Time  Nick      Message
09:00 sapier    hmmm I've got some ideas to get mobf even faster but as there's conflict about movement mechanism to use I'm not sure if they ever will be added. Basicaly there are two ways to do movement
09:00 sapier    1) velocity/path based (with optional client side smoothing in path mode) --> simple mobs uses velocity
09:00 sapier    2) acceleration based (client already does prediction) --> mobf
09:00 sapier    both have advantages and dissadvantages
09:01 Jordach   disadvantages for 1
09:01 sapier    first one for example is more usefull if a mob tries to get to a special location while second one is very good to get a random movement
09:01 sapier    disadvatages for one is you can't get smooth movement if you don't use paths while you will get outdated movement if you use paths
09:04 sapier    you should consider most mobs beeing in random move mode almost 100% of time
09:06 sapier    and of course you could combine acceleration with paths too ... it'd just get even more complex ;-)
09:07 Jordach   minecraft uses a acceleration system
09:07 Jordach   as if youre running from a creeper, they have smooth turns like a vector
09:08 sapier    jordach I already added a pull request that'd make same in minetest ... its about half a year old
09:09 sapier    but as long as pilzadam and me didn't agree for a movement mechanism patches improoving acceleration based movement aren't added that fast ;-)
09:09 sapier    -didn't + don't
09:43 proller   sapier, add players number to serverlist please!
09:49 sapier    vanessae already complained about the flags so I don't think it'll get better if I add playernumber too ... we need to find a better way to show that information
09:59 proller   players number is most important
09:59 proller   it was before formspec without problem
09:59 proller   FLAGS 5/100 server namee e e ee ee
10:07 Taoki     sapier: I support the movement wat which allows for least lag and most normal movement. Since currently that's what I found to be the biggest issue with mobs in Simple Mobs at least
10:07 Taoki     I don't use mobf cuz the default animals don't have such nice models. simplemobs has artistically accurate animals
10:07 Taoki     I should try the other one too though
10:08 Taoki     Acceleration based sounds a bit better though. Should be more realistic movenent
10:16 Taoki     I also think mobf has excessive items and such. Like vombie trap, net, etc. Those would be too much for default IMO
10:23 sapier    mobf is a framework, any mob is a single mod
10:23 sapier    so by default mobf has not any item ... except of path marker
10:24 sapier    and lag is an issue for all movement styles ... depends on current situation which one will "hide" the effects of lag better ... I don't think either one is better in general
10:25 sapier    proller indention issue is even worse for player numbers as they don't even have a fixed digit count
10:26 sapier    maybe font used before is different than now
10:26 proller   but server list without players - its stupid and useless
10:26 proller   add witout identation, then think how to fix
10:28 sapier    already did that for flags and got complains, until this is fixed I won't make it worse by a single request ... I always have to add additional features to fix issues added by a request of a single person ... if you manage to gett supporters I gladly add it but as long as its 1:1 I don't add it
10:29 proller   it was before your changes
10:29 proller   you broke it
10:29 sapier    and now  it isn't.
10:29 proller   then add it now
10:29 sapier    I never intended to create a 100% identical replacement but provide a base for custom changes
10:30 sapier    and I told that from begining if you want this feature then add it and fix all issues resulting from your changes
10:30 Taoki     Trying mobf now. Focusing on the system though
10:30 Calinou   inb4 crash
10:30 Taoki     The most huge issue of mobs imo is the lack of angle smoothing. Which is also an issue for players and of course all lua entities
10:30 Taoki     Why do we have movement smoothing but no angle smoothing yet?
10:31 sapier    taoki there's a pull request
10:31 Taoki     What's the link? And why isn't it in yet?
10:31 proller   sapier, identation - its not issue
10:31 sapier    proller it is ask vanessae
10:31 proller   VanessaE, !!!
10:33 sapier    taoki https://github.com/minetest/minetest/pull/418
10:33 Taoki     thanks
10:33 Calinou   Taoki: I tried doing angle smoothing once, and epicly failed 8)
10:33 * Calinou would like to see entity pitch too
10:33 sapier    but I don't know if it aplies cleanly I last rebased it in may
10:33 Calinou   and a visual representation of pitch
10:33 Calinou   (rotate model by 25% of pitch?)
10:33 Taoki     Angle smoothing is needed for players too a lot
10:33 Calinou   yes
10:34 Calinou   try coding it, you're probably better at C++ than me :P
10:36 Taoki     It will probably be put on hold for months like my fog color is, so I feel a bit less motivated to work on code now
10:36 Calinou   le "just now" comment
10:38 sapier    taoki I'm not sure if current mobf version uses the automatic rotating completely as the lua style rotation needs to be disabled to get full benefit of automation
10:38 sapier    and I can't diable lua rotation while the patch isn't added
10:38 Taoki     sapier: The code isn't in so it can't use it
10:39 Taoki     Even Lua rotation would look good however, if only there was angle smoothing in >.<
10:39 Calinou   it's probably easy to do
10:39 Taoki     And no one did it for years :P I need to do other things since I'll be away soon, so can't really look into it now
10:39 sapier    in most cases you don't need smooting a cow for example looks into their movement direction 99% of time ;-)
10:40 Calinou   Taoki: this is FOSS 8)
10:40 Calinou   noone does anything for years
10:40 Taoki     heh
10:40 Calinou   sapier: you do
10:40 sapier    what shall I do?
10:44 Taoki     Uh... even vombies lag all over the place in singleplayer :(
10:44 sapier    vombies are agressive animals those require a lot of cpu power
10:45 sapier    did you try  with current git or 0.4.7?
10:45 Taoki     I wonder if for movement we could add a direct data stream system. And only leave tcp packets for one-time things
10:45 Taoki     That would be of most help
10:46 sapier    mobf is only tested against 0.4.7 don't know if git added regressions
10:47 Taoki     le server crash in singleplaer (prolly vombies, everything just stopped working)
10:47 Taoki     yah, can't even shut it down, prolly gotta kill the process
10:47 sapier    yes I know there's a bug in 0.4.7 stopping agressive animals from work at all
10:48 Taoki     And minetest is stuck at 13% CPU usage
10:48 Calinou   Taoki: tip: hyperthreading fakes CPU usage
10:48 Taoki     And about 800 MB of ram
10:48 Calinou   you should double it to get "actual" CPU usage
10:48 Calinou   if you use HT
10:48 Taoki     ok, then 26% :P
10:48 Calinou   you use HT?
10:48 sapier    doesn't change situation ;-)
10:48 Taoki     yeah, had to kill minetest
10:48 Taoki     mob framework is not even SAFE >.<
10:49 sapier    no it's as safe as it can be with core bugs causing a crash ;-P
10:49 sapier    I already workaround lots of core bugs sorry for missing one
10:49 Calinou   for work
10:50 Calinou   ITT: mobf is NSFW
10:50 sapier    guys stop complaining and report errors
10:51 sapier    and by reporting errors I don't speak about "it's slow" and "it doesn't" work .... especialy if after days of investigation result is that some other mod you happened to install same time skrewed up
10:53 sapier    mobf has lot's of features I ain't capable of achieving 100% feature test on my own anymore if there won't be help it's most likely gonna die some time in future
10:53 sapier    at least if I have to fight against mobf haters same time :-P
10:56 Taoki     Yeah. I can quickly tell Simple Mobs is SO much more consistent and even realistic (only except movement)
10:56 Taoki     And way less intensive
10:56 sapier    ok so simplemobs is consistend except of MAIN feature?
10:57 Taoki     It seems to be less intensive and even locally, less laggy
10:57 sapier    it IS less intensive as working around that glitches sm has requires cpu power
10:57 sapier    this is wash me but don't get me wet argument taoki
10:58 Taoki     hmm?
10:58 sapier    mobf checks for all that invalid movement states that occur in sm ... checking can't be done without using cpu
10:59 sapier    therefore mobf does require more cpu power
10:59 Taoki     ok
11:00 Taoki     I still tend to like it more currently. Just think movement and collision detection needs some improving though
11:00 sapier    if you improove simple mobs in that areas you end up at mobf
11:00 sapier    because that are the main (performance) differences
11:01 Taoki     hmm...
11:01 sapier    second main difference is sm requries any mob to implement same functions over and over again while mobf already contains it
11:01 sapier    so if I fix a bug it's fixed for ALL mobs
11:01 Taoki     prolly some architecture difference too. Mob defnitions seem a little easier to understand in simple mobs. Not saying mobf is bad though... surely both are good
11:01 * Taoki   nods
11:02 sapier    yes they are more easy as things you have to implement in sm can just be enabled by definition in mobf
11:02 Taoki     Ah, and I forgot you're the owner of mobf >.< Yeah, I'm not saying yours is bad that's for sure. Just thinking which seems to be most usable at this point
11:03 sapier    it's like deciding between a prototype not implementing challanging features to a full blown implementation needing bugfixing
11:03 Taoki     Sadly, I have to admit that mobf just crashed minetest earlier (had to kill the process) and got it stuck at 13% CPU usage + 600 MB of RAM. Which of course happens... nothing can be fully stable nd all, but just reporting it
11:04 sapier    mobf is in no state to be merged to core of course
11:04 Taoki     With simple mobs I seem to be at 11% CPU too, but (not sure if related) about 200MB of ram only
11:04 sapier    how much mobs are implemented in sm and mobf?
11:04 sapier    you need to compare same things taoki
11:05 Taoki     SM has fewer mobs, MOBF more. I do compare at the same quantity of existing surrounding mobs though
11:05 proller   meeelting  https://github.com/proller/minetest/compare/master...weather
11:06 sapier    which isn't very accurate as it depends on what type of mob is around too
11:07 sapier    50 sheep for example are way less cpu consuming than 25 vombies wolfs or archers
11:07 Taoki     yeah
11:07 sapier    this results primary by sheep don't attack player thus don't need to look for players around them
11:08 Taoki     sapier: BTW. Are you considering making a version of MOBF with the mobs in SM? Mostly the same models and perhpas different functionality.
11:08 Taoki     True too yes
11:08 Taoki     I need to test the aggressibe mobs in SM
11:08 Taoki     But can't find any
11:08 sapier    call mobf_settings you can disable them by checkbox
11:08 Taoki     There are some sand monsters but they're being nice >.<
11:08 sapier    I don't see any benefit of removing mobs if it's that easy to disable ... the model thing is on my list yes
11:09 Taoki     ok. Will mob definitions from SM work in MOBF?
11:10 Taoki     Noticed a little something else: In SM mobs can jump to bet up blocks. In MOBF they just appear a step higher (same issue in Minecraft IIRC)
11:10 sapier    no they never will work
11:10 Taoki     ok
11:11 sapier    it's completely different
11:11 sapier    mobf definition for most sm's is way more simple in mobf
11:12 Taoki     Trying to get something to attack me in sm but no mobs are
11:12 sapier    jumping up is difficult but possible
11:13 sapier    no not difficult .. but require additional cpu power again :-)
11:13 Taoki     Anyway, I hope my test pointed out one of the bugs. mobf can crash the server with latest Minetest GIT, so something goes wrong
11:13 sapier    a mob needs to decide if it wil be able to reach the block above by calculating it's distance acceleration andvelocity
11:14 sapier    hmm latest git you say? mobf is 0.4.7 only atm but I guess I need to fix that issue for 0.4.8 anyway :-)
11:15 sapier    did you do anything special to crash it?
11:15 Taoki     Also having mobf from git
11:15 Taoki     sapier: No. Just walking around with 4-5 vombies around me
11:15 Taoki     Then suddenly, all fell to silence
11:15 sapier    hmm I wonder if that bug in 0.4.7 isn't fixed in core yet
11:16 sapier    but I need to complete mainmenu first
11:19 Taoki     sapier: Well a bunch of sand monsters are attacking me in sm. No intensivness however
11:19 sapier    I guess pilzadam did completely different agressive handling
11:19 Taoki     The way they're running after me is also very good looking
11:20 Taoki     But again I'm not saying your mod is bad :( You did a nice work on mobf too
11:20 sapier    did you try what happens while different player attack sandmonsters?
11:20 sapier    or you have a sibling attacking sandmonsters along with you?
11:20 Taoki     Trying in singleplayer now, no one else to test with
11:20 sapier    I don't think sm is prepared for this
11:21 sapier    I know you can do attacking very simple ... maybe mobf just checks "too much" and players are completely ok with the glitches that occur that way
11:22 sapier    or maybe there are even major bugs in
11:22 Taoki     Well... barely and CPU usage or memory usage or slowness with 2 sand monsters attacking me
11:23 sapier    I'll have a look what pilzadam does there but fighting is quite advanced in mobf ... I guess sm does the basic thing I added first ... about  a year ago
11:24 Taoki     True. The intensivness in mobf is likely because it's more advanced indeed
11:24 PilzAdam  sapier, it always attacks the nearest player
11:24 sapier    so most simple way as I guessed
11:24 Taoki     3 sound monsters now. minetest is barely using any resources :P
11:25 sapier    mobf doesn't stop attacking a player once it started attacj
11:25 sapier    unless player runs away to far
11:25 Taoki     But yeah... if any system is to be added to builtin, I am in favor of sm. Mobf can be very useful too for anything that needs more complexity, so it would not be useless
11:25 sapier    latest version even usses aggro to decide whom to attack
11:25 * Taoki   would feel bad if mobf got lost, since sapier prolly put a lot of work into it too
11:26 sapier    if sm was added I'm gonna drop mobf as it'd be wasted time to continue
11:26 Taoki     That would suck :<
11:26 sapier    this is not a threat but just a consequence
11:27 Taoki     I wonder if there's any way to slowly improve sm with code from mobf, and perhaps get to a merge between the two mobs. That results in simplest but also most complete functionality
11:27 sapier    mobf never ever will have a chance against a in core mob implementation
11:27 Taoki     I know, it makes sense
11:27 Jordach   or: move mob ai to core
11:27 sapier    it'd end up at mobf complexity
11:27 Jordach   problem solved, and have extra stuff inbetween
11:27 sapier    with exactly same issues
11:28 Taoki     I think mobf features could slowly be moved over to sm. Testing each one for how much resource usage it adds and how well it works
11:28 sapier    and of course you understand I'd not be very glad with "simple mobs by pilzadam" while all the challanging issues are taken by mobf ;-P
11:28 Taoki     I am curious though: What extra feature does mobf has over sm? Without looking at the code, I se ethe same detail in both. Including how mobs attack me
11:28 Taoki     sapier: I know. I'd feel bad if your work got lost... I know how I'd feel if that happened to me :(
11:29 PilzAdam  Jordach, ok, then please write the API (not the code, but the docs) for it
11:29 sapier    collision detection, movement prediction, drop prevention
11:29 sapier    siblings
11:29 sapier    ridable mobs
11:29 Jordach   PilzAdam, i can do that in a day
11:29 PilzAdam  Jordach, go ahead
11:29 sapier    guards
11:29 sapier    patrol mode
11:29 Jordach   sapier, scale the model down then up again
11:29 Taoki     nice. Lack of collision detection is what I dislike most in smobs
11:29 sapier    yes can be done in mobf too
11:29 PilzAdam  sapier, movememnt prediction cant be done without client side changes
11:30 sapier    it can
11:30 Taoki     What's movement prediction do exactly? And what does drop prevention mean?
11:30 PilzAdam  and collision detection is builtin in core....
11:30 sapier    acceleration is predicted by client
11:30 Taoki     oh
11:30 sapier    collision detection is done within mobf to decide where to reroute a mob too
11:30 PilzAdam  no, not acceleration, velocity is predicted based on acceleration
11:31 sapier    drop prevention means a mobf mob won't drop into water for example
11:31 sapier    flying mobs
11:31 sapier    fish
11:31 Taoki     Well again, I really don't want to make you feel bad about mobf >.< But those sound like they can be added to smobs easily too. Possibly in a not so intensive way
11:31 PilzAdam  acceleration always stays the same until you get a new package from the server
11:31 Taoki     smobs doesn't have flying and fish?
11:31 Taoki     Ah, so not falling off cliffs and stuff
11:32 Taoki     PilzAdam: We really need angle smoothing, the same way we have position smoothing. For both players and mobs
11:32 sapier    taoki yes once work has been done what to do some features can be added ... others won't be possible as design doesn't match
11:32 Taoki     I don't get why in years it wasn't added, since it should be pretty basic :P
11:34 Taoki     PilzAdam: When could collision detection be added in simple mobs? In a simple way of course (since that's the point of SIMPLE mobs) but enough to keep them from blindly going into walls. Also perhaps drop avoidance, sop they don't jump off cliffs like fools
11:34 sapier    and no I won't help developing sm just because its a core developer mod
11:34 PilzAdam  Taoki, the collisiondetectioin is built-in in the core
11:34 Taoki     sapier: I don't like or dislike it because it was made by a core dev, that has no influence on me. I just... like the design more I guess
11:35 Taoki     PilzAdam: Ah, so it already should work? Last time I saw a sheep pushing int a wall
11:35 Taoki     Or maybe sm doesn't use this builtin option yet
11:35 PilzAdam  they dont go through the wall, thats collission detection
11:35 sapier    Taoki you like the designe because it's still simple due to lack of features
11:35 Taoki     ok, maybe I saw an older version then
11:35 sapier    what you're talking about Taoki is collision prediction not detection
11:36 Taoki     sapier: Truth is I don't feel a lack of features in practive between mobf and smobs. Both seem to have the same features. Again, from what I can feel without knowing the code
11:36 Taoki     In both mobs roam around, in both they attack, etc
11:36 sapier    Yes you don't feel it because it's the minor glitches that cause 99% of work
11:37 Taoki     The ONLY differences I noticed is that in mobf, mobs turn around more gradually as they walk, whereas in smobs turning is very stiff (I don't mind though). Also that in smobs mobs jump, whereas in mobf they just appear off higher blocks
11:37 sapier    it's no effort to get mobs running aroud I did this in a day
11:37 * Taoki   nods
11:37 Taoki     Oh yes, and mobf has birds and fish. I don't know if smobs can do those yet
11:38 Taoki     That is indeed a nice thing
11:38 PilzAdam  sapier, I did whole simplemobs in a day
11:38 sapier    I did mind about the stiff movement that's the main reason for acceleration based movement .. if you don't care about it go with sm ... but this issue can't be fixed there
11:38 sapier    PilzAdam you did read mobf first
11:38 PilzAdam  of course not
11:38 Taoki     I meant stiff rotation. But stiff movement is true too; smobs have no acceleration
11:38 PilzAdam  nobody understands your code :-p
11:39 Taoki     PilzAdam: Can acceleration be added as well to smobs?
11:39 sapier    Yes I know you even asked about git access to mobf code before you startet sm
11:39 PilzAdam  oO
11:39 sapier    so don't tell me you didn't look at it
11:39 Taoki     Great... what a nice reason for a dev fight again >_>
11:39 * Taoki   le sigh
11:39 sapier    it's not a fight ;-)
11:39 PilzAdam  I asked you to put it in a repo so its easier to keep it up to date
11:39 PilzAdam  Taoki, :D
11:40 sapier    no acceleration can't be added because sms design doesn't match to it
11:40 Taoki     BTW: Do either of the mods allow spawning mobs with chat commands?
11:40 sapier    both do
11:40 Taoki     nice
11:40 Jordach   /spawnentity mobs:sheep
11:40 Taoki     Just got 5 sand monsters lashing after me lol
11:40 PilzAdam  sapier, can you explain me how mobf uses acceleration?
11:41 sapier    PilzAdam looking at how many mods you cloned you'd created a own mod either way ;-)
11:41 sapier    mobf doen't set velocity but acceleration
11:41 Taoki     haha, 6. Still no extra cpu usage though. I like testing this ^^
11:42 sapier    to do that in a sane way it periodicaly checks where a mob will be in and adjusts it's acceleration to avoid collisions or drops
11:42 PilzAdam  sapier, and how does that add prediction for moving in a not straight way?
11:42 Taoki     sapier: Truth be told, I am a bit unsure how much acceleration is needed for *mobs*. I mean in real life when you walk, you barely have any noticeable acceleration. You just get up and walk... in less than a second you're running at "chosen speed" ;)
11:42 Taoki     Even if it is a nice addition sure
11:43 Taoki     Acceleration woulkd be useful for large mobs like dragons however
11:43 sapier    actually that "less than a second" matters it's one of those minor issues humans notice and realize as artificial
11:44 Taoki     Well less than half second or so. Though it can be noticeable yeah
11:44 Taoki     PilzAdam: Where could the mobement be changed to acceleration in smobs too? Doesn't what work at all for smobs?
11:44 Taoki     Can't test it now thought since I'll be more busy till I head off in a week
11:45 sapier    it's not enough to acceleration taoki you need to switch handling completely if you don't use velocity anymore
11:45 Taoki     ah
11:46 PilzAdam  sapier, are you going to answer may question?
11:46 sapier    I still try to understand your question ;-)
11:46 PilzAdam  bbl
11:46 sapier    what do you mean wit prediction in a non straight way?
11:47 sapier    if you use acceleration you always get fluent direction changes
11:48 sapier    unless you use very high acceleration values but that's a problem for other reasons
11:48 sapier    did that answer your question or did I misinterpret it?
11:49 Taoki     sapier: Angle still can't be set fluently yet, sadly
11:50 sapier    I already told an hour ago there is a pull request that will do this
11:50 sapier    I can't do anything against core dev's not adding things
11:50 Taoki     Right, so nothing other than that pull request. ok
11:51 sapier    and maybe a small fix in mobf to disable the workaround in there
12:17 proller   https://github.com/proller/minetest/compare/weather
12:18 proller   PilzAdam, you must say "NOOOO" 8)
12:19 proller   or maybe only 2 commits - https://github.com/proller/minetest/commit/fad2b4e589a863a82196f81ec704484e25a82dea
12:20 Exio4     can i ask why you change those NoiseParams?
12:20 proller   https://github.com/proller/minetest/commit/6d7daf9b4202d26e88e9bb4b0236911e08541a27
12:21 Exio4     and.. other thing
12:22 Exio4     are you using the C API to wrap back to lua code?
12:23 Exio4     https://github.com/proller/minetest/commit/6d7daf9b4202d26e88e9bb4b0236911e08541a27#L3R236
12:24 proller   because heat not used now
12:25 proller   yes, it for dropping torch when ice below melt
12:25 Exio4     if i understand correctly the code, that can be done inside lua
12:26 Exio4     https://github.com/proller/minetest/commit/77f7c824025201d811dc923eb8cf03d973f5aba5#L4R306
12:26 Exio4     you don't just move a variable from private to public
12:28 proller   Exio4, done what? slow abm?
12:28 Exio4     is it used directly in the ABM?
12:28 proller   yes, abm must call nodeupdate when melt
12:29 proller   Exio4, why not move?
12:30 Exio4     well, you are wrapping to lua anyway, i don't think that would matter..
12:30 proller   but ok. will make get*
12:30 Exio4     proller: the notation
12:31 Exio4     proller: something "good" would be changing the code that uses the m_ though
12:32 Exio4     as that would be a "small" cleanup :P
12:40 Taoki     https://github.com/proller/minetest/commit/77f7c824025201d811dc923eb8cf03d973f5aba5#L4R306 Why would antone HARD-CODE weather types >___<
12:41 Taoki     +  u8 getRain(ServerEnvironment *env, v3s16 p);
12:41 Taoki     +  u8 getSnow(ServerEnvironment *env, v3s16 p);
12:41 Taoki     Totally against if any words like "rain" or "snow" are added to the code. weather should only be defined in Lua IMHO
12:41 Taoki     Only thing I agree should be added for weather to the code is support for sky particles and fog color changes.
12:41 Exio4     +      n.setContent(n.getContent() == c_water_source ? ndef->getId("default:ice") : ndef->getId("default:snow"));
12:42 Exio4     +      map->addNodeWithEvent(p, n);
12:42 Exio4     are you hardcoding ice and snow name
12:42 Exio4     +s
12:42 Taoki     Even worse IMO
12:44 Exio4     if the ABM needs to "know that much" about lua, it should be defined in lua, not C++
12:44 Exio4     neighbors.insert("default:furnace_active");
12:44 Exio4     you are hardcoding a hell lot of nodes
12:47 Taoki     proller: The idea itself is good. But it's very very wrong to hard code nodes at this day when we have the Lua API. Code is aimed at containing mechanics for defining things in Lua. Nodes, weather types (in this case), etc. must be defined in Lua only
12:48 Taoki     So the only thing that IMO should be coded is support for sky particles. And Lua-side fog colors, though those are part of another plan of mine to allow custom skies
12:48 proller   its version 0.001
12:48 proller   in first i want to make something working
12:49 proller   in second - make some on uglua
12:49 Taoki     Oh, ok. If that's not intended to be final then sure
12:50 Taoki     Idea is, please don't hard code any nodes nor weather types. Just the system to allow weather., Types of weather, their sounds, their particles, their fog color, and all that should be in Lua scripts
13:04 PilzAdam  sapier, I want to compare your system to the one where you send the target pos to the client
13:06 PilzAdam  you get really smooth curves in the target system, while the accelration one will always delay
13:13 PilzAdam  proller, NOOOO!
13:13 PilzAdam  :-p
13:13 PilzAdam  why do you do that in core?
13:13 PilzAdam  it would work perfectly fine in lua with this slow interval and low chance
13:22 proller   PilzAdam, its faster
13:23 PilzAdam  it doesnt need to be fast with this low chance and slow interval
13:28 proller   its must analyze blocks around
13:28 proller   and it must freeze big oceans
13:29 PilzAdam  and you hardcode nodenames? oO
13:29 PilzAdam  bbl (sry)
13:32 proller   harcode for melting
13:33 proller   for preserve level
13:33 proller   _flowing-> snow
13:33 proller   _source->ice
13:33 proller   or maybe add melt definition
13:33 proller   and freeze
13:40 proller   is any objections about first two commits?
13:48 proller   okay, willl commit
13:48 Exio4     ehm
13:49 Exio4     that is not
13:49 Exio4     good
13:49 Exio4     the idea of asking for other core dev is waiting for *him* to accept, not "objections"
13:50 Exio4     silence doesn't mean "everyone is ok and happy with it in the actual form"
13:51 proller   but it works
13:52 proller   ok, accept it please!
13:52 Exio4     http://dev.minetest.net/Git_Guidelines
13:52 Exio4     You can push something to upstream [1] only if some other member of the core development team agrees on it [2]
13:52 Exio4     *only* agrees, not nobody *disagrees*
13:53 proller   even if PilzAdam disagrees and somebody  agrees?
13:53 Exio4     in that case you'll need to hear to PilzAdam and try to see, maybe it is something trivial? but the change can be completly wrong or breaking other thing or so
14:01 proller   its 51% joke
14:02 proller   m_game_time returned
14:03 Taoki     http://forum.minetest.net/viewtopic.php?pid=99636#p99636 Might be an interesting improvement to consider
14:06 Taoki     PilzAdam: ^ Also related to mob movement, which IMO this could improve a lot
14:19 PilzAdam  Taoki, even the first sentence is wrong
14:19 PilzAdam  we use UDP
14:19 Taoki     PilzAdam: Oh, sorry about that. I remembered TCP for some reason
14:20 Taoki     Corrected
14:20 PilzAdam  proller, hardcoding node names is a complete nogo
14:21 PilzAdam  this neighbor thingy can be done in Lua too without any major performance regressions
14:21 PilzAdam  hardcoding these ABMs is just plain wrong
14:22 PilzAdam  and how is melting related to "weather"?
14:22 proller   its first step
14:23 PilzAdam  move these ABMs to minetest_game and add API calls for get_heat() etc.
14:23 proller   falling snow will fill all around without melting
14:24 proller   now minetest use 100-200% i5 cpu
14:24 proller   with this abm near ocean it will be 300+
14:24 Exio4     200%? how are you doing that black magic
14:24 Exio4     proller: 100%, singlethread
14:25 proller   23:30 126.17% minetest
14:25 proller   now 4 threads
14:27 PilzAdam  your performance argument is wrong; Lua wont decrease the performance that much
14:27 Exio4     yes, 126%, you can't make lua multithreaded
14:27 Exio4     and with luajit, it can even be faster
14:27 PilzAdam  maybe the ABM itself is the problem, but hardocing wont help then
14:30 PilzAdam  !join #wolfgame
14:30 PilzAdam  oops
14:36 proller   PilzAdam, i make .frozemelt property for snow and waters
15:04 Taoki     What's wolfgame even? I hear a lot about it here :)
15:04 Taoki     a minetest game?
15:04 PilzAdam  -> #minetest
15:54 kahrl     Any news on #822/#827? I think it's quite an important issue because it prevents servers from running
16:01 proller   removed hardcode https://github.com/proller/minetest/compare/weather
16:04 proller   and support for it in game: https://github.com/proller/minetest_game/compare/weather
16:04 proller   accept someone 8)
16:04 proller   its not final and disabled by default
17:06 Jordach   hmmmm, do you have your custom MGV7 lua settings?
17:06 Jordach   please pastebin me a copy, i could do with one atm
17:07 PilzAdam  Jordach, Pilztest uses v7
17:07 Jordach   PilzAdam, i want to use minetest_game with MGV7
17:07 Jordach   not your game
17:07 Jordach   so snow biomes, desert etc
17:07 PilzAdam  you can create them yourself
17:08 Jordach   logs it is then
17:27 Jordach   PilzAdam, found it: http://pastebin.com/1kjU5tN7
17:49 proller   Jordach, is this definitions in git?
17:50 Jordach   nope.
17:50 Jordach   no MGV7 defines in git#
18:10 celeron55 kahrl: if it works and doesn't do anything that potentially breaks something, i'm okay with it
18:17 kahrl     well, considering there are some bad pointer accesses in minetest that can in theory overwrite anything with anything, any change can potentially break something
18:18 kahrl     if we assume those don't happen I'm quite certain my patch is safe
18:18 kahrl     though I neither proved it nor tested it exhaustively
18:20 Jordach   can we fix MGV7 so that trees grow out of the ground as they do in MGV6
18:20 Jordach   without the need for a schematic
18:21 VanessaE  if 827 goes in, I'd suggest fixing the whole nodeid limit entirely as soon as practical
18:21 VanessaE  or at least, in the news post, make a mention that the "C++ exception" isn't so much a bug as an error message
18:23 kahrl     VanessaE: you mean the message old clients get on your server?
18:24 VanessaE  kahrl: I mean the news post you proposed
18:24 kahrl     well, to be frank, the C++ exception message is a bug
18:24 VanessaE  though if that's fixable too, you might want to at least catch it and...oh
18:25 kahrl     it's the serialization bug I explained yesterday or some time
18:25 VanessaE  right, I know
18:26 VanessaE  but it's only as much a bug as your willingness to allow > 4069 node ids :-)
18:26 kahrl     hmm? how would that change anything?
18:26 kahrl     the limit's there in old clients, can't change it now
18:26 whirm     Hi all
18:26 VanessaE  oh nevermind, I was trying to make a small joke :)
18:27 kahrl     sorry for not getting it :P
18:29 whirm     I'm having a very weird error, I have whole areas of my maps which are completely dark
18:30 whirm     torches work, but its just as if I was underground, the sky is dark, no stars or moon whatsoever
18:31 kahrl     whirm, I think that's more suited for #minetest
18:33 kahrl     once it's known that it's caused by the core and not by some malfunctioning mod, it can be discussed here of course
18:34 whirm     kahrl: I've tried creating a new map with the minimal game and its even worse
18:34 whirm     the whole map is dark
18:35 kahrl     very strange then, I've never seen this happen with just the standard mods
18:35 whirm     with technic_game there are both dark and normal parts
18:35 kahrl     whirm: and you're not using bumpmapping?
18:35 whirm     I've tried with irrlich 1.7 and 1.8 with the same results
18:36 kahrl     try enable_shaders = 0, enable_bumpmapping = 0
18:36 whirm     kahrl: shaders are disabled
18:36 whirm     let me check bumpmapping
18:36 kahrl     shouldn't have any effect if shaders are already disabled
18:36 * whirm   tries just in case
18:37 whirm     nope, exactly the same
18:37 whirm     can a texture pack influence this?
18:38 kahrl     not really unless the textures are all dark... but you said there are lit parts too
18:38 whirm     yes, the textures are ok
18:39 kahrl     does this happen with all mapgens?
18:40 whirm     I had an old map that was OK, but I deleted it and any new I create its like this
18:40 whirm     kahrl: I think so, let me try
18:41 kahrl     on my end, I see lighting glitches like that with the math mapgen but not the others
18:42 PilzAdam  kahrl, I think the mapgen outputs pure dark areas, and plantlife in techinc forces recalc where trees grow
18:42 whirm     v6, v7, indev are completely dark
18:43 whirm     the math one is mostly OK above 0, but dark under water
18:45 kahrl     PilzAdam: this is without mods
18:45 PilzAdam  <whirm> with technic_game there are both dark and normal parts
18:45 kahrl     oh, you mean that
18:45 PilzAdam  just a guess
18:45 kahrl     that would explain it, right
18:46 VanessaE  I would think any of the objects routed through plants_lib would do that, really
18:47 whirm     VanessaE: yes, it looks like its not dark in areas where moretrees trees spawn
18:54 whirm     is there anything else I can do to help debug the thing
18:54 whirm     ?
18:54 kahrl     whirm: I'm a bit stumped, sorry
18:55 kahrl     does this happen with a clean minetest.conf too?
18:55 whirm     kahrl: let me see
18:56 whirm     kahrl: yes, I deleted minetest.conf and its the same
18:56 kahrl     very weird
18:57 kahrl     if you have lots of time, you could try bisecting to find out when this started
18:57 whirm     indeed
18:58 whirm     kahrl: Ill try for a while
18:58 whirm     any point in particular you think I should start from?
18:58 kahrl     not really, but you could try the releases
18:58 whirm     ok
18:59 kahrl     start with 0.4.7 and go backwards
18:59 whirm     I think it was OK at 0.4.7
18:59 kahrl     if it is, bisect between 0.4.7 and now
19:00 kahrl     there were lots of mapgen related commits since then so I can't point at a single one
19:00 whirm     kahrl: ok, I'm on it
19:02 whirm     0.4.7 is fine
19:04 hmmmm     do you think it could be your configuration file
19:05 kahrl     hmmmm: already eliminated that possibility
19:05 hmmmm     it could also be a mod, because mods can set mapgen flags now too
19:06 proller   hmmmm, why NoiseParams nparams_biome_def_heat {50, 50, v3f(500.0, 500.0, 500.0), 5349, 3, 0.70};
19:06 proller   50,50 ??
19:06 proller   average 50c ?
19:06 hmmmm     proller, that's my half-assed attempt to get it ranging from 0-100 with an average of 50
19:06 hmmmm     but it's wrong
19:06 proller   why 0-100
19:06 hmmmm     good question
19:06 proller   -70..60 on earth
19:06 VanessaE  hmmmm: the ramp is, as far as I could tell, something more like -25 to +150C
19:07 kahrl     proller: where is it s stated that it's celsius
19:07 VanessaE  (as defined by snow mod and plants_lib)
19:07 hmmmm     proller, vanessae, like i said it's wrong
19:07 kahrl     -s
19:07 proller   15, 60 - now works good
19:07 VanessaE  hmmmm: I know, just in case you feel like changing it I mean
19:08 hmmmm     i would fix them to have a sane range when i go to actually add in all the biomes
19:09 proller   https://github.com/minetest/minetest/pull/830/files want to commit
19:09 hmmmm     proller, i'd really appreciate it if you didn't go and change the noise params
19:10 proller   no problem
19:11 hmmmm     when you change that, everybody's biome definitions are completely wrong
19:11 hmmmm     that's a great way to break things
19:12 proller   but now no biome defs in git
19:12 hmmmm     that doesn't matter
19:13 hmmmm     there are quite a few mods i can think of that already use it
19:13 hmmmm     even though it's completely undocumented and i never said anybody could...
19:14 proller   then you must change it to more correct
19:14 proller   it will be harder after, when more mods would use it
19:14 hmmmm     erm
19:15 hmmmm     i'm really not sure if weather should be something in the core
19:15 VanessaE  hmmmm: only if it doesn't rain/snow inside a building :)
19:16 proller   its long way to make complete weather
19:16 hmmmm     what's wrong with the weather mod?
19:16 proller   its first small steps
19:17 proller   and in pull- only freeze-melting
19:17 whirm     2e292b67a0a02b045969034c06aaf92b42a83a81 is the first bad commit
19:17 whirm     Add Lua on_mapgen_init callback, and minetest.set_mapgen_params API
19:17 hmmmm     proller,  env->getMap().transforming_liquid_add(p0);
19:18 hmmmm     what happens if the server map is NULL?
19:18 proller   coredump?
19:18 hmmmm     or what if the environment doesn't exist yet?
19:18 proller   but how it can happens?
19:18 hmmmm     this would've been okay if env was still its own object
19:19 whirm     kahrl: ^^
19:19 kahrl     proller: why is minetest.transforming_liquid_add even needed?
19:19 kahrl     it's not used anywhere as far as i can see
19:19 hmmmm     I suppose he wants to use it for the lua portion of his own weather mod
19:20 proller   now - not, can remove for now
19:20 proller   but it can be useful for something related to liquid
19:20 hmmmm     i don't really think these ABMs and what not should be in the core
19:21 hmmmm     something like liquid flowing, sure okay, i can see that
19:21 hmmmm     but where does it end?
19:21 kahrl     hmmmm: agreed
19:21 hmmmm     next someone's going to want to move lava cooling into the core
19:21 hmmmm     and then every time someone wants to add an ABM, where does it go?
19:21 hmmmm     if you start hardcoding things like this, hell, why not hard code door prediction into the client ....
19:21 hmmmm     then it'll be back to the way it was in 0.3.x
19:22 proller   its faster than lua
19:22 hmmmm     yeah... but..
19:22 kahrl     whirm: only way I can see that commit introducing the bug would be if something sets the nolight mapgen flag
19:22 hmmmm     and these are such mundane ABMs, all they do is set a certain node to another node
19:22 whirm     maybe it gets set by default somehow?
19:22 proller   these abms must work with huge amount of blocks
19:23 VanessaE  proller: if something's just too slow but unsuitable for the core, there IS a way to do C++ mods now, kinda-sorta that is (in the form of loadable binary libs, ask RBA, he knows more about that)
19:23 kahrl     whirm: perhaps try grep -Ri nolight .
19:23 kahrl     in the minetest directory
19:25 kahrl     when I execute that it finds matches in minetest.conf.example, src/mapgen_* and some binary files
19:26 whirm     let me rebuild the bad commit
19:26 kahrl     I'm mostly interested in whether there's any lua files that contain nolight
19:26 proller   100000+ blocks of water and ice around me
19:27 hmmmm     proller, it's just that i don't want people to add ABMs as a knee-jerk reaction to a new feature
19:27 hmmmm     like here i guess it's fine because you need to do a lot of setting, but it's just that i don't really like how the core is becoming more... specific
19:28 hmmmm     and you have specific values set for fog/melting temp/freezing temp
19:28 whirm     kahrl: http://paste.debian.net/16061/
19:29 proller   no problem to make adjustable this values, and make lavacooling on it
19:29 proller   but on next steps
19:29 kahrl     whirm: so nolight is enabled in the world "bad"
19:30 kahrl     don't know why but it is
19:30 whirm     kahrl: yes, this is a world created from the first bad commit
19:31 proller   hmmmm, about env->getMap(). - its used in 10+ places in api
19:31 hmmmm     proller, yeah, that's something that should probably be fixed, but we should ask sapier first
19:31 hmmmm     (well, or test it)
19:32 hmmmm     couldn't be arsed to do minetest related things aside from chatting right now, sorry
19:33 hmmmm     whirm, i suppose you *tried out* a mod that turns off lighting and then it got written to map_meta.txt, and you removed the mod but the setting still stayed there
19:33 hmmmm     that could be the problem
19:33 hmmmm     do you guys see now why I didn't want lua to have control over mapgen settings?  :(
19:38 whirm     hmmmm when bisecting I created a new world every step using the minimal game with no mods enabled
19:39 celeron55 to me this sounds like an uninitialized variable
19:39 kahrl     celeron55: my thoughts, but I can't find that variable
19:39 celeron55 (well, sounds until proven otherwise)
19:42 hmmmm     well, it's lua -> map_meta.txt -> world specific config -> global config -> config defaults -> constructor defaults in terms of precedence
19:42 hmmmm     and it's definitely not an uninitialized variable, that's basically impossible
19:44 celeron55 it's valgrind time!
19:45 celeron55 i have no idea what it could tell, but that's why one needs to try it
19:46 hmmmm     erm, a bunch of well-placed printfs would work way better than valgrind ever could
19:46 hmmmm     for this situation in particular
19:46 celeron55 of course; my order of preference starts with valgrind though if i haven't been working on given code in a long time 8)
19:48 celeron55 especially as such random behavior screams to me "memory corruption"
19:51 whirm     it doesnt look random to me, before that commit the maps are ok, after it completely dark, its not random
19:52 hmmmm     keep in mind this is *only* you having the problem
19:52 hmmmm     i'm going to say that it's due to some stealthy mod or an errant setting in the config
19:53 whirm     oh
19:54 hmmmm     "before/after that commit" isn't really significant, we know it's working like it should because you clearly have nolight in your map_meta.txt.  the question right now is how it got there
19:54 kahrl     checked out 2e292b67a0, compiled with -O0, started valgrind, selected minimal, made new world, opened world, waited until I saw some generated blocks, closed game
19:54 kahrl     nothing that stands out in the valgrind log
19:55 kahrl     I mean, there are lots of uninitialized value warnings, but nothing mapgen related
19:55 celeron55 whirm: so is the case that any new world started with a version before that commit does not have the problem, and any world started after it has?
19:56 whirm     celeron55: exactly
19:56 celeron55 also i must say that these new minetest_game sounds are ridiculously bad
19:56 hmmmm     i personally think they're better than cracking eggshells
19:56 celeron55 (i just compiled a new version since... a lot ago)
19:56 hmmmm     but it is still bad
19:57 hmmmm     it's hilarious, we have so many non-technical people interested in contributing to minetest but our sounds and textures suck really hard
19:57 hmmmm     whirm:  did you check any other possible places where mapgen settings might be set?
19:58 whirm     hmmmm: for instance?
19:58 celeron55 sand sounds like you're wealking in some kind of mud, and grass sounds like... nothing from real life
19:58 hmmmm     the gravel step sounds good at least
19:58 hmmmm     whirm, [03:42 PM] <hmmmm> well, it's lua -> map_meta.txt -> world specific config -> global config -> config defaults -> constructor defaults in terms of precedence
19:58 celeron55 also on proper speakers grass actually is so loud on high frequencies that it's completely unenjoyable
20:00 Jordach   uhoh -- all formspecs in master are literally broken
20:01 Jordach   literal clean server running master + master of minetest_game: http://i.imgur.com/5SSt7wJ.png
20:02 Jordach   i'll try a more recent client, see if it fixes it
20:02 whirm     hmmmm: I have no minetest.conf, no mods enabled, using minimal_game and the world is new at each bisect iteration
20:02 hmmmm     what does that last clause of your sentence mean exactly...?
20:03 hmmmm     "at each bisect iteration" ??????
20:03 whirm     hmmmm: I did a git bisect to find out which was the bad commit
20:03 Exio4     proller: u32 get_game_time() { return m_game_time; }
20:04 Exio4     should that function be called getGameTime or something like that?
20:04 hmmmm     whirm, the problem isn't finding the commit "that broke it"
20:04 hmmmm     the cause of this is exactly known
20:04 proller   Exio4, maybe yes, but its ugy make onemore name for var
20:05 Exio4     what?
20:05 proller   game_time GameTime gameTime
20:05 hmmmm     just use the current master, start a new world without any mods that change mapgen flags, making sure your game-specific conf (if one exists) and global conf doesn't have nolight in mgflags
20:05 proller   camel sux
20:05 Exio4     i don't like it but it is a guideline
20:05 hmmmm     proller, we do have code guidelines you know, right?
20:06 proller   ok
20:06 hmmmm     for like a half year it seems
20:06 hmmmm     http://dev.minetest.net/Code_style_guidelines
20:08 proller   fixed
20:11 proller   PascalCase, method names are camelCase., unix_lowercase_underscore_style().   -- 3 types in one gudeline
20:11 whirm     hmmmm: I compiled master, created a new world with no mods loaded, started it once, nolight was set, exited, removed the flag from the config file, deleted map.sqlite, started again, everything is fine.
20:12 whirm     I guess mods can't modify those flags even if I don't specifically enable them, right?
20:12 hmmmm     no, mods are able to explicitly modify any mapgen flags at creation time as they'd like
20:13 whirm     so I should delete the mods folder then I guess
20:14 hmmmm     I thought you said you weren't using any mods...
20:14 whirm     I wasnt using them, but I had them on the mods folder
20:14 hmmmm     try removing them and seeing if that helps
20:20 whirm     empty mods/ dir, only minetest_game on games/, still nolight set when creating a new world
20:21 hmmmm     and you're absolutely sure nolight isn't in a global config file, or something like that?
20:22 hmmmm     by the way, which mapgen are you using?
20:22 kahrl     I think at this point we should try the printf debugging framework
20:23 hmmmm     kahrl, right.  whirm, you don't mind modifying source code, right?
20:24 hmmmm     in server.cpp:811, add printf("=== flags: 0x%08x", mgparams->flags); and add the same thing on line 816, after m_emerge->initMapgens()
20:25 whirm     hmmmm: there's no minetest.conf, where can it be if it's not there?
20:25 hmmmm     it could be in your home directory, hidden maybe, if it's not compiled with run_in_place
20:26 Exio4     ~/.minetest/minetest.conf
20:26 whirm     hmmmm: └──> find ~ -iname "*minetest.conf*"
20:26 whirm     gives no results
20:26 whirm     and its compiled with run_in_place
20:27 whirm     let me modify the shource
20:27 kahrl     it should at least print minetest.conf.example
20:27 hmmmm     maybe he doesn't have the source in his home directory
20:27 whirm     kahrl: its not in my home
20:28 kahrl     ah, okay
20:28 hmmmm     i don't think the install script installs the example
20:28 kahrl     what are the results when you run that find command on the source directory?
20:31 whirm     kahrl: ./minetest.conf.example
20:31 whirm     hmmmm: I'm not installing anything by the way
20:32 celeron55 have you ever played around with the nolight flag anywhere?
20:32 hmmmm     he is indicating that he did not, i'm telling him to just add those printfs so we can see where it is being set exactly
20:33 * whirm   building
20:33 celeron55 well it seems printfs would be the way to go in any case
20:33 kahrl     the game looks for $path_user/minetest.conf, $path_user/../minetest.conf, $path_user/../../minetest.conf
20:34 whirm     celeron55: I did some time ago, but nothing remains from then
20:34 kahrl     path_user is the minetest directory (that contains bin, src, ...)
20:34 celeron55 whirm: as kahrl said, minetest has some sneaky (somewhat legacy) load paths for minetest.conf 8)
20:35 whirm     === flags: 0x0000005f=== flags: 0x0000005f22:35:00: ACTION[main]:         .__               __                   __
20:35 hmmmm     whoops, forgot \n
20:35 hmmmm     but whatever
20:36 hmmmm     so anyway it's being set in one of the config files
20:36 hmmmm     oh hey whirm, copy minetest.conf.example to minetest.conf in the directory and try that
20:36 celeron55 whirm: when you start minetest with a deleted minetest.conf, it creates it to where you suppose it'd be, right?
20:37 celeron55 if that is the case, then you don't have any sneaky config anywhere
20:37 celeron55 if not, then you *have*
20:37 hmmmm     0x5f = 0x40 (nolight) | 0x1f  (trees, caves, dungeons, jungles, biome blend) <-- these are the default flags
20:38 hmmmm     so you can see it's clearly not an uninitialized variable
20:43 whirm     so, it looks for the file on the parent dir of the source?
20:43 kahrl     that and the parent of that
20:44 whirm     fuck, this is where I had backed up the config file
20:44 kahrl     (the latter on in RUN_IN_PLACE builds)
20:44 VanessaE  isn't that a little...well...weird?
20:44 kahrl     only*
20:44 Exio4     i don't use RUN_IN_PLACE builds because i like having ALL the configs in a single place, am i weird? :P
20:45 hmmmm     erm... why would anybody set nolight manually in the config
20:45 VanessaE  hmmmm: a moon map maybe?
20:45 hmmmm     the mod should do that then
20:45 Exio4     nop
20:45 Exio4     +e
20:45 Exio4     nolight should be used only and only, and ONLY for mods
20:45 Exio4     no? :D
20:46 hmmmm     you know something
20:46 hmmmm     it may be handy to include some sort of "spit out config" option to prevent future cases of this
20:46 Exio4     "spit out" what would it do?
20:47 hmmmm     it'd output what the current settings are and what config file it was set in
20:47 Exio4     ah
20:48 hmmmm     for example you'd have:  CONFIG: mg_flags = "blahblah" - set by "../minetest.conf", and then CONFIG: something_other = "whatever" - set by "default setting"
20:48 kahrl     postfix has this and it can be really useful
20:48 VanessaE  hmmmm: absolute paths
20:48 hmmmm     ofcourse absolute paths
20:48 VanessaE  good idea at any rate
20:49 VanessaE  useful for bug reports in general, or if we get some kind of automated bug reporting tool some day like what ff has
20:50 whirm     anyways, looking for a config file in so many places is unexpected, at least minetest should print which config file is it reading the config from, it would saved a lot of time
20:50 hmmmm     whirm, I definitely agree
20:50 hmmmm     also eduke32 could use the same
20:52 hmmmm     thus ends another case of "i told you so"
20:52 VanessaE  at this point, is there any real reason to read it from anywhere except minetest top level, ~/.minetest, or maybe some file in /etc ?
20:52 hmmmm     lol, minetest in /etc
20:53 VanessaE  hmmmm: I've seen stranger stuff placed there :)
20:53 hmmmm     if it were up to me it'd look in ~/.minetest, the current working directory, and that's it - absolutely nowhere else
20:53 VanessaE  yes.
20:53 hmmmm     but this is already an established thing and we need to work with what we're given
20:54 VanessaE  well, current working directory is iffy..
20:54 kahrl     I disagree with current working directory
20:54 VanessaE  cwd might be minetestdir/bin
20:54 VanessaE  or just minetestdir
20:54 hmmmm     yeah, i don't like putting things in bin though
20:54 hmmmm     i mean, come on, it's two files
20:54 hmmmm     it'd make sense if there were 50000 DLLs and what not
20:54 VanessaE  true
20:54 kahrl     someone could place a specially crafted minetest.conf and mod directory in the current working directory that executes some malware lua code
20:55 hmmmm     such is the case for games in windows
20:55 kahrl     don't remember off the top of my head if minetest.conf can modify mod paths
20:56 hmmmm     sapier already had a lot to say about this and attempted to fix the problem but it never took off because it'd break backwards compatibility
21:01 kahrl     I'm going to push #727 now if no one objects
21:01 VanessaE  827*
21:02 kahrl     oops
21:02 sfan5     kahrl: thats not how the "core dev must agree" thing is supposed to work
21:03 sfan5     but #872 sems fine to merge
21:03 Exio4     pilzadam tried and 'agreed' to it
21:04 kahrl     sfan5: celeron55 agreed to it
21:04 sfan5     ah, ok
21:04 sfan5     IIRC there were cases where someone just said "if nobody disagrees [...]"
21:05 Exio4     proller'
21:05 kahrl     sfan5: yeah, without explicit agreement that should only be done for reasonably trivial commits
21:05 sfan5     Exio4: thats what I wanted to avoid saying
21:05 Exio4     or for things that are .. like v7
21:05 Exio4     in that case hmmmm has the right to do it without asking ;P
21:07 Exio4     kahrl, sfan5, hmmmm, shouldn't that be "explicit"? for fixing documentation, trivial-and-simple changes and so on, that "rule" can be avoided?
21:07 sfan5     yeah, probably
21:09 kahrl     yeah, that stuff can be pushed without asking, but sometimes you want to ask regardless
21:10 Exio4     i think for fixing documentation, or cosmetic fixes (like changing comments, small "cleanup" like adding tabs, or so) should be explicited "noted"
21:10 Exio4     i mean, just for noting it in a "paper", it won't hurt anyone
21:11 kahrl     yeah, it can be noted in http://dev.minetest.net/Git_Guidelines
21:16 Exio4     yep :P
21:20 hmmmm     if a change might screw something up or you're changing other peoples' code, really
21:20 hmmmm     like if I'm going to modify the way scriptapi works for example, i'd ask sapier first because he'd know if it breaks anything
21:21 whirm     off to bed now, thanks for the help
21:21 hmmmm     if i were to modify the way the emerge manager works, i wouldn't necessarily ask anybody because i know that code best
21:21 Exio4     the emergemanager guy ;P
21:21 hmmmm     the whole "ask another core dev" "rule" is misguided
21:22 Exio4     yeah
21:23 hmmmm     for neutral code such as mapnode.cpp, a general consensus is needed more or less
21:24 VanessaE  yay!  my server works again :)
21:24 Exio4     haha
21:24 VanessaE  er, wrong channel
21:28 sapier    Taoki I investigated mobf slowness for vombies in current git ... for my system cpu usage drops drasticaly if pathfinding is disabled
21:29 Exio4     if they are dumb they are fast8
21:29 Exio4     s/8/*/
21:29 kahrl     making a News post, does this look fine? http://paste.dy.fi/ko6
21:32 sfan5     kahrl: looks fine to me
21:32 VanessaE  kahrl: I'm a little dubious about my server being singled out :P but seems okay otherwise :)
21:32 kahrl     VanessaE: well, how'd you phrase it? :P
21:33 VanessaE  you had to ask. :)
21:35 VanessaE  just leave those parts out entirely, since you already have a good explanation of what happens anyway
21:35 VanessaE  I already posted an update to my server thread......and ShadowNinja already signed on :D
21:35 sapier    PilzAdam if you want to re mobs in simplemobs while using your core dev privileges to add those features you need to do what mobf worked around due to not having chance to get fixes in quick do it... but don't expect me to do a lot of work in this area. I will maintain and improove mobf as long as there is any chance for it to be usefull ... if there's another implementation of mobs added to core I'll drop mobf ... this is same as micro
21:36 kahrl     VanessaE: well, not every player might look into your server topic
21:36 kahrl     especially if they found your server via the in game menu
21:36 VanessaE  true
21:37 sapier    VanessaE you can but the note in server description
21:37 VanessaE  there are a disturbing number of users who find it that way, even with the server thread link mentioned in the motd.
21:37 sapier    as the desscription is new atm users might tend to look at it for some days
21:42 Exio4     does anyone know an easy way to add a 250ms "forced" latency?
21:43 VanessaE  Exio4: downgrade to 33.6k dial-up? :D
21:43 kahrl     Exio4: loopback or other interface?
21:43 Exio4     kahrl: loopback, yes
21:43 Exio4     VanessaE: my nether already has a 250ms latency to anywhere outside my lan
21:43 Exio4     nether.. i mean network
21:43 kahrl     not easy in general, minetest has an INTERNET_SIMULATOR that could be extended like that
21:44 Exio4     hm
21:44 kahrl     VanessaE: so, should I post my version or without mentioning your server?
21:45 Exio4     i'll just run it on this server and a super-reduced game
21:45 Exio4     (minimal's default only)
21:45 VanessaE  kahrl: it's up to you
21:46 kahrl     Exio4: try this: tc qdisc add dev lo root netem delay 50ms
21:46 kahrl     250ms*
21:46 Exio4     kahrl: iirc it drops packets for adding delay
21:46 kahrl     Exio4: yeah, INTERNET_SIMULATOR only drops packets
21:47 Exio4     is that desired?
21:47 kahrl     VanessaE: it's your server, so your choice
21:47 kahrl     Exio4: dunno, never used it
21:47 Exio4     :P
21:47 sapier    it should be quite easy to add a random nanosleep to receive
21:47 Exio4     i'll just run the server here (a "shell provider" of  friend)
21:48 kahrl     sapier: that's not the same, is it?
21:48 sapier    if we use receive in blocking way this should be almost same (at least for a single client)
21:48 kahrl     you want to only increase latency, your solution would increase latency and decrease bandwidth
21:48 sapier    not for multiple clients of course
21:49 kahrl     it would decrease bandwidth even for a single client
21:49 kahrl     unless I'm misunderstanding
21:49 Exio4     that is why adding client-side prediction is hard! :P
21:49 Exio4     we need more lag!
21:49 sapier    it does but as far as I know we use synchronized client server communication not parallelizing requests
21:51 sapier    but maybe I'm wrong with that synchronized thing
21:51 proller   https://github.com/proller/minetest-mod-weather/tree/weather - now dynamic
21:52 proller   but needs my core and _game patches
21:52 kahrl     sapier: kind of right what you're saying, but every player movement packet would cause a random nanosleep
21:52 sapier    btw I guess those predicion things fail on a simple issue ... pathfinding is expensive ... too expensive
21:52 kahrl     this would add up quickly
21:53 sapier    but isn't that exactly what happens?
21:53 kahrl     a normal network doesn't increase its latency like that
21:54 sapier    if you use synchronized packed handling it does
21:54 kahrl     movement packets are not synchronized afaik
21:54 sapier    if we send multiple packets at once too the sleep should be done where decision to send multiple packets is done
21:54 Exio4     well, other question, where is the chat predicted?
21:55 sapier    ok if there are different cases it's not that simple of course
21:55 kahrl     Exio4: Client::typeChatMessage
21:55 Exio4     where is the "chat_backend->addmsg("<player> message")" :P
21:55 Exio4     kahrl: k, thanks!
21:55 Exio4     was searching in the wrong file
21:57 sapier    guy last call for decision if noone decides what to do about background/overlay thing I'm gonna implement it the way I think it's correct tomorrow
21:58 Exio4     hmm, the client doesn't know anything about the commands managed by the server?
21:59 sapier    and I think celeron55's description of how it should be is more consistent
22:04 kahrl     Exio4: correct
22:33 VanessaE  another issue I'm seeing is the client crashing if you open the inventory while there's too much of the map being received all at once
22:33 VanessaE  probably requires "preload visuals" to be turned off.  May require technic/unified inventory, but it's a client crash, not server.
22:36 sapier    VanessaE I'll implement overlay/background as celeron suggested ... as I wont start until tomorrow evening you may have a last chance to get a consens for your flavor to do it
22:36 VanessaE  if you wait several seconds after connecting before requesting the inventory, it does not crash.
22:36 VanessaE  sapier: eh?
22:36 VanessaE  oh the load order of the menu images?
22:37 sapier    yes I asked for comments 2 days ago I didn't get any celeron wants this you want that and pilzadam wants something similar to what you want
22:37 sapier    that's not anything I can rely on
22:37 VanessaE  well just go with whatever everyone else agrees on.
22:38 sapier    exactly that wont happen I'll do it a last time any further changes wont be my issue anymore
22:38 PilzAdam  sapier, I guess everybody agrees that singleplayer tab shouldnt fall back to default ones
22:38 sapier    exactly you "guess"
22:39 PilzAdam  well, I said so last time and celeron didnt complain about that
22:39 sapier    but only 3 ppl did even comment ant those don't really have a consens
22:39 PilzAdam  he only complained about overlay, and I dont care about that
22:40 sapier    I agree with not falling back to default too that was already in first of celerons suggestion
22:44 sapier    there are only 3 issues on my todo list for formspec menu all others are either major feature enhancements to formspec or things I can't fix ... after those are done I consider formspec menu finished