Minetest logo

IRC log for #minetest-dev, 2017-04-15

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

All times shown according to UTC.

Time Nick Message
00:06 MrIbby left #minetest-dev
00:16 wayward1 joined #minetest-dev
00:25 wayward1 left #minetest-dev
01:17 proller joined #minetest-dev
01:26 paramat joined #minetest-dev
02:56 Tmanyo joined #minetest-dev
02:58 Tmanyo joined #minetest-dev
03:01 Tmanyo joined #minetest-dev
03:41 Puka joined #minetest-dev
04:17 sofar joined #minetest-dev
04:23 kaeza joined #minetest-dev
05:09 sofar bug related question
05:09 sofar does class GUIFormSpecMenu know or have a member to indicate that it is being executed in a MAINMENU context or in a GAME/CLIENT context?
05:10 sofar #5593
05:10 ShadowBot https://github.com/minetest/minetest/issues/5593 -- Game exits when [tab] key pressed in server login page at [user] field
05:10 sofar can fix this bug if I can somehow determine `if (in_mainmenu)`
05:11 sofar m_client == NULL?
05:14 sofar ahhh yes, that actually works
05:16 sofar yay, fixed my own weird bug
05:29 Hunterz joined #minetest-dev
05:40 rdococ hello
05:40 rdococ I'm here. You should go and hide.
05:42 octacian joined #minetest-dev
05:57 YuGiOhJCJ joined #minetest-dev
07:06 nerzhul joined #minetest-dev
07:11 YuGiOhJCJ joined #minetest-dev
07:23 kaeza joined #minetest-dev
07:25 nerzhul merging #4981
07:25 ShadowBot https://github.com/minetest/minetest/issues/4981 -- Partial damage cheat fix: node damages server side by nerzhul
07:44 cx384 joined #minetest-dev
07:45 Krock joined #minetest-dev
08:08 cx384 joined #minetest-dev
09:29 DS-minetest joined #minetest-dev
10:03 nerzhul joined #minetest-dev
10:16 Krock nerzhul, what about https://github.com/minetest/minete​st/pull/5493#discussion_r111555540 ?
10:17 nerzhul Krock, yeah, sorry i forget to answer
10:18 nerzhul on first test when i'm doing this overwrite on m_bgcolor there was a problem, i need to re-test since last changes
10:22 Fixer joined #minetest-dev
10:36 DS-minetest joined #minetest-dev
10:56 celeron55 Fixer: try this https://github.com/celeron55/mine​test/tree/mesh_update_block_cache
11:00 kilbith joined #minetest-dev
11:01 celeron55 or wait for a moment, i'm fixing a thing
11:03 celeron55 ok now
11:17 Fixer celeron55: will try but little later @ dinner
11:17 Krock ^ throws a weird MSVC error about converting "const v3s16" to "size_t" is not possible with "CachedMapBlockData *"
11:18 Krock will look into this one deeper
11:18 celeron55 wut?
11:18 celeron55 well i suppose it tells you a line number
11:20 Krock I suppose it's something with the cpp11 containers: https://pastebin.com/raw/GBub9E1x
11:20 DI3HARD139 joined #minetest-dev
11:21 Krock actually thrown in mesh_generator_thread.cpp
11:25 Krock looks like it can't generate a hash from v3s16 as map key o.0
11:30 nerzhul yes
11:30 nerzhul std::unordered_map doesn't support v2xxx and v3xxx as they are not hashable containers celeron55
11:31 Krock fixed by replacing with std::map
11:35 celeron55 oh, oops; switching to std::map
11:37 Krock but it works.. so far
11:43 celeron55 it's a performance improvement; what you should be looking for is the almost complete disappearance of stutter from adding tasks to the mesh generator
11:44 Krock yes, I'm testing it right now
11:44 celeron55 there are other kinds of stutter, i don't know if people have learned to notice that exact kind of stutter :P
11:44 celeron55 this is a reasonable test config to make it more apparent:
11:44 celeron55 max_simultaneous_block_sends_server_total = 100
11:44 celeron55 max_simultaneous_block_sends_per_client = 100
11:44 celeron55 mesh_generation_interval = 0
11:45 celeron55 (on a high end system)
11:46 celeron55 also if someone makes a windows build, throw that to me, i can test on my media pc that doesn't have a compiler or anything
11:46 Krock okay, will upload
11:50 Krock https://www.dropbox.com/s/bpwfpqwohdnn79o/mine​test_170415_1c851e0_mesh_update_block_cache.7z
11:51 Krock 32-bit and std::map patched manually, so it's the most recent version
11:59 Krock ah, increasing the minimal view range helps.
12:02 celeron55 anything that causes more stuff to be copied to the mesh generator will exaggerate the difference
12:05 celeron55 which is, anything that causes the server to send more mapblocks
12:06 celeron55 it can be masked by mapblocks that have heavy use of node metadata though as it's able to cause a different stutter that can be much higher than the copying-mapblocks-for-meshgen one
12:08 Krock http://i.imgur.com/X51IW7M.png local world
12:09 Krock but there weren't so many processed blocks in the new one. will have to do more testse
12:09 Krock *tests
12:12 Krock but overall feels smoother now
12:33 Krock Pushing this fix in 15 minutes: https://pastebin.com/raw/qN2A6ZE5
12:39 lisac joined #minetest-dev
13:04 Fixer celeron55: i'm back, will look now
13:09 Fixer celeron55: don't see much difference, default config, spawned near jungle, stuttered like crazy https://i.imgur.com/QgP4bSx.png (tested without std::map commit)
13:11 Fixer will try std::map version
13:16 celeron55 it doesn't look like what i would expect, are you sure you built the right version?
13:24 Fixer celeron55: is not krock's screenshot also suggests there is tons of stutter anyway
13:24 Krock right, there's also sometimes no sleep time for the main thread
13:26 Krock but what counts is the speed of copying the map data
13:34 Fixer tried krock's build, right away in forests observed very visible shitty stuttering https://i.imgur.com/p2Z26RI.png
13:34 celeron55 Fixer: actually, looks like in that screenshot basically all of the block receive related stutter is gone, not counting the single big spike
13:35 celeron55 that shows the same thing
13:36 celeron55 there's both _other and _draw stutter left that isn't related to the toclient_blockdata handler
13:36 celeron55 what CPU and GPU does this system have?
13:37 Fixer core i3 2120 3.3ghz + 8gb ram + amd hd6870 1gb vram
13:38 Fixer 0.095 ms mainloop_dtime... karl
13:38 Fixer mainloop_draw*
13:38 Fixer i have no idea how it helps since game stutters like crazy anyway
13:38 Fixer default config
13:39 Fixer 0.095 sec*
13:39 celeron55 basically the same cpu as my media pc then (this has 2100)
13:42 celeron55 yeah, i can see the stutter i set out to get rid of is gone
13:42 celeron55 and that there are at least 2 other stutters left
13:53 celeron55 the stutters left in mainloop_other don't correlate with anything that's currently graphed, the stutters left in mainloop_draw seem to partly correlate with # processed meshes and partly not with anything that is currently graphed
14:32 proller joined #minetest-dev
14:44 YuGiOhJCJ joined #minetest-dev
14:51 celeron55 https://github.com/minetest/minetest/pull/5595
15:06 kaeza joined #minetest-dev
15:12 Krock poor travis got a lot to do right now :/
15:13 paramat joined #minetest-dev
15:22 paramat Zeno` are you +1 for #5519 ? i moved the temporary old move code to the end of the file to make things cleaner
15:22 kilbith why the dominant male ain't allowed to push straight ahead if he wants to
15:22 ShadowBot https://github.com/minetest/minetest/issues/5519 -- Sneak: Add option for old move code by paramat
15:22 celeron55 kilbith: why would i want to; these nice people will catch my errors so i don't need to
15:22 celeron55 lol
15:23 kilbith I remember hmmmm enforcing the rules on you
15:24 kilbith you need some praetorian guards to defend your rights of dictator IMO
15:25 Krock we'd get nowhere if our dictator could do what he wants :P
15:25 celeron55 well someone would try enforcing them for sure, and if i made any errors it'd be kind of bad
15:25 nerzhul lol the dictator era is finished in minetest (not on earth, we have three WW3 competitors)
15:30 xerox123 joined #minetest-dev
15:31 Krock duh, if the order of including our header files matter, then something's going wrong with the structure
15:31 celeron55 yeah i added an initial commit that adds the include into util/thread.h
15:31 celeron55 util/thread.h was broken
15:34 celeron55 should be fixed now
15:38 paramat celeron55 since you are concerned about the issue, are you able to +1 #5519 ?
15:38 ShadowBot https://github.com/minetest/minetest/issues/5519 -- Sneak: Add option for old move code by paramat
15:41 paramat or maybe just a concept review
15:42 troller joined #minetest-dev
15:44 paramat nore ^ any opinion on that?
15:44 nore paramat: need to think about it
15:45 paramat ok thanks
15:46 celeron55 i think the concept is fine
15:47 celeron55 the only thing left from that in the long term is the byte added to the protocol
15:52 paramat hmm from the extra physics override
15:52 paramat i had not considered that
15:53 kaeza joined #minetest-dev
15:53 paramat new sneak code is going well, we now have all ladders working as before, we have 2 node sneak jump, and basic sneak behaviour only has minor issues left
15:56 celeron55 (the code is fine to me also)
15:56 celeron55 are we doing these partial reviews nowadays?
15:56 nerzhul what do you mean with partial review ?
15:57 celeron55 there's no actual rule for how to take into account something like someone saying "+1 for concept", or is there?
15:57 celeron55 i wrote about it a long time ago but i guess it ended up as a suggestion
15:57 nerzhul no rule, concept != quality
15:59 celeron55 if someone says +1 (implying everything) and then three others say "+1 for concept", "+1 for code", "tested to work fine", i guess it sums up as two full +1s
16:00 celeron55 except that then you can forget to check the protocol or something
16:00 celeron55 do we even need 2 +1s for anything else than concept
16:04 celeron55 i'd suggest 2 +1s should be required for concept and persistent data (being network protocol and disk formats), and for everything else one +1 is enough
16:05 Remy_ +1
16:05 nerzhul celeron55, you mean less +1 ? and what about coredev ?
16:06 celeron55 i mean all of those from coredevs
16:07 nerzhul it's already the case for coredevs,
16:07 nerzhul 2 +1 for external , 1 for coredev PR
16:07 paramat no concept rule is not an official thing, i was just asking your opinion on the concept (since you probably don't have time to review fully)
16:08 paramat sorry i mean 'concept approval'
16:09 paramat a +1 is only official if it is a full review: code and concept
16:10 celeron55 really the only reason i ask is because personally i like to review concepts and protocols
16:10 paramat and a core dev's approval of their own PR counts towards the 2, which helps
16:11 troller joined #minetest-dev
16:15 * Fixer .oO(minetest stone soup)
16:46 nerzhul okay guis #5475 is now ready and has bidirectional migration, like map
16:46 ShadowBot https://github.com/minetest/minetest/issues/5475 -- WIP: player data to Database by nerzhul
16:46 nerzhul guys*
16:47 nerzhul the only missing thing is a little bit more documentation on minetest wiki linked into client
17:07 nerzhul who manages wiki.minetest.net accounts ?
17:07 Calinou nerzhul: me
17:07 nerzhul Calinou, can you create me an accouhnt ?
17:08 Calinou nerzhul: sure, PM me your e-mail address and desired username
17:08 Calinou (note that username must start with an uppercase character)
17:11 YuGiOhJCJ joined #minetest-dev
17:12 nerzhul i PM you
17:50 ShadowNinja The first Minetest developer meeting is in 10 minutes.  :-)  nore, sofar, celeron55, sfan5, paramat: you said this time worked for you, are you available today?  nerzhul, Zeno`: I didn't hear from you, please tell me if this time (18:00 UTC) works for you.  All devs: please join -staff for private discussion if you have not joined already.
17:50 nore ShadowNinja: I'm there
17:51 kilbith it's 3:50 AM where Zeno` lives
17:52 * sofar yawns
17:52 sofar I thought it was 50 minutes ago
17:52 ShadowNinja kilbith: Oh, I suppose he won't be able to participate then. :-|
17:53 kilbith and can you please explain why the development discussion has to be that opaque ?
17:53 nore I just rebased two of the pulls https://github.com/Ekdohib​s/minetest/tree/pull-3299 and https://github.com/Ekdohib​s/minetest/tree/pull-3651 btw
17:53 nore one of these is completely untested (the build one), the other only had a quick testing though
17:54 ShadowNinja kilbith: What do you mean?  The -staff part?  That's mainly just for stuff like the meeting notes (they're editable to anyone with the URL).
17:55 Fixer minetest-dev feels very b-rocratic
17:55 kilbith yeah, I don't know it has to be private, there are people actually living on MT and they need to anticipate from what has been decided in upstream
17:56 nore ShadowNinja: isn't there a way to have a public url that isn't editable?
17:57 sofar brt
17:57 ShadowNinja nore: Not with the service that I'm using (AFAIK), but we could use something else.  Any ideas?
18:00 ShadowNinja Good evening and welcome to the first edition of the Minetest weekly(?) developer meeting.
18:00 ShadowNinja The purpose of this meeting is to quickly reach a decision on a number of PRs at a time when enough developers are available to quickly approve/deny them.  This meeting is intended to be fairly fast-paced.  If we decide to merge a PR it will be assigned to someone and should be merged shortly after the meeting.  If a PR warrants an extended discussion we'll delay it until after the meeting.  We're going to focus on the engine for now.
18:00 ShadowNinja First order of business: #1188  I don't see any reason why this shouldn't be merged, although it does require some rebasing.  Any objections to merging?  If not, any volunteers to do the rebase and merge?
18:00 ShadowBot https://github.com/minetest/minetest/issues/1188 -- getTime refactoring by Selat
18:00 nore it was a good PR, so +1 for merging from me
18:00 nore I can try to rebase it
18:02 sofar careful
18:02 sofar I rewrote a bunch of gettime.h
18:02 sofar so there's conflicts for the core parts
18:03 ShadowNinja nore: Alright, great.  Moving on...  #3580 / #4664  (alternate implementations) Merge or close?
18:03 sofar but, yes we should merge what we can from this
18:03 ShadowBot https://github.com/minetest/minetest/issues/3580 -- Added delayed shutdown feature by pinkysnowman
18:03 ShadowBot https://github.com/minetest/minetest/issues/4664 -- Implement delayed server shutdown with cancelation by nerzhul
18:04 nore I would say concept-wise, it is an useful feature
18:05 sofar normally I'd agree that we want mods to do things where applicable, but this is still something I don't mind seeing in c++
18:05 sofar doesn't help that hmmmm and est31 are no longer here :)
18:05 sofar and Zeno`'s alseep
18:05 nore I guess we could rebase #4664 and merge it probably
18:05 ShadowBot https://github.com/minetest/minetest/issues/4664 -- Implement delayed server shutdown with cancelation by nerzhul
18:05 nore (as it seems more complete than 3580)
18:06 sofar if nerzhul rebases I'm fine with it
18:06 paramat here
18:06 sofar but I'm looking at others more, I was already ok with it before
18:08 ShadowNinja IMO nrz's implementation is overcomplicated.  The simple Lua implementation should work just about as well, right?
18:09 nore ShadowNinja: it doesn't allow cancelling
18:09 ShadowNinja nore: True, but that could be added in Lua.
18:09 nore ShadowNinja: yes, but it's not easy because it uses minetest.after
18:10 nore I mean it can be done but most of the PR would have to be rewritten
18:11 ShadowNinja Yeah, I guess I just like the simplicity of the Lua solution.  But if you all prefer the C++ impl, we'll do that.
18:11 ShadowNinja It seems we're in agreement to merge at least one of them though?
18:11 nore it looks like so, yes :)
18:12 nore next item on the list then, I guess?
18:12 ShadowNinja paramat, sofar: opinions on which to use?
18:13 paramat no opinion
18:13 sofar nrz's version, since he's here to redo it
18:13 paramat are we going through old PRs oldest first?
18:13 ShadowNinja nerzhul: Will you rebase and merge your PR?
18:14 ShadowNinja paramat: Yep.
18:14 ShadowNinja (more-or-less)
18:14 celeron55 i like the C++ one more
18:15 paramat ah you're going by least recently updated, i was thinking oldest by number
18:15 celeron55 (altough, it's almost criminal how wasteful duration_to_string is about lines of code use)
18:17 ShadowNinja Alright, nerzhul's taking too long, I'll assign the PR to him for now.  Next: #2085 -- Merge or close?
18:17 ShadowBot https://github.com/minetest/minetest/issues/2085 -- Disable font scaling on non-Android platforms by cheapie
18:18 nore is it still relevant or not?
18:18 nore I mean, is font scaling still there?
18:19 ShadowNinja I'm not sure.
18:21 ShadowNinja The PR is old and it seems font-scaling is working nore-or-less fine now, so I vote close.  Does that sound good?
18:22 nore hmmm, with quick testing (xrandr --scale 0.5x0.5, where I get text ~2 bigger), it looks like there's no problem
18:22 nore so I would say close as well
18:22 celeron55 i don't think making formspec behave differently on different devices is going to make anything better to anyone
18:22 paramat i'm fine with close
18:23 nore (interestingly, with --scale 2x2 I get text twice smaller, but some elements don't scale... well)
18:23 celeron55 it would be nice to add a proper reference to this log in the closing messages
18:24 ShadowNinja Alright, done.  #3651: Seems ready to merge, just needs someone to rebase(trivial) and merge.  Any volunteers?
18:24 ShadowBot https://github.com/minetest/minetest/issues/3651 -- Disable android leveldb by default by MoNTE48
18:24 celeron55 otherwise nobody will remember even after a few days
18:24 nore already rebased it, but untested
18:24 nore https://github.com/Ekdohib​s/minetest/tree/pull-3651
18:24 ShadowNinja celeron55: I'll add it in.
18:25 nore I can't test however, but this is a rebased version of it
18:25 ShadowNinja nore: Well, since you've rebased it I think in makes sense for you to test/merge it.  Is that fine?
18:25 paramat could you open a new PR then? so we can close this one?
18:26 nore ShadowNinja: as I said, this is android build and I can't test it
18:26 celeron55 i would trust MoNTE48's choices on android
18:26 nerzhul oh it's now ? :)
18:26 nore that's why it is completely untested
18:26 nerzhul sorry for to be late then
18:27 paramat others can test the new PR
18:27 sofar Wayward_One ?
18:27 nerzhul yes i can rebase 4664 without doubt
18:28 ShadowNinja nore: O.K., uh, anyone else to test it?  (if not, I'd just say go for it)
18:29 nore --> opened #5596 as paramat requested
18:29 ShadowBot https://github.com/minetest/minetest/issues/5596 -- Disable android leveldb by default by Ekdohibs
18:29 paramat PR opened. so close 3651?
18:29 nerzhul paramat, yes
18:30 paramat will do
18:30 paramat done
18:30 ShadowNinja Alright, nore: the Travis build will probably do.  I'll leave you to merge it as soon as it's ready.
18:31 ShadowNinja Next: #1439  -- Reabse and merge or close?
18:31 ShadowBot https://github.com/minetest/minetest/issues/1439 -- Add configurable key bindings for selecting next/previous hotbar item and changing the volume by Wuzzy2
18:31 nore Didn't est31 said he could still help with android builds btw?
18:31 nerzhul travis deosn't test android build
18:31 nerzhul nore est31 left us
18:32 nore nerzhul: I know, but he said he could still help with something I seem to remember was android building
18:32 nore but anyway
18:32 nerzhul i do the android builds for release, i can test it, but not know i'm rebasing the delayed shutdown pr
18:32 nore I like 1439
18:32 ShadowNinja I'll leave nore to figure it out, I'd like to keep moving.  :-)
18:33 nore it's simple enough, so I'd just say rebase & merge
18:33 nore however, paramat thought more keybindings were not needed
18:34 nore depends on what others want then
18:35 Hunterz joined #minetest-dev
18:35 paramat 1439 i think there's a new PR that covers this?
18:36 celeron55 i think 1439 is fine, altough maybe the defaults should be nothing
18:36 ShadowNinja There are complaints that the keybindings aren't very flexible, but that's really just an issue with our input system, not this PR in particular.
18:36 celeron55 as most people probably won't use them
18:36 Player_2 joined #minetest-dev
18:36 paramat i think there's a newer PR that allows keybindings for hotbars of any size, up to 23
18:36 celeron55 ShadowNinja: the PR is good for going to that direction
18:37 celeron55 it adds more things to bind to, which is needed anyway
18:37 celeron55 as it's useless to have a better system for binding things without anything to map inputs to :P
18:37 celeron55 any actions, i mean
18:38 ShadowNinja Alright, I support merging.  I don't care whether they're enabled by default sonce they're configurable.
18:38 paramat i still don't think these are very useful, next/previous when we have number key access
18:38 celeron55 well you can always think of disabled people or something
18:39 paramat volume is not much needed in-game and can be controlled at hardware level
18:39 nore paramat: some people think they are useful, and it causes no problem to have more keys to bind
18:39 celeron55 if mouse wheel becomes freely bindable some day, then that's probably the code it's bound to
18:39 ShadowNinja paramat: Well, it doesn't work with extended hotbars (>10 hotbar elements).
18:39 ShadowNinja (it = number keys)
18:39 celeron55 (and the previous mouse wheel code is removed)
18:39 paramat there's another PR for extended hotbars
18:39 nore paramat: let's say I want to mute only minetest
18:40 ShadowNinja paramat: It's fine if there's more than one way to do it, they're just defferent input styles that different people will prefer.
18:40 paramat yeah mute could be useful
18:40 nore #4929 indeed, but the two PRs are not mutually exclusive
18:40 ShadowBot https://github.com/minetest/minetest/issues/4929 -- Make direct item selection keys freely bindable by Wuzzy2
18:41 paramat so, i'm not stringly opposed, more neutral
18:41 nore also, maybe you only want to bind 10 keys because your keyboard hasn't an infinite number of keys, so the other keys for next/previous are still useful
18:41 ShadowNinja Anyways it seems like everyone but paramat is in favor of merging, correct?  Are there any volunteers to merge?
18:41 nore hmmm, I can try to rebase
18:43 paramat *strongly, not stringly :]
18:43 ShadowNinja nore: Alright.  Maybe someone else can step up for the later ones... :-)
18:44 nore heh, I already rebased #3299 ;)
18:44 ShadowBot https://github.com/minetest/minetest/issues/3299 -- Added afternoon/morning red tint by kaadmy
18:44 ShadowNinja Next: #4035  -- ~~Needs rebase~~, merge or close?
18:44 ShadowBot https://github.com/minetest/minetest/issues/4035 -- Additional hooks for inventory manipulations callbacks. by Foghrye4
18:44 ShadowNinja er...
18:45 paramat 3299 strongly opposed
18:45 ShadowNinja Lets do 4035 first, it _does_ need a rebase.
18:46 paramat foghrye4 has been inactive in MT since Nov
18:47 nore 4035 is interesting, it allows to track *every* item movement
18:47 nore I don't think that's possible without it
18:47 ShadowNinja paramat: Yeah, but if someone else can fix it up we could merge it.
18:48 nore i.e. you can't even track movement of items within a player's own inventory I think?
18:48 paramat i know :] added adoption needed, didn't mean that to mean close it
18:48 nore I vote for rebasing
18:49 nerzhul please note #4664 is now rebased and mergeable, i tested it :)
18:49 paramat i meant, author not likely to rebase it
18:49 ShadowBot https://github.com/minetest/minetest/issues/4664 -- Implement delayed server shutdown with cancelation by nerzhul
18:49 nore (& merge)
18:49 celeron55 4035 is massive, i wonder if anyone has done any kind of code review
18:49 sofar it touches a lot of parts that nerzhul has modified
18:49 nore yeah, that's the problem with it
18:49 sofar hence the conflicts, perhaps nerzhul can take a look at it?
18:50 nerzhul huge callback pr
18:50 nerzhul i don't have time to rebase this PR as i already have player to db pr as a huge pr :p
18:50 nerzhul and i am finishing the chat protocol rewrite, it's nearly complete
18:51 nerzhul (for better CSM/SSM integration with chat
18:51 nore okay, maybe it can wait a bit though -> leave it open and rebase it some time later, except if someone else can rebase it?
18:51 ShadowNinja Before we rebase it, are we in agreement that the concept is good, regardless of the current implementation/code quality?
18:52 nore +1 from me
18:52 celeron55 to me, 4035 is fine in concept (but a lot of potential for bad code)
18:52 paramat 4035 seems potentially useful
18:52 nerzhul it needs some cleanups, i don't like the GameScripting interface passed to inventorylist
18:53 nore ^ yeah, same, the concept is good but I don't like that point of the current implementation
18:53 ShadowNinja Alright, I'll put it on hold for the next meeting, avaiting someone to adopt it and clean it up a bit.
18:54 nerzhul maybe we should have some singleton in core at a moment to have better code interfaces, for example a singleton for the GameScripting and CSM will permit to call callbacks without adding those things, and these are singleton by essence
18:54 nerzhul i see he prepared CSM part, it's interesting to take it, but we need more time
18:54 celeron55 no; the codebase is designed to work in a situation where you have multiple Servers for whatever weird reason
18:54 nerzhul celeron55, very strange situation :p
18:55 celeron55 but IGameDef is designed for fetching server- or client-wide singletons
18:55 celeron55 when you need something like that
18:55 nerzhul it's designed for but not used as is
18:55 celeron55 well, it can be added for that purpose to anywhere if needed, that's its purpose
18:55 nerzhul imagine: ClientScripting::get_instance()->on_use(blah);
18:55 nore I guess it could be useful for things such as server redirections or the like
18:56 ShadowNinja Let's keep moving please.  Next: #3299: Already rebased by nore, merge or close?
18:56 ShadowBot https://github.com/minetest/minetest/issues/3299 -- Added afternoon/morning red tint by kaadmy
18:56 celeron55 some complex testing might want to compare two server instances or whtaever
18:56 celeron55 whatever*
18:56 nerzhul celeron55, except unittests i don't see where we can have a such thing
18:56 celeron55 it's a design decision i made and we should stick with it
18:56 nore paramat says he's strongly opposed to 3299 but didn't explain why yet
18:57 paramat explained in thread
18:58 nore hmm
18:58 paramat has issues and i feel should be in shaders
18:59 nore bottleneck of meshgen -> not at all, the bottleneck is not there but in getTileInfo
18:59 ShadowNinja It looks like it will cause color changes in an underground bunker with no natural light,
18:59 nore however yes, there's the second problem
18:59 nore maybe it should wait for the change of lightbanks meaning
18:59 paramat ok bottleneck comment not valid :]
19:00 ShadowNinja The effect is nice, but I think it needs a better implementation.
19:00 nore ShadowNinja: note the new implementation is even simpler: https://github.com/Ekdohibs/minetest/commi​t/aedc81bf7b3999e9f166dad69d343821fbdce173
19:00 paramat it's level of fanciness means best in shader code
19:00 nore (it's literally just modifying the function giving the sunlight color)
19:01 paramat and at least needs to be optional
19:01 ShadowNinja nore: That works the same and doesn't affect only artificially-lit areas?
19:02 nore ShadowNinja: unfortunately it does as lightbanks meaning was not changed yet
19:03 nore which is something we need to change I would say, as this would allow setting the sunlight color
19:04 ShadowNinja Alright, I'll mention in the natural/artificial light thread and close it for now.
19:05 nore ah no, it should work in most cases with artificially-lit areas
19:05 nore if it's only artificially-lit it should work
19:06 nore however if it's both artifically-lit and sunlit there won't be red tint
19:07 ShadowNinja nore: Still buggy, so my position's the same.
19:08 ShadowNinja We're done with PRs for now.  sofar mentioned that he's be willing to host a mumble server if we wanted to use it for future meetings.  What are your thoughts on that?
19:08 paramat even if it doesn't affect artificial light, i think it needs to be in shaders and optional
19:08 paramat eh? i was just getting into this
19:09 nore well, we need to change lightbanks meaning, but if we do so, it would be easy to set sunlight color as a texture to fetch from so it would be easy to make it optional, just change the texture
19:09 sofar we should keep it to an hour, I need to pack boxes and feed kids atm
19:09 paramat IRC is fine, why use elsewhere
19:10 nore I don't really want to use voice if possible, but I can put up with it if needed
19:10 sofar voice chat may help to keep the pace up, but, everyone should be OK with it
19:10 sofar so, irc is ok for now
19:11 celeron55 some of us (like me) aren't native english speakers; it would probably just slow things down
19:11 nerzhul except if you want to hear my french english accent i prefer IRC :p
19:11 paramat voice chat =/ pace up is the problem, writing helps with thought
19:11 celeron55 i and nerzhul would slow it to a standstill trying to say words 8)
19:11 paramat IRC has logs too
19:11 ShadowNinja Voice would probably be faster, but not everyone's comfortable with it and we have non-native speakers, so I think we should pass on that.
19:12 nore same as celeron55 & nerzhul, I have a problem of parsing what others say as well when speaking, it's easier when written :)
19:12 sofar alright then
19:12 nore hmmm maybe try to find a time for the minetest_game meeting
19:12 nerzhul can we finish the remaining points ?
19:13 sofar I have to leave
19:13 nerzhul if we don't sofar will be ate by his kids
19:13 ShadowNinja Next: C++11?  It's supported by basically everything now and we already use some of its features.
19:13 nore there are less devs active for it though (sofar, paramat, sfan5 and myself I guess?)
19:13 nerzhul ShadowNinja, you forgot #5451
19:13 paramat (i wouldn't participate in voice chat anyway :] too shy)
19:13 ShadowBot https://github.com/minetest/minetest/issues/5451 -- Rename doc/*.txt to *.md. by ensonic
19:13 celeron55 to me, 3299 is fine in concept but it needs code review AND i want to take an artistic look about it as i've kind of designed the sunrise/sunset stuff
19:14 ShadowNinja nerzhul: Oh, you're right.  We'll do that right after C++11.
19:14 celeron55 or, i mean, not really code review but an optimization review
19:15 nore celeron55: you can test the rebased version if you want, but anyway, maybe changing light banks meaning first would be good
19:15 nore it would fix a lot of problems as well
19:15 nerzhul for C++11 it's time to switch, compilers are a little bit late, norm was released in 2011, and GCC 4.9 and Clang 3.6 have all C++11 features which can be required by minetest
19:15 nerzhul the only problem i need to test is android build, which can be a problem
19:15 nore note that the android build scripts seem to use GCC 4.9 now anyway
19:15 nerzhul two years ago NDK didn't permit C++11 i should test with last NDK
19:16 ShadowNinja Any objections to switching to C++11?
19:16 nerzhul nore the problem is not GCC for android but NDK
19:16 nore nerzhul: oh, ok
19:16 celeron55 can we try disabling non-C++11 compilers and just seeing how many complaints we get, before actually starting to use C++11?
19:16 celeron55 like, a week
19:16 nerzhul no no, wait for release :)
19:17 nerzhul Debian 8 doesn't have C++11 compiler
19:17 nerzhul after release, Debian 9 will release (on August or September) with a C++14 compiler
19:17 nore hmmm, maybe wait until that time then
19:17 nerzhul nore, for NDK, some stdlib parts was missing on my last test in Android
19:18 nerzhul but two years ago on Android 4.1 :p
19:19 ShadowNinja O.K. I guess we'll delay it for Debian 9.
19:19 nore but I'm for switching to C++11 as soon as it is released
19:19 nerzhul everybody is okay for switching to C++11 after MT release ?
19:19 nerzhul (June)
19:19 celeron55 can we all agree debian isn't really doing any benefit to open source projects by being like that when closed source projects have been using C++11 since 2011
19:19 ShadowNinja I have to go now, so I'll let celeron55 take over.  #5451 is next.
19:19 ShadowBot https://github.com/minetest/minetest/issues/5451 -- Rename doc/*.txt to *.md. by ensonic
19:20 nerzhul see you ShadowNinja :) maybe have a shortest list next time
19:20 nerzhul celeron55, nore okay for that after 0.4.16 ?
19:20 nore nerzhul: ok for me
19:21 paramat can't judge but no objection
19:21 nerzhul paramat, constexpr can benefit on your code parts in client :)
19:21 nerzhul or lambdas
19:21 celeron55 ok for me at whatever point debian 9 is released; i don't think that is set in stone
19:22 paramat 5451 next? weird order :]
19:22 nerzhul yeah #5451
19:22 ShadowBot https://github.com/minetest/minetest/issues/5451 -- Rename doc/*.txt to *.md. by ensonic
19:22 nore hmm, both are fine with me for 5451
19:22 nerzhul in CSM we started to wrote it using md, because it was new and permit to have only documentation on github
19:23 nerzhul md are easily readable on vim (with support md syntax) if needed
19:23 kilbith it's too fancy (tm)
19:23 celeron55 do we think windows users need to be able to read the README without extra software or not?
19:24 paramat hard on the eyes for me, and can't link to single lines
19:24 celeron55 the issue is, regular windows is quite handicapped with almost everything
19:24 nerzhul windows user doesn't use console, in notepad++ it's supported too, and in some IDE too (like clion and VS with a plugin)
19:24 celeron55 yeah i mean we have users that aren't developers, have you noticed
19:25 nerzhul will a non developper read a lua developper API ?
19:25 paramat client lua api should have been .txt for consistency
19:25 celeron55 wait, this  doesn't touch README
19:25 celeron55 it's ok to me then, lol
19:27 paramat the MD formatting of he docs needs improving
19:27 paramat *the
19:27 nore hmmm, I must go now, but I don't think there's anything left to discuss anyway
19:28 nerzhul seems we don't have any consensus :)
19:28 nore (I meant except that and I'm neutral on it :p)
19:29 Calinou celeron55: I don't think our userbase is even knowledgeable enough to open a README :/
19:29 Calinou I don't think they need to read it anyway
19:29 paramat Krock has issues with it too
19:29 Calinou it's mostly information about compiling
19:31 celeron55 maybe there should be a second readme that would be used in compiled windows packages
19:33 celeron55 so does anyone actually object 5451?
19:37 nerzhul not really in both cases
19:39 celeron55 we're leaving it hanging around collecting better ideas then i guess
19:39 nerzhul yes
19:39 Calinou most people don't read READMEs anyway (except really curious people like me :P)
19:40 nerzhul for the last point, clang-format just delay it on next meeting when more users will be present
19:40 Calinou if we have official Windows installers ready for prime-time, we can show an information window, this is very easy to do
19:40 Calinou (why do we still lack installers? .zip for Windows should be the secondary installation option for portable uses, definitely not the primary one)
19:40 Calinou like… most software today
19:41 nerzhul celeron55, can you review #4664 i rebased it :)
19:41 ShadowBot https://github.com/minetest/minetest/issues/4664 -- Implement delayed server shutdown with cancelation by nerzhul
19:41 nerzhul (and tested it)
19:45 celeron55 do we translate any server chat messages?
19:45 nerzhul it depends
19:45 nerzhul with the chat protocol i'm designing it can be possible more easily
19:45 nerzhul as we have a proper distinct between chat, announces, system messages and raw messages
19:46 nerzhul then announce and system can be catched by gettext to be translated :)
19:46 celeron55 but currently?
19:46 nerzhul no
19:46 paramat i object 5451 i thought that's obvious
19:46 nerzhul the only server message which are translated are standard denied messages  (not custom)
19:46 celeron55 paramat: yeah i forgot to acknowledge that
19:47 paramat :]
19:47 paramat Krock too :]
19:49 nerzhul i updated http://wiki.minetest.net/Database_backends with postgresql, 1 year after merging not too late :p
19:50 Calinou (updated page, added a missing "s")
19:51 Sokomine_ joined #minetest-dev
19:53 celeron55 nerzhul: does /shutdown -1 cancel a shutdown?
19:54 celeron55 it doesn't really handle that very well; i think it says "<name> shuts down server" and cancels the shutdown 8)
19:54 celeron55 and is not documented in params
19:55 celeron55 wait, no, it says nothing
19:55 celeron55 should be in params as it's a useful feature
19:56 Taoki[mobile] joined #minetest-dev
19:56 Sokomine joined #minetest-dev
20:04 paramat Krock nerzhul any opinions/objectons (not necessarily reviews) for #5519 ? i'm considering merging it later
20:04 ShadowBot https://github.com/minetest/minetest/issues/5519 -- Sneak: Add option for old move code by paramat
20:04 Krock looks like I've missed quite much here. However, I'm neutral on that pull
20:05 paramat ok
20:07 Krock nerzhul, can't believe that postgres wasn't added yet. What do you think about replacing http://dev.minetest.net/CMake_Options with a README link? That page is potentially quite outdated
20:07 Krock and a duplicated documentation too
20:12 kilbith_ joined #minetest-dev
20:17 halt_ joined #minetest-dev
20:32 nerzhul Krock, as we maintain readme seems lgtm to replace it with link yes
20:32 nerzhul paramat, sorry i'm not sneak aware :(
20:32 paramat no problem
20:33 nerzhul celeron55, sorry for to be late, yes negative value cancel it, i'm adding documentaiton
20:33 nerzhul and the server announce cancel is shutdown :)
20:34 paramat i'll merge 5519 later if no objections
20:38 nerzhul paramat, please factorize a little code with existing code
20:40 nerzhul celeron55, thanks for the review, i'm waiting for travis and will merge
20:42 paramat the duplication is intentional for various reasons
20:43 nerzhul there is no reason to duplicate code here as it's very very similar, don't do this
20:44 paramat there is
20:44 nerzhul explain with good arguments ?
20:45 paramat see thread comments
20:46 paramat also, the code will be removed at the end of 0.4.16, a separate block of code makes it easier to remove
20:46 celeron55 yeah it's explained already
20:47 paramat if it was permanent i would agree with you of course
20:47 kaeza joined #minetest-dev
20:47 nerzhul i don't understand, it's a temporary commit which will be removed before release ?
20:48 paramat at end of 0.4.16
20:49 paramat so at 0.5
20:51 Krock what about "readU8"? what does it return when the end of the stream is reached? (i.e. old server)
20:52 Krock looks like it's 0, which would make it to default true, which is intended
20:54 celeron55 maybe the boolean should be added as a u8 bitfield so as not to waste 7 bytes at a time for each new boolean value...
20:54 celeron55 7 bits*
20:55 Fixer Calinou: portable option is essential too, since a lot of people distrast installers those days (because of crapware/adware/malware included)
20:55 Fixer zipped one i mean
20:56 celeron55 do you know what's probably the stupidest thing ever? the way the MT client unloads mapblocks and then the server just sends all of the unloaded ones again: http://i.imgur.com/ysOslwd.png
20:57 celeron55 i think the thing i'm doing next is ripping off the mapblock sending code from the farmap PR
20:57 Krock >.<
20:58 celeron55 i want my PR reviewed and merged though, i'm going to want to build these on top of each other
20:59 celeron55 nobody dared to +1 it yet
20:59 Krock same feeling on my PRs
20:59 nerzhul which PR celeron55 ?
20:59 nerzhul i also have the same feeling but we don't work on same parts :p
20:59 Krock #5595
20:59 ShadowBot https://github.com/minetest/minetest/issues/5595 -- Cache MapBlock data in the mesh generator thread for less stutter by celeron55
20:59 nerzhul i started to review it , as you see :p
21:00 nerzhul the idea is excellent i was thinking about exactly same concept when the Fixer complains appear
21:01 nerzhul celeron55, what is the cache memory usage ? i think about android build
21:03 paramat i'll test 5595
21:04 celeron55 nerzhul: it quite directly depends on the view range
21:04 celeron55 what range does android generally use?
21:05 Fixer nerzhul: it is not a silver bullet though, you can hardly see much difference since it stutters like crazy anyway %)
21:05 celeron55 at 50 the cache seems to peak at something like 8 MB
21:05 nerzhul active_block_range 1 viewing_range 25
21:06 Fixer nerzhul: it kinda fixes one stutter but elefant stutters remain
21:06 Calinou Fixer: it's essential but shouldn't be the promoted option
21:06 Calinou it's not convenient for newbies at all
21:07 celeron55 nerzhul: generally on the client the most memory hungry thing is mapblock meshes; the base mapblocks don't really use a lot of memory even if all of them were left in the cache forever
21:07 nerzhul okay
21:07 celeron55 well, 4kB each
21:08 celeron55 wait
21:08 celeron55 i mean 16kB
21:11 celeron55 being a cache it does exchange memory for lessened processing though, obviously
21:15 celeron55 my best guess for viewing_range=25 is 3 MB
21:16 celeron55 does it need to conserve memory more aggressively?
21:17 nerzhul no it seems reasonable
21:17 sofar we hundreds of mb of ram elsewhere
21:18 sofar that needs to go on a diet
21:18 sofar not this :)
21:19 nerzhul sofar, did you get some time ? :)
21:19 nerzhul #4664 merged
21:19 * sofar eats lunch, needs to run again
21:19 ShadowBot https://github.com/minetest/minetest/issues/4664 -- Implement delayed server shutdown with cancelation by nerzhul
21:19 Fixer minetest 32bit has problems staying below that limit
21:20 nerzhul sofar, when you get time i finished the player db backend, for using it you should use --migrate-players options in server
21:20 celeron55 with large ranges it can take like 50 MB
21:21 nerzhul maybe we should use sqlite3 by default on new worlds for players ? and let existing worlds migrate ?
21:24 celeron55 i just flew around with a range of 320 and got the cache to 90 MB... i think i'll do something to this, it shouldn't need _that_ much in any use
21:44 thePalindrome Think we could get #2727 for 0.4.16? (Just asking "can this be done without breaking things")
21:44 ShadowBot https://github.com/minetest/minetest/issues/2727 -- Sound pitch
21:44 DI3HARD139 joined #minetest-dev
21:45 nerzhul thePalindrome, i don't know, there are many things to finish and present in the milestone
21:49 thePalindrome Alrighty, I figured, just more wondering if it would break compatability
21:50 Megaf joined #minetest-dev
21:50 nerzhul i don't think
22:00 VanessaE celeron55: you need to tell github that your fork of minetest != the official upstream, btw
22:01 VanessaE nevermind
22:01 VanessaE the website's just being obscure.  as usual.
22:14 VanessaE anyway your PR does help some.
22:18 halt_ joined #minetest-dev
22:19 Grandolf joined #minetest-dev
23:09 Fixer joined #minetest-dev
23:11 nore #5597
23:11 ShadowBot https://github.com/minetest/minetest/issues/5597 -- getTime refactoring by Ekdohibs

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