Minetest logo

IRC log for #minetest-dev, 2017-08-27

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

All times shown according to UTC.

Time Nick Message
00:58 ssieb joined #minetest-dev
01:21 Tmanyo joined #minetest-dev
01:33 Fritigern joined #minetest-dev
02:28 Lunatrius` joined #minetest-dev
03:19 DI3HARD139 joined #minetest-dev
05:17 fwhcat joined #minetest-dev
05:30 Fritigern joined #minetest-dev
05:34 Fritigern joined #minetest-dev
06:12 Hunterz joined #minetest-dev
06:30 fwhcat joined #minetest-dev
08:10 DI3HARD139 joined #minetest-dev
08:57 Megaf joined #minetest-dev
09:13 YuGiOhJCJ joined #minetest-dev
09:27 nore #6324
09:27 ShadowBot https://github.com/minetest/minetest/issues/6324 -- Statbars: fix incorrect half-images in non-standard orientations by Ekdohibs
09:44 Megaf joined #minetest-dev
09:45 Megaf joined #minetest-dev
10:20 Fixer joined #minetest-dev
10:29 behalebabo joined #minetest-dev
11:10 lisac joined #minetest-dev
12:32 paramat joined #minetest-dev
13:01 rubenwardy merging #6315 in 5 minutes
13:01 ShadowBot https://github.com/minetest/minetest/issues/6315 -- More readable Default Controls in README by lupoDharkael
13:03 nore rubenwardy: ok
13:09 paramat wow number of issues dropped, great work
13:10 nore #6324 should kill another issue
13:10 ShadowBot https://github.com/minetest/minetest/issues/6324 -- Statbars: fix incorrect half-images in non-standard orientations by Ekdohibs
13:11 rubenwardy mergin...
13:13 rubenwardy paramat, I closed like 50 the other day
13:14 paramat yes thought it was you :D
13:15 paramat i'll try to sweep through them again sometime
13:36 Lunatrius joined #minetest-dev
14:16 paramat is #4912 actually fixable in MT?
14:16 ShadowBot https://github.com/minetest/minetest/issues/4912 -- Pixels flicker on nodes when using Minetest without shaders
14:18 rubenwardy Irrlicht issues our usually our issues
14:18 rubenwardy except GUI and Input
14:20 Fixer <YuGiOhJCJ> hello, on SONOF Raspberry Pi server, with Minetest 0.4.16 and Irrlicht 1.8.1, my client often crashes after playing some minutes: https://pastebin.com/QzujXh7Q do you know why?
14:23 sfan5 Fixer: probably because the GL driver on the pi is experimental
14:23 sfan5 if you want properly running mt on a pi you'll need to fight with the gles/egl stuff
14:51 fwhcat joined #minetest-dev
15:13 Bobr joined #minetest-dev
15:31 fwhcat joined #minetest-dev
15:36 paramat please can we close #6322 ?
15:36 ShadowBot https://github.com/minetest/minetest/issues/6322 -- Collaboration with Terasology project?
15:38 sfan5 i'd keep it open for now
15:39 juhdanad joined #minetest-dev
15:49 paramat people can still discuss it if its closed :]
15:50 sfan5 no that's dumb
15:51 rubenwardy leave it open for now
15:51 juhdanad There is place for collaboration. For example allowing the other project to use some algorithms from there.
15:52 rubenwardy juhdanad, that's a good point. Worth suggesting that
15:53 juhdanad I also thought they could help making our Android port better, but they don't have an Android version.
15:54 sfan5 their heavy shaders won't work on any mobile GPU anyway
16:04 paramat and they're Java
16:04 rubenwardy algorithms translate across languages though
16:05 paramat yes, i was referring to Android
16:05 rubenwardy oh ah
16:05 rubenwardy well, our Java part is very very thin
16:05 rubenwardy it's an NDK application, so most of the Android features are in C++
16:05 paramat now the author is attacking me, good grief
16:07 paramat kilbith joining in, unsurprisingly
16:10 sfan5 paramat: shaders are not written in java
16:11 Krock joined #minetest-dev
16:11 sfan5 not that it matters any for interoperability
16:11 paramat anyway, i guess it could be left open for a while, but it seems like a waste of time, if a dev of either project wants to work with the other project they will, it can't be forced, discussion seems pointless
16:12 paramat ruben me too, but we're doing ok
16:13 rubenwardy I agree it's mostly pointless, but immediately closing does give a bad impression. The guy is trying to help
16:13 sfan5 ^ this
16:13 rubenwardy one of the big worries I have about Minetest is us becoming stagnated - lots of good developers have left recently (hmmmm, est31, T4im, etc)
16:14 rubenwardy so I feel that there is a need to make development easier by cleaning, simplifying, and decouplying code
16:14 rubenwardy and also documenting more
16:15 rubenwardy but as always, there's too much to balance
16:18 paramat i agree
16:18 paramat i'm ok with it being open for a while :]
16:19 celeron55 i've never worried whether minetest will do well or not; it's simply not a goal or a non-goal to me
16:19 celeron55 it's kind of superficial
16:19 juhdanad joined #minetest-dev
16:20 paramat it's doing very well, better than ever
16:20 paramat we just don't have a large number of devs, so stuff moves slowly
16:22 rubenwardy the project is currently doing pretty well
16:22 Krock merging #5445 and #6318 in 10 mins
16:22 ShadowBot https://github.com/minetest/minetest/issues/5445 -- lua API for deregistering single biomes by zeuner
16:22 ShadowBot https://github.com/minetest/minetest/issues/6318 -- core.get_objects_inside_radius: Omit removed objects by HybridDog
16:22 rubenwardy but that's mostly due to paramat and nerzhul
16:23 paramat i'm concerned about T4im, there was a sudden stop in his Github activity while he was enthusiastically involved, hope he's ok
16:24 nore can someone review almost-trivial #6324 btw?
16:24 rubenwardy i noticed that too
16:24 ShadowBot https://github.com/minetest/minetest/issues/6324 -- Statbars: fix incorrect half-images in non-standard orientations by Ekdohibs
16:26 celeron55 developers leaving isn't strictly a bad thing, it gives a lot of flex to development in a loosely lead project like this and it's much easier to set new directions... like 0.5.0
16:26 paramat i asked lhofhansl if he was still ok for being a dev but have not had a response
16:26 rubenwardy devs leaving isn't a problem as long as it's not due to a "bad" reason
16:26 paramat and juhdanad is another one to keep in mind
16:26 rubenwardy bad meaning our fault
16:26 rubenwardy paramat, either of those are fine by me
16:27 nore paramat: hmm btw are you still against #5687 now since we are breaking backwards compatibility for 0.5?
16:27 celeron55 it's worth trying to understand why each person left of course
16:27 ShadowBot https://github.com/minetest/minetest/issues/5687 -- Artificial bank by juhdanad
16:27 paramat lhof said he was ok with it at one point before, i just wnated to check again
16:27 Krock nore, will test.
16:27 nore Krock: thanks
16:27 paramat i still don't like 5687
16:28 Player_2 joined #minetest-dev
16:28 celeron55 rubenwardy: do you suppose there was a big fault from MT's part?
16:28 paramat such a change in map format needs more justification
16:28 rubenwardy I'm not sure, celeron55
16:28 juhdanad 5687 has a critical bug: conversion from old to new format sometimes creates dark map blocks. I don't know the cause, any help is appreciated.
16:29 celeron55 OC mixed things so much that i'm not sure i could pinpoint anything for this year
16:29 paramat it wasn't MTs fault for hmmmm, est or zeno
16:29 juhdanad I tried to document my functions well, that should help.
16:29 rubenwardy I think hmmmm was getting disillusioned with code quality etc
16:29 rubenwardy but predominately it was due to lack of time, I believe
16:29 nore juhdanad: hmm, will look
16:30 paramat hmmmm lost interest, didn't play the game, he said he felt 'the rug was being pulled out from under him'
16:30 celeron55 i think hmmmm just learned to be lazy after being a bit young and naive at first
16:31 paramat i recall he said he has his own ideas of how to code a voxel game, bu tsaid he has no time to start his own
16:31 paramat he also said he didn't enjoy dev anymore
16:32 paramat i suspect my presence in a small way made it easier for him to leave because i can (just) keep mapgen going
16:36 paramat also i don't think 0.5.0 makes 5687 any more acceptable, because we are not breaking map format
16:38 nore paramat: but we are breaking backwards compatibility anyway
16:38 paramat it's just that the stated advantages seem a bit 'meh' to me, useful but not enough for such a fundamental change :]
16:38 paramat not for map format
16:38 nore well, you know how much light updates are costly though
16:39 nore they'll get half as costly in some cases
16:39 paramat yes that's the most useful point
16:39 nore and it also provides a lot of advantages, both for modding and for visuals
16:39 rubenwardy is there a convertor?
16:39 nore rubenwardy: yep
16:40 nore basically it recomputes the lighting
16:40 paramat well, those other advantages are very minor
16:40 juhdanad rubenwardy: If merged, Minetest will automatically convert and save the loaded map blocks in the new format.
16:41 juhdanad paramat: think of it as maximising the information content of param1.
16:42 paramat but, maybe i will be convinced, i just feel it needs more work and more consideration
16:42 paramat with a map format change obviously we have to be super-cautious
16:43 nore juhdanad: btw, is there a comment in the issue about the problem you're currently having with it?
16:44 paramat btw juhdanad if asked would you be prepared to be a MT dev?
16:46 YuGiOhJCJ joined #minetest-dev
16:47 shivajiva for the record Zeno specifically expressed it wasn't OC that made him come to a decision...
16:51 YuGiOhJCJ sfan5, GL is only used on the client side isn't it? so why it matters to run the server on a pi with gles/egl instead of GL driver?
16:51 shivajiva like it or not how mt is developed and the behaviour of the people involved were factors. Not a single person but an accumalation of things
16:51 paramat the graphics of 5687 need work too, which is intended to be done though, obviously this is not good https://github.com/minetest/minetest/pull/5687#issuecomment-298822185
16:52 rubenwardy shivajiva, did he express in IRC? Do you know the date?
16:53 rubenwardy http://irc.minetest.ru/minetest-dev/2017-08-08
16:53 Krock s/ru/net/
16:53 sfan5 YuGiOhJCJ: i thought you were running the client on a pi
16:54 YuGiOhJCJ oh ok that's not the case, I am running on desktop, the server is on a Raspberry Pi
16:55 sfan5 the issue is quote hard to pinpoint without a `bt full`-style backtrace and a mt+irrlicht debug build
16:55 nore btw http://irc.minetest.net/minetest-dev/2017-08-08#i_5036407 <-- do you know anything about that? never heard anything
16:55 rubenwardy I think he means we're developing MT only for MultiCraft's gains
16:55 rubenwardy ie: "hey he pays well" is sarcastic
16:55 rubenwardy but unsure
16:56 nore hmmm possible indeed
16:56 YuGiOhJCJ sfan5, ok so I will not report this problem on GitHub until I get a better trace with debugging information
16:56 sfan5 YuGiOhJCJ: sounds good
16:57 celeron55 finally commented to #5687; should have done this much earlier but was lazy
16:57 ShadowBot https://github.com/minetest/minetest/issues/5687 -- Artificial bank by juhdanad
16:58 sfan5 "should have done this much earlier but was lazy"
16:58 sfan5 me irl
16:58 Krock ^ me, irl
16:58 sfan5 i still want to rewrite the formspec system
16:58 sfan5 but >effort
16:58 rubenwardy agreed
16:59 celeron55 you'll get a lot of fame for it
16:59 celeron55 i mean, like a blog post
16:59 celeron55 totally worth!
16:59 Krock could be done using Json, which can be extended easily
17:00 rubenwardy the problem isn't the format, but how it works
17:00 paramat i disagree with what zeno wrote in dev on that day, however there is this which gives a different impression http://irc.minetest.ru/minetest-hub/2017-08-08#i_5036596 that seems a better explanation of why he left
17:00 sfan5 Krock: what with the love for json? are lua tables not good enough
17:00 rubenwardy ie: not responsive, doesn't scale well, co-ordinates are inconsistent, basically no client-side prediction support, no containers
17:00 sfan5 what's*
17:01 juhdanad joined #minetest-dev
17:01 Krock sfan5, well, surely you could serialize Lua tables but JSON offers the feature to bring it into a nice readable text form
17:02 rubenwardy they're basically identical, except for the fact that Lua doesn't differentiate between dicts and arrays
17:04 sfan5 rubenwardy: how would you design containers in formspecs?
17:04 rubenwardy you can't very well, which is why you also need a new format to support it
17:04 rubenwardy but we do have face containers
17:04 rubenwardy *fake
17:04 sfan5 i dont mean with the existing format, how would you approach it with an entirely new format
17:05 nore merging #6324 now since the travis is ok
17:05 ShadowBot https://github.com/minetest/minetest/issues/6324 -- Statbars: fix incorrect half-images in non-standard orientations by Ekdohibs
17:06 juhdanad paramat: I think I would like to become a developer, but I won't be able to put more time to the project than currently.
17:08 rubenwardy similar to common graphics libraries. Components such as tabs can have children, one for each tab page. These children would be one of several types of containers: linear, tabs, canvas. But it's quite complicated, so it's worth investigating GUI libraries for this
17:12 paramat juhdanad there's no time obligations, we have core devs who disappear for months or years :]
17:12 paramat you just work when you feel like it here
17:13 paramat often i feel someone has to be crazy to be a core dev, luckily crazy people exist :]
17:13 lisac joined #minetest-dev
17:16 paramat i sometimes think, why do i do this? i must be crazy, then i remember i am crazy, so it's ok :}
17:21 paramat i do think 5687 should be worked on and considered, because if it turns out to be worth it, before 0.5 would be the ideal time to do it. i'm not saying 'i will always oppose this', just that it needs work and more consideration of whether it is justified
17:21 shivajiva rubenwardy: you seem to have the right points in the log
17:23 paramat also, lighting should be visually identical to how it is now
17:26 juhdanad I remember how you took care to keep visuals identical when I first introduced hardware colorization too.
17:53 paramat yeah, i made a fuss :}
18:49 behalebabo joined #minetest-dev
19:03 paramat closed #6321
19:03 ShadowBot https://github.com/minetest/minetest/issues/6321 -- On systems with lower ram, minetest lags going large distances in flat mapgen, uses up most of the system ram and swap space
19:08 paramat wow issues + PRs down to 950, less than 1000 again
19:09 Krock thanks to rubenwardy
19:13 paramat for 0.5 we could aim for < 100 PRs and < 1000 total
19:16 rubenwardy my aim was < 100 PRs by the end of august
19:16 rubenwardy may be pushing it now
19:16 rubenwardy mid september ?
19:17 juhdanad joined #minetest-dev
19:17 rubenwardy as for #6321, I reckon maybe the "solution" is to detect how much RAM there is left, and drop some blocks if so
19:17 ShadowBot https://github.com/minetest/minetest/issues/6321 -- On systems with lower ram, minetest lags going large distances in flat mapgen, uses up most of the system ram and swap space
19:19 rubenwardy then again, they seem to be moving in a straight line, so blocks should be unloaded
19:20 behalebabo joined #minetest-dev
19:21 cx384 joined #minetest-dev
19:22 celeron55 hmm
19:23 celeron55 the farmap-without-farmap branch has a feature that is able to unload blocks very fast, accurately and efficiently in situations like that
19:24 celeron55 i guess i need to turn it to a PR and rebase for 0.5
19:24 rubenwardy please do :)
19:25 celeron55 i mean... not really "a feature" but it just swaps the old system out and puts a new system in
19:25 rubenwardy Does it improve view range / chunk sending at all?
19:25 celeron55 it manages those better
19:26 celeron55 rendering speed doesn't change
19:28 celeron55 altough, more accurate block sending _is_ a prerequisite for rendering improvements that focus on range
19:28 celeron55 and also for saving memory
19:29 celeron55 hmm... accurate might be a bad term; maybe more like systematic
19:30 rubenwardy I wonder how easy it would be to put together a benchmarking tool for Minetest
19:30 rubenwardy so we can objectively check whether performance has reduced during a release cycle
19:31 rubenwardy rendering / graphics is hardly my area though. It's all black magic
19:32 celeron55 well the only thing you can really do (and should?) is to load up a world with the player in the same position, wait for it to be loaded and check fps
19:33 celeron55 that could be automated
19:33 rubenwardy if CSM could access the client's FPS and jitter, you could automate that
19:33 rubenwardy well
19:34 rubenwardy that would be the easiest way, in C++ is also possible
19:34 celeron55 C++ patches are kind of a pain as the code changes all the time
19:35 celeron55 CSM would be a backwards compatible interface so it would work well for the purpose
19:35 rubenwardy yeah, that's why I suggest CSM
19:37 nerzhul joined #minetest-dev
19:38 celeron55 if there's some way in config or on the command line to force loading a client-side mod, it's just a matter of writing a mod that runs the test, outputs the result somewhere and closes the game, and writing a script that builds many minetest versions and runs them with the mod and the same world
19:38 celeron55 https://github.com/celeron55/minetest-worldtest
19:39 celeron55 this was a similar test for testing minetest world compatibility between versions
19:40 celeron55 that would probably still work just fine if patches for new versions where written in rules.d/
19:40 celeron55 maybe it should still be used?
19:54 rubenwardy could be good. Means that if you change the benchmark script, you don't need to manually run all previous versions
19:54 rubenwardy and in the right place in game.cpp, a patch could be easily port able
19:59 paramat celeron55 this might interest you https://forum.minetest.net/viewtopic.php?f=50&amp;t=18355 a subgame remaking MT 0.2
20:03 celeron55 lol, it sounds like this guy hates what he's doing
20:03 celeron55 junglegrass not turning grass to dirt? heresy!
20:03 celeron55 it's the result of my careful and artful design
20:05 celeron55 i think the 0.2.20110922 actually already differs from my original stuff a bit too much
20:05 celeron55 something earlier from that year is better
20:06 celeron55 like, before github
20:08 celeron55 i think 2011-07-23 is pretty much the pinnacle of minetest-c55
20:15 lisac joined #minetest-dev
20:17 Fixer joined #minetest-dev
20:58 proller joined #minetest-dev
21:05 paramat joined #minetest-dev
21:08 proller joined #minetest-dev
21:43 Lone-Star joined #minetest-dev
21:43 Lone-Star when i build from source, how can i make a deb from it?
21:50 YuGiOhJCJ joined #minetest-dev
21:50 Lone-Star left #minetest-dev
21:57 paramat closed #2205
21:57 ShadowBot https://github.com/minetest/minetest/issues/2205 -- Android - Drowning bubbles don't always show.
22:15 paramat any support for me adding a setting for #6101 ? before i start work
22:15 ShadowBot https://github.com/minetest/minetest/issues/6101 -- Do not use signed integer for registering nodes to allow more than 32768 registered nodes
23:02 paramat long overdue #6326
23:02 ShadowBot https://github.com/minetest/minetest/issues/6326 -- Advanced settings: Make 'controls', 'graphics', 'sound' more accessible by paramat
23:12 paramat luckily settingtypes makes it easy and quick to move stuff around, far better than messing around with formspecs
23:25 Fixer_ joined #minetest-dev
23:47 YuGiOhJCJ sfan5, I now have a full trace of the problem: https://pastebin.com/m5G5016n (irrlicht and minetest are built with debugging information)
23:47 bigfoot547 joined #minetest-dev
23:48 YuGiOhJCJ do you see something useful?
23:48 YuGiOhJCJ in this output?

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