest-dev
12:25 blaze joined #minetest-dev
12:25 TBC_x oh, because the second call don't lock a mutex over it
12:25 proller joined #minetest-dev
12:35 Sketch2 joined #minetest-dev
12:35 Calinou <TBC_x> opengl 1.x is not
12:35 Calinou <leilol> heresy!!!
12:36 Sketch2 there's someone name Faker on Australia Skyblock claiming to be a dev. using a hacked client
12:37 TBC_x don't you know banhammers in australia?
12:38 Sketch2 he says he's testing for bugs in the game
12:38 TBC_x haha
12:38 Sketch2 anybody know the admin?
12:39 sfan5 !server skyblock
12:39 ShadowBot sfan5: server [--{name,address,ip,players,ping,port} <value>]
12:39 sfan5 oh right we have ShadowBot here
12:39 sfan5 Sketch2: this is something for #minetest, not this channel
12:39 sfan5 nothing we can do
12:41 Sketch2 sorry sfan5, I kinda realized that after I typed it.  I was originally thinking bcause he claimed to be a dev I'd verify in here first
12:48 RealBadAngel Sketch2, use a smoked kangooroo to smash that imposter
12:58 RealBadAngel Sketch2, but it can be that was cornernote, skyblock creator ;)
12:59 Sketch2 lols
13:00 RealBadAngel was it cornernote?
13:02 RealBadAngel Sketch2, i was also banned from VE's server a few days ago ;)
13:02 RealBadAngel was just testing my new tablet lol
13:05 H-H-H joined #minetest-dev
13:09 RealBadAngel ok. https://imgrush.com/GAIRcVrUFHVf <- after
13:09 RealBadAngel https://imgrush.com/w0AgWWov1350.png <- before
13:09 johnnyjoy joined #minetest-dev
13:10 RealBadAngel i solved fsaa problems, but it requires some lua changes to node defs
13:10 RealBadAngel same problem as we had before with grass
13:11 RealBadAngel we do need definition in node def saying if tile is tileable
13:11 RealBadAngel quite change for mod makers
13:13 RealBadAngel code is ready for that, just defs in game shall be updated
13:19 Gael-de-Sailly joined #minetest-dev
13:21 blaze joined #minetest-dev
13:37 TBC_x why can't this be for all plantlikes?
13:37 nrzkt RealBadAngel i want to run minetest on my TI-82, can you fix your minimap please
13:37 Sketch2 that's great
13:37 TBC_x default
13:39 RealBadAngel target is to run it on ZX-81
13:41 nrzkt perfect, thanks :D
13:41 TBC_x ClientInterface in clientiface.cpp was not designed to be multithreaded
13:41 RealBadAngel https://upload.wikimedia.org/wikipedia/commons/thumb/9/93/ZX81_kit.jpg/220px-ZX81_kit.jpg
13:42 RealBadAngel thats a pretty decent piece of hardware
13:42 RealBadAngel just twice as old as open gl 2.1 ;)
13:43 RealBadAngel nrzkt, https://upload.wikimedia.org/wikipedia/commons/thumb/1/14/Sinclair_ZX81_Setup_PhotoManipped.jpg/1280px-Sinclair_ZX81_Setup_PhotoManipped.jpg
13:44 RealBadAngel are you sure our network protocol supports tape recorders? ;)
13:44 TBC_x it will get rewritten anyway
13:46 Sketch2 I remember my Commodore64 had a 300 baud modem, wonder if that's fast enough to run a server
13:46 TBC_x we can then add support for pack mules
13:46 TBC_x that carry packets
13:48 Sketch2 I'd imagine it'd be possible.  what print them out, then scan them back in with OCR when they reach their destination?
13:49 Sketch2 wonder what the ping would be on that route
13:49 TBC_x we'll need to raise timeout a little
13:50 TBC_x if we make a clever mule deserialization, I think we can skip the OCR part
13:51 RealBadAngel ive seen c64 running web browser
13:51 TBC_x and WebGL
13:51 TBC_x ;)
13:51 Sketch2 They were pretty mean little machines
13:54 RealBadAngel and thats why i do want in the future one in
13:55 RealBadAngel exactly this one: http://libz80.sourceforge.net/
13:56 RealBadAngel nowadays pcs are able to make 15yrs old computers in a node
13:56 Sketch2 no kiddin
13:56 RealBadAngel eloraam made 6502
13:57 RealBadAngel i want z80 and cp/m
13:57 RealBadAngel *eloraam is creator of RedPower mod for mc
13:58 RealBadAngel computers there are fully programmable and able to control logic in the world
13:58 TBC_x mt should allow native code mods
13:58 TBC_x at least source-compatible
13:58 RealBadAngel fully means they are real computers with own IO, RAM and video
13:59 RealBadAngel once such lib is merged theres no problem
14:00 RealBadAngel z80lib is open source, compatible with us
14:00 Calinou native code mods suck
14:00 TBC_x gpl not lgpl
14:00 RealBadAngel this wont be a mod
14:00 Calinou not only it provokes people into making their mods proprietary, they are also a huge pain to make platform-independent
14:00 RealBadAngel but a feature
14:01 Calinou people should not be taught into trusting random .dll's and .so's
14:01 RealBadAngel no
14:01 TBC_x have you seen c55's buildat?
14:01 RealBadAngel i mean including the sources
14:01 RealBadAngel later on you can run whatever z80 soft you like
14:02 VanessaE RealBadAngel: I removed that ban when you reported it.
14:02 RealBadAngel https://www.youtube.com/watch?v=vb0Q9htsbBI
14:02 RealBadAngel VanessaE, i know, thx
14:03 RealBadAngel TBC_x, watch that vid
14:03 RealBadAngel its quite old
14:03 TBC_x i know that, I was playing with RP a lot
14:03 RealBadAngel its 6502 and forth interpreter build on top of it
14:04 RealBadAngel point is that this forth is an app
14:04 TBC_x i know
14:04 RealBadAngel for an engine feature
14:04 RealBadAngel z80 is lightweight
14:05 RealBadAngel as for nowadays cpus
14:05 RealBadAngel but can be lotsa fun
14:06 RealBadAngel we do lack such features, like wireing and computers
14:06 RealBadAngel mods cannot do that properly
14:06 RealBadAngel we had nearly perfect circuits PR there
14:06 RealBadAngel even c55 said its brilliant
14:07 RealBadAngel what happened to it?
14:07 TBC_x circuits in minetest_game?
14:07 RealBadAngel no
14:07 RealBadAngel in core
14:07 RealBadAngel no game can handle such thing
14:07 TBC_x like support for connections in core?
14:07 RealBadAngel not connections
14:07 RealBadAngel wires logic
14:08 RealBadAngel how could it be you have not seen that PR?
14:08 TBC_x core needs generic implementation
14:08 RealBadAngel there was one
14:08 RealBadAngel fm took it
14:09 RealBadAngel guy spent lotsa time on it
14:09 RealBadAngel SHAME we havent it merged
14:10 RealBadAngel now rebasing stuff propably could take ages
14:10 TBC_x we need at least smooth animations, smooth inventory handling and rewrite connection.cpp for next release
14:12 TBC_x and smarter node selection box that don't cause mesh updates
14:12 RealBadAngel youre all attempting ass side
14:12 RealBadAngel while basics are broken you do care bout face lifting :P
14:13 TBC_x is there a todo list of things to do?
14:13 RealBadAngel before i could get my ideas to work like parallax it took me over a year to get what ive imagined
14:13 RealBadAngel after such period you can see results of my work
14:14 RealBadAngel what i do plan is always long term
14:14 RealBadAngel even if im making stupid mistakes from time to time
14:15 TBC_x also... mesh generator needs occlusion culling
14:25 TBC_x why server thread calls SetBlocksNotSent?
14:42 Amaz joined #minetest-dev
14:50 COINTELBRO joined #minetest-dev
14:50 TBC_x it is more intertwinned that I expected
14:58 TBC_x that's a gordian knot
14:59 hmmmm joined #minetest-dev
15:00 Hunterz joined #minetest-dev
15:00 TBC_x hi hmmmm
15:05 blaze joined #minetest-dev
15:17 COINTELBRO joined #minetest-dev
15:19 sloantothebone joined #minetest-dev
15:25 rubenwardy joined #minetest-dev
15:37 H-H-H joined #minetest-dev
15:46 hmmmm hey
15:47 hmmmm looking through the logs
15:47 hmmmm dammit est
15:47 hmmmm this is exactly why I wanted unit tests added
15:49 hmmmm TBC_x:  I just looked through your thread sanitizer logs.  it seems like shared access of UDPPeer is the only current issue, which we haven't seen because modifications to the UDPPeer list are incredibly incommon network events
15:50 hmmmm TBC_x:  collecting more crash logs isn't going to help diagnose the actual problem sooner, just change the SharedBuffer to Buffer and be done with it
15:50 hmmmm so at least people won't crash
15:56 amadin joined #minetest-dev
15:56 amadin Please help server crashed every 3-6 hours https://forum.minetest.net/viewtopic.php?f=6&amp;t=12962
16:21 leat joined #minetest-dev
16:21 amadin buggest server in the world
16:42 twoelk joined #minetest-dev
16:45 Gael-de-Sailly joined #minetest-dev
16:57 hmmmm hrmm
16:58 hmmmm we should've told him to try disabling mods
16:58 hmmmm i'm not quite sure how, but it seems like his mod is corrupting Lua's internal state to the point where it has a runtime error attempting to generate a backtrace
17:09 lag01 joined #minetest-dev
17:30 leat joined #minetest-dev
17:40 FR^2 left #minetest-dev
17:43 celeron55 in case someone is following the github code of conduct bullshit and their recommendations for projects to have a CoC, here's a good one for us so this doesn't have to discussed at any point: https://github.com/ciafwywcoc/ciafwywcoc/blob/master/CODE_OF_CONDUCT
17:44 sfan5 wtfpl for for CoC
17:44 sfan5 :D
17:44 celeron55 yes; this wins the internet
17:45 sfan5 i like this one too: https://github.com/domgetter/NCoC
17:45 COINTELBRO nothing cuter than geeks being macho
17:45 celeron55 that works too
17:54 nrzkt joined #minetest-dev
17:59 alket joined #minetest-dev
18:04 COINTELBRO joined #minetest-dev
18:12 kilbith joined #minetest-dev
18:24 luizrpgluiz joined #minetest-dev
18:34 twoelk|2 joined #minetest-dev
18:35 GunshipPenguin joined #minetest-dev
18:45 luizrpgluiz left #minetest-dev
18:46 MinetestForFun joined #minetest-dev
18:52 twoelk|2 joined #minetest-dev
18:55 FR^2 joined #minetest-dev
19:06 TBC_x what is code of conduct about anyways?
19:06 Calinou "promoting diversity"
19:06 Calinou and avoiding harrassment
19:06 COINTELBRO basically a way of eliminating people who are shitheads in public without any further discussion
19:07 nrzkt where did you see this code of conduct ?
19:07 sfan5 <celero​n55> in case someone is following the github code of conduct bullshit and their recommendations for projects to have a CoC, here's a good one for us so this doesn't have to discussed at any point: https://github.com/ciafwywcoc/ciafwywcoc/blob/master/CODE_OF_CONDUCT
19:08 COINTELBRO alternatively, a plot by the united nations to steal our guns and with it our freedoms
19:08 sfan5 >guns
19:08 sfan5 do you live in usa?
19:08 TBC_x apperntly
19:09 TBC_x ICAN'T SPEEL
19:09 TBC_x ...
19:17 TBC_x I don't feel like having a CODE_OF_CONDUCT file
19:18 TBC_x strlen(CODE_OF_CONDUCT)
19:18 TBC_x too long
19:27 Gael-de-Sailly joined #minetest-dev
19:29 TBC_x looks like setBlocksNotSent is causing the heap corruption
19:31 COINTELBRO1 joined #minetest-dev
19:35 TBC_x how does std::vector<T> copy constructor work? deep copy or shallow copy?
19:39 kahrl TBC_x: deep in the sense it invokes the copy constructor for each element
19:39 H-H-H joined #minetest-dev
19:39 kahrl you can make it shallow by making T a pointer type
19:40 TBC_x changing Server::SetBlocksNotSent(std::map<v3s16, MapBlock *>& block) to Server::setBlocksNotSent(std::vector<u16>& clients, MapBlock *>& block)
19:41 TBC_x hmm
19:41 TBC_x this is inconsistent, type& var or type &var?
19:42 kahrl what < is closed by the last >?
19:43 TBC_x std::map<v3s16, MapBlock *> &block vs std::map<v3s16, MapBlock *>& block
19:44 kahrl doesn't make a difference to me
19:44 TBC_x look at & position
19:44 TBC_x &-space-variable or space-&-variable
19:44 kahrl that's just code style
19:44 TBC_x yes
19:45 kahrl I thought you were fixing the heap corruption
19:45 TBC_x well... yeah
19:45 TBC_x but I don't want to mess up code style
19:45 kahrl the guidelines recommend space-&
19:46 TBC_x ok
19:46 kahrl (I personally don't care, not even if it's inconsistent)
19:47 hmmmm TBC_x: type &var
19:47 TBC_x kk
19:47 hmmmm but
19:47 hmmmm you don't have to change existing code
19:47 hmmmm just don't use non-const references in new code
19:48 hmmmm if you're modifying some kind of value, use type *val;
19:48 hmmmm if you want to do a zero-copy pass of some kind of value, use const type &val
19:48 hmmmm never use "type &val"
19:48 hmmmm it's so misleading
19:48 TBC_x ok
19:49 hmmmm iirc, most other code standards ban non-const reference parameters as well
19:49 hmmmm with great reason!
19:50 TBC_x hmm... should I avoid method overloading?
19:51 TBC_x because I want to rename a method to setBlocksNotSentToAll
19:51 TBC_x or just setBlocksNotSentAll
19:55 kaeza joined #minetest-dev
19:55 RealBadAngel damn, i hate coding changes to tile.cpp it recompiles then all the shit around ;)
19:57 hmmmm TBC_x, try to avoid it if you can
19:57 hmmmm I don't think it's explicitly prohibited
20:01 TBC_x removing ClientInterface::getClientNoEx
20:02 TBC_x the mutex in there comes too late
20:02 nore joined #minetest-dev
20:04 TBC_x damn... the entire thing is wrong
20:04 RealBadAngel question: fsaa textures fix for plants requires defining if texture is tileable or not. shall i make plantlike non-tileable by default or mods shall have to define that in nodedefs?
20:05 RealBadAngel the second means all the existing nodedefs will have to be changed to get rid of a glitch
20:05 TBC_x clients get mutexed, but the whole list does not and clients can be freed at any time while other threads may be working on them
20:06 RealBadAngel hmmmm, what do you think?
20:06 TBC_x can't there be any default?
20:07 RealBadAngel default was tileable
20:07 RealBadAngel and when it comes to fsaa or any effect that modifies textures thats not so good
20:07 TBC_x what tileable exactly mean?
20:08 RealBadAngel that two tiles placed next to each other are tiling seamlessly
20:08 TBC_x so, every node using drawtypes is wrong?
20:09 RealBadAngel no, the problem is with just some of them
20:09 RealBadAngel but all the plants for example
20:10 TBC_x can't a drawtype define a default tileability?
20:10 RealBadAngel thats what im asking for, but that could create a mess if some of drawtypes will have another defaults
20:11 TBC_x what happens when you define all drawtypes nontileable?
20:11 RealBadAngel without fsaa, mipmapping and parallax - nothing propably
20:12 luizrpgluiz joined #minetest-dev
20:12 RealBadAngel but with any of the effects above connections between tileable nodes will be broken
20:12 TBC_x then go and take measurements
20:12 RealBadAngel i did
20:12 RealBadAngel ^^
20:12 TBC_x why did you say probably?
20:13 RealBadAngel well, forget "propably" ;)
20:13 RealBadAngel for sure
20:13 RealBadAngel in such case we shall define then which nodes have seamless faces
20:14 RealBadAngel and that could be even worse imho
20:14 RealBadAngel because most of the nodes are seamless
20:15 RealBadAngel http://pastie.org/10330112
20:16 RealBadAngel ^^ above shows whats needed to define which face is not seamless - in case of dirt with grass it cannot tile vertically, but can horizontally
20:17 RealBadAngel defaults are true - so only one face has to be modified
20:17 RealBadAngel erm, one tile - it goes for 4 faces in this case
20:21 TBC_x do Server class really span of several files?
20:29 RealBadAngel https://imgrush.com/OA94MUbCowIk.png
20:30 RealBadAngel jungle sapling has defined flags for its tiles
20:30 RealBadAngel so dirt with grass too
20:32 alket_ joined #minetest-dev
20:35 RealBadAngel ive found a bug while playing with those saplings
20:36 RealBadAngel https://imgrush.com/IF7Gb2fTsbBt.png
20:37 RealBadAngel accacia tree replaces node below the sapling with trunk node
20:37 TBC_x unfortunate for skyblockers
20:37 RealBadAngel not sure about other trees, going to check now
20:37 luizrpgluiz left #minetest-dev
20:42 nore joined #minetest-dev
20:43 TBC_x which thread handles server packets?
20:46 TBC_x Server and ClientInterface are pretty messed up
20:46 werwerwer joined #minetest-dev
20:49 crazyR_ joined #minetest-dev
21:02 OldCoder joined #minetest-dev
21:06 Taoki joined #minetest-dev
21:08 dzho joined #minetest-dev
21:31 lag01 joined #minetest-dev
21:36 RealBadAngel so, only accacia tree is bugged
21:36 RealBadAngel rest of the trees seems to be fine
21:37 TBC_x I wish I had your problems right now
21:37 RealBadAngel this is not mine but paramat's problem rather
21:38 RealBadAngel propably the schematic for the tree is wrong
21:49 TBC_x it is a problem of whoever the tree mod maintains
22:04 GunshipPenguin joined #minetest-dev
22:15 TBC_x Thos are my changes so far http://sprunge.us/NfEH
22:17 cvtsx joined #minetest-dev
22:18 TBC_x and I don't know whether it fixes the problem completely
22:19 TBC_x because EmergeThread may get into its way when Server is handling packets
22:19 TBC_x fixing that would require a rewrite
22:19 TBC_x probably
22:21 VanessaE joined #minetest-dev
22:22 TBC_x hi VanessaE
22:22 sloantothebone joined #minetest-dev
22:22 TBC_x got any new crash reports?
22:23 VanessaE hey
22:23 VanessaE *looks at stats*
22:23 TBC_x I think I may have partially fixed the problem
22:24 VanessaE I don't see anything in the past 24h
22:24 TBC_x partially because to fix it completely it probably will need a rewrite
22:25 VanessaE ( http://digitalaudioconcepts.com/vanessa/hobbies/minetest/stats.html -- 6am is the scheduled restart.  The big spike right after is just routine - either minetestmapper or tar'ing up the downloadable worlds, I forget which)
22:25 VanessaE oh yeah?
22:25 TBC_x I haven't made a PR yet
22:25 VanessaE (er 6am my time, 12:00 noon on the graph)
22:25 TBC_x http://sprunge.us/NfEH
22:25 TBC_x this is the patch
22:26 VanessaE you said "partially"..  what's missing?
22:26 TBC_x it is a gordian knot
22:27 VanessaE heh ok
22:27 VanessaE patch applied, recompiling now.
22:27 TBC_x the problem is, that packet handler is modifing client data structures directly
22:28 TBC_x and emergeThread may step in
22:28 TBC_x and trample each other's work
22:28 VanessaE ah
22:28 TBC_x in an undefined way
22:29 TBC_x I said `it may`, because I don't have it in logs and I don't believe it
22:29 TBC_x I mean, I don't believe the code
22:31 TBC_x I have removed ClientInterface::getClientNoEx because It potentially leaks clients outside of the Interface mutex
22:31 TBC_x which is happening anyway with the packet handler code
22:32 TBC_x so the problem still persists
22:32 TBC_x I just blocked one entry point
22:32 VanessaE you know, I think I actually understood that :)
22:33 TBC_x I would like to fix the problem once and for all, but given feature freeze won't allow me to do It without very very very nasty hacks
22:34 TBC_x and I refuse to be a nasty person
22:34 VanessaE well to be fair,
22:34 VanessaE you're fixing a bug here, not adding a feature
22:34 VanessaE so I'd say it's fair game
22:34 VanessaE (the real problem isn't the advent of the freeze, it's more the length of it - time required for testing any fixes_)
22:35 TBC_x yeah
22:35 proller joined #minetest-dev
22:35 TBC_x but I don't know how others will react to refactored code within freeze window
22:36 TBC_x on the other side, I don't want to have this bug sitting in the released version
22:41 VanessaE well if worse comes to worst, extend the freeze window long enough to iron out any new bugs caused by a refactor, seeing as how you'd need to do that ANYway to fix the original bug
22:43 TBC_x also, a RC could be released instead
22:45 VanessaE that's a thought
22:45 VanessaE we haven't done an RC-alike release since....um..   0.4.0-d1?
22:45 VanessaE maybe once more since then
22:46 kilbith no, 0.4.12 had a RC
22:46 kilbith https://forum.minetest.net/viewtopic.php?f=18&amp;t=11227
22:48 VanessaE it did?
22:48 VanessaE huh.
22:48 VanessaE I'd quite forgotten about those builds.
22:48 VanessaE still, they weren't actually *tagged* as -rc in the repo were they?
22:51 TBC_x does minetest have any vertical content yet?
22:53 TBC_x except caves
22:53 VanessaE nope.
22:54 VanessaE there are mods that add bedrock and some Nether-like region, and a couple that add stuff up in the sky like nyanland and floatlands
22:54 VanessaE but nothing in-core or in minetest_game
22:54 VanessaE (excuse me, nyancat heaven I think the first was called)
22:54 TBC_x I would really welcome something from df universe
22:54 VanessaE df?
22:54 TBC_x dwarf fortress
22:54 VanessaE dwarf?
22:55 VanessaE right
22:55 TBC_x hmm
22:56 TBC_x because I've got heap-use-after-free in my log, I doubt that the patch fixed the problem
22:56 TBC_x as I said, it fixed single entry point
22:56 * TBC_x is looking at the logs again
22:57 TBC_x I've just fixed concurrent write
22:57 TBC_x from not-a-packet-handler
23:00 TBC_x expanded template parameter is so unhelpful
23:10 TBC_x the solution is to remove any RemoteClient *getClient() from ClientInterface
23:10 TBC_x because that is not how interfaces work
23:19 TBC_x https://github.com/t0suj4/minetest/commit/3449f8df5592f98c115c3884b6ac74a636feb2cd
23:20 TBC_x I'm not gonna create a PR from that though
23:20 VanessaE should I nuke the previous patch and pull ^^^^ that?
23:20 TBC_x it's the same
23:21 TBC_x I put it up just for educational purporses
23:21 TBC_x :)
23:22 VanessaE oh heh, ok
23:32 TBC_x ok, lets get hands dirty
23:32 VanessaE uh oh.
23:56 TBC_x wtf is setting full_block_send_enable_min_time_from_building?
23:56 VanessaE an anti-lag measure
23:57 VanessaE if a player places a node, wait ^^^^ that amount of time before sending new mapblocks to them
23:57 VanessaE or something similar anyway.  not sure if that function still does anything now though
23:58 TBC_x max_simul_sends_usually is awesome variable name
23:58 VanessaE hah

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