Time  Nick        Message
01:15 paramat     trivial docs update #4651 will merge later
01:15 ShadowBot   https://github.com/minetest/minetest/issues/4651 -- Lua_api.txt: Clarify 'override_meta' bool in 'set mapgen setting' by paramat
03:28 paramat     damage sounds for consideration game#1336 sofar bbl
03:28 ShadowBot   https://github.com/minetest/minetest_game/issues/1336 -- Default: Add damage sounds by paramat
04:01 sofar       paramat: not bad!
04:45 paramat     i'll merge #4573 #4651 in a moment
04:45 ShadowBot   https://github.com/minetest/minetest/issues/4573 -- Damage flash: Reduce maximum alpha. Avoid fade overload by paramat
04:45 ShadowBot   https://github.com/minetest/minetest/issues/4651 -- Lua_api.txt: Clarify 'override_meta' bool in 'set mapgen setting' by paramat
04:51 paramat     merging
04:53 garywhite   So paramat I know that 4.14 came out in May, so my guess is the next version will come out in spring 2017?
04:54 paramat     around Dec 21st
04:55 paramat     so it can be a solstice / xmas thing
04:55 paramat     merged
05:23 garywhite   night all
05:24 * garywhite wanders off to bed
11:11 _Megaf      Hi folks
11:12 _Megaf      est31: do you know if Rogier uses IRC?
11:12 _Megaf      He is doing a great work with the compression thing and other things too
11:13 est31       _Megaf: he's here only very rarely, if I recall correctly
11:13 est31       yes, rogier is doing great work
11:13 _Megaf      That last 30% increase in speed got me, that was unexpected
11:14 _Megaf      Anyhow, I'd like to be able to choose the compression lib for my server
11:14 _Megaf      we can choose the db backend, so why not choosing the compression lib to?
11:15 _Megaf      Every server has it's different needs
11:15 est31       because it needs to be baked into the protocol
11:15 est31       every client must ship with all compression libraries
11:15 est31       essentially we must bundle all compression libraries we allow in the source code
11:15 est31       it isnt as simple as with db backends
11:15 _Megaf      Oh
11:15 _Megaf      I see
11:16 _Megaf      with db is easier because clients don't need them
11:16 est31       yes
11:17 nrzkt       est31: i did the requested changes on the PR and answer on the getPlayerSAO() point
11:17 _Megaf      ok, so let's just get the fastest that don't increase the db size.
11:21 Megaf       "@est31: I did some more investigation using perf and strace, and it turns out the malloc library is actually the culprit! It will repeatedly munmap() a free()d 5MB area of memory, only to mmap() it again on the subsequent malloc() in the zstd code. The original code even mmap()ed and munmap()ed 2 areas of 5MB and one of 2MB for every compression operation... Every time again, at least one of the 5MB areas is zeroed with memset(). How's that for a
11:21 Megaf       way to avoid having to turn on the heater when it's cold :-)"
11:21 Megaf       That explains a lot on low power CPUs
11:54 Fixer       it is not mine, just look how chat is done, just add entry field below chat text output and scroll and you are done
12:01 sfan5       est31: right now it looks like that *IF* we replace zlib it would be a good idea to go with brotli because of reduced size, what do you think?
12:51 est31       sfan5: yes, currently it looks like brotli is the best option
12:51 est31       but it would be great if we could test dictionaries before that
12:51 est31       whether they give a benefit or not
12:52 est31       (and whether zstd then performs better than brotli then)
12:56 nrzkt       sfan5: i don't agree space is a good point except for network reasons, depending on the diff. The most important thing is to serialize/deserialize mapblocks fast with a good compression ratio, but not the best
12:57 Calinou     LZ4 seems the most appropriate here
12:57 Calinou     disk space is cheap, very cheap, CPU is definitely not so
12:57 Calinou     and bandwidth is relatively affordable in general
12:57 nrzkt       yes, i also think LZ4 is the best approach, it consume a little bit more space but CPU overhead is just ridiculous
12:58 Calinou     LZ4 is ridiculously fast :p
12:58 PR4530lives Calinou: some admins will disagree
12:58 est31       O.o getting github unicorns
12:58 PR4530lives Calinou: it is pain to backup a large world
12:58 Calinou     the worlds won't be larger by that much
12:58 nrzkt       why ZFS has LZ4 ondisk compression + gzip + lzo instead of brotli and zstd ?
12:58 nrzkt       PR4530lives: LZ4 world size only increase ~5-10%
12:58 Calinou     https://facebook.github.io/zstd/
12:58 PR4530lives Calinou: bandwidth is important for many since they are running server on their machinesz
12:58 Calinou     LZ4 reads at 3,100 MB/s
12:59 Calinou     PR4530lives: no, the majority of medium/large-sized servers run on VPSes or dedis, which have unmetered bandwidth (or several TBs of metered bandwidth)
12:59 Calinou     home servers are kind of fading away, since the price of VPSes is going down for better specs and service
12:59 PR4530lives Calinou: minetest servers?
12:59 Calinou     yes
13:00 PR4530lives i have opposite impression
13:00 nrzkt       i use LZ4 as a compression method on my server to store mapblocks into postgresql and there is not decompression/compression penalty, my CPU was very very lightweight now
13:00 PR4530lives people still prefer home hardware
13:00 nrzkt       PR4530lives: take the serverlist, whois each server and verify :)
13:00 PR4530lives it is probably in between
13:00 PR4530lives nrzkt: actually a good idea
13:01 Calinou     yeah but I get the feeling most servers are in datacenters, not in people's homes
13:01 Calinou     especially if you exclude servers with a low amount of players on average
13:01 nrzkt       decompression of non loaded mapblocks or compression to save mapblocks take ~10% used CPU in a running server. LZ4 reduce this to 2-3%
13:02 PR4530lives Calinou: i believe both just test, just test ii and xanadu with lots of players are on home hardware
13:02 nrzkt       to be precise, when i did my last tests on LZ4 on my server with valgrind, the most consuming process after ABM was GZIP compression
13:02 Calinou     yeah but they have a good amount of bandwidth
13:02 Calinou     they're not hosted on a low-end ADSL line, I believe
13:02 nrzkt       with LZ4 i didn't see LZ4 compression in the top 20 actions performs by the server
13:03 est31       LZ4 takes more than double as much space
13:03 nrzkt       this is wrong est31
13:03 nrzkt       did you test with very heavy mapblocks or empty ?
13:04 est31       https://cloud.githubusercontent.com/assets/6202053/19588218/853eeb94-9764-11e6-8738-8db608984707.png
13:04 est31       https://github.com/minetest/minetest/issues/4495#issuecomment-255305550
13:04 Calinou     ah
13:04 Calinou     is this really a problem though? I figure people host servers on HDD machines, and can afford the added storage
13:05 Calinou     SSDs are also getting larger, although some low-end VPSes stick with 20-40 GB SSDs (but you shouldn't host a large server on those...)
13:05 sfan5       >and bandwidth is relatively affordable in general
13:05 sfan5       no it isn't
13:05 nrzkt       est31, Calinou: my db size with 100% generated mapblocks only uses 15GB
13:05 Calinou     sfan5: it's unmetered on my VPS here :)
13:05 sfan5       dont think about your vps
13:05 sfan5       think about mobile users
13:05 nrzkt       for mobile users CPU is the principal problem
13:05 Calinou     do we care about people playing on crap mobile networks? I think it's a worthless idea
13:05 sfan5       and don't come citing your mobile plan with 20GB you pay 10€/mo for
13:05 Calinou     it is doable to play on 4G, but other than that, just stop it
13:05 sfan5       people in brazil dont have that shit
13:05 nrzkt       not the BW as the 4G is everywhere now
13:06 Calinou     we are not developing a game for mobile clones with ads.
13:06 Calinou     we are developing a game for us :)
13:06 sfan5       4g is not everywhere either
13:06 Calinou     playing a game on 3G is asking for trouble
13:06 sfan5       okay then go on and delete the android port
13:06 sfan5       since we don't need it apparently
13:06 Calinou     except if it's chess or something like that
13:06 nrzkt       in fact our mobile experience is just bad in fact as we don't have proper controls for android :p
13:06 nrzkt       but maybe some clones have modified it heh
13:07 Calinou     also for what it's worth, Minetest is pretty lightweight on bandwidth compared to Minecraft.
13:07 sfan5       and that justifies making the mobile experience even worse?
13:07 Calinou     so we can double our bandwidth requirements, we should still be below Minecraft. (or on a similar level)
13:07 est31       so, lets make it worse
13:07 Calinou     nrzkt: first person games on a phone are an oxymoron
13:07 est31       because minetest has to be worse than minecraft, no?
13:07 Calinou     I'd consider them remotely playable on a 10" only
13:10 PR4530lives Calinou: i've looked most popular servers, and it is evident that most popular once a on home hardware
13:10 nrzkt       Calinou: i think we should use the accelerometer for mobile users with minetest to permit them to move in the world xD
13:10 PR4530lives are*
13:10 PR4530lives Ukraine is rolling out 4G only now, and that East Europe for you
13:11 PR4530lives most of servers with lots of players are on home hardware it seems
13:11 PR4530lives bandwidth is important for some of them
13:11 PR4530lives since many ISPs got SHITTY upload rates
13:12 Calinou     I could play Minetest perfectly with 0.3 Mb/s upload back in 2012 :P
13:12 Calinou     (as a client)
13:12 Calinou     also, isn't MapBlock sending, as a client, mostly downloading?
13:13 Calinou     nrzkt: when you say you have a fully generated world, is it from -31000,-31000,-31000 to 31000,31000,31000?
13:14 nrzkt       Calinou: yes
13:15 est31       tbh, I can't believe it only takes 15 GB
13:15 est31       nrzkt: do you use postgres compression?
13:15 nrzkt       i have added a background processor on my server which emerge all blocks periodically from 0 to abs(X),abs(Y),abs(Z)
13:21 nrzkt       est31: no it's not a good option for our use case
13:21 nrzkt       and on my disk i don't compress to increase the fetch time for my pgsql
13:21 nrzkt       and i also use a SSL connection to my pg
13:24 nrzkt       note: it's a pg 9.6
13:24 nrzkt       but only since 2 weeks :p
13:24 Fixer       Calinou: it is not about client, thats about server upload
13:25 Calinou     don't hosts servers on ADSL connections ._.
13:25 Fixer       Calinou: for example Cash Wolrd starts to have problems somewhere at 20 players and he has 2Mbit of upload iirc
13:25 Fixer       Calinou: people will since it is just a hobby
13:26 nrzkt       Fixer: with 20 players generally it's not upload the problem but engine itself, especially with many lua and ABMs
13:26 Fixer       nrzkt: no
13:26 nrzkt       minetest bw is very lightweight
13:26 Fixer       nrzkt: max_lag is nice, network lag is big
13:26 Fixer       nrzkt: look at xanadu, it has lots of mods and still works nice with 40 players
13:27 Fixer       if you don't use heavy mods it works nice most of the time
13:28 Fixer       jt II with 28 players, max_lag 0.6
13:29 Fixer       Calinou: i believe some vps are rip off, look at some servers, like LinuxGaming, with no players it has huge max_lag 7 because of some BS VPS
13:29 Fixer       same with LibertyLand
13:30 Fixer       one thing is dedicated server, another thing is some virtualised crap
13:45 Calinou     if you want high-performance gaming, you need a dedi anyway
13:46 Calinou     people happily pay €40+ a month to host Minecraft servers
16:03 Fixer       not sure if happily though
16:05 sfan5       <Calinou> also, isn't MapBlock sending, as a client, mostly downloading?
16:05 sfan5       clients don't send any mapblocks
16:05 sfan5       only changes to individual nodes
18:54 paramat     sfan5 rubenwardy game#1336
18:54 ShadowBot   paramat: Error: Delimiter not found in "Page is too big or the server took too much time to answer the request."
18:54 paramat     https://github.com/minetest/minetest_game/pull/1336 add damage sounds
18:55 Krock       I expected some kind of "OUCH!" but this is good too :)
18:55 Krock       Wilhelm Scream, for example
18:59 nrzkt       github down in france because SERVFAIL in DNS... :(
18:59 nrzkt       who is the operator \o`
19:00 Krock       \o` = Zeno`
19:00 nrzkt       backslash Zeno :) thanks Krock
19:01 Krock       operator? Looks more like voice to me
19:01 Krock       np
19:01 \o`         heh, yeah voice
19:02 garywhite   o/
19:02 \o`         \o/
19:02 nrzkt       voice yes :)
19:03 paramat     so, we're still looking for some hero to complete #4530
19:03 ShadowBot   paramat: Error: Delimiter not found in "Page is too big or the server took too much time to answer the request."
19:04 paramat     "Fix superflous shader setting updates"
19:04 \o`         #4530
19:05 ShadowBot   \o`: Error: Delimiter not found in "Page is too big or the server took too much time to answer the request."
19:05 \o`         grr
19:05 nrzkt       lol you have same problem as everybody
19:05 nrzkt       https://www.dynstatus.com/incidents/nlr4yrr162t8
19:06 paramat     https://github.com/minetest/minetest/pull/4530
19:06 \o`         oh that one. I'm staying clear of it :(
19:07 paramat     i might have to bug hmmmm
19:09 paramat     greg est and the 'ninj don't have time
19:11 sofar       nobody can read PRs atm :)
19:11 sofar       I'm surprised I even get mail messages delivered from github atm
19:11 sofar       since smtp servers can't reverse lookup the origin
19:14 \o`         I read the PR
19:15 \o`         github is working fine for me
19:15 \o`         although I've noticed that it's not for a lot of people (not only mentioned here but in other channels)
19:16 \o`         rock and roll (apparently) https://www.youtube.com/watch?v=jSKJQ18ZoIA
19:17 * \o`       dances on his desk
19:17 Krock       babala allahu ahkbar ala
19:19 nrzkt       \o`, in australia problems are far away from US :p
19:20 \o`         nrzkt, yeah they are. https://www.youtube.com/watch?v=csaUvkYOkLY
19:21 \o`         I remember that song fondly from the soundtracker version on Amiga lol
19:22 \o`         we should really add tracker support, but it's not as easy as I first thought it might be
19:23 \o`         on the Amiga a soundtracker player was short and sweet... but on PC it's very complicated (almost, if not have to, emulate the sound Amiga's chip)
19:24 \o`         I'm halfway there using raspberry pi
19:24 \o`         if it all turns out ok I can port it to minetest
19:30 nrzkt       \o`, i think we should port MT on Amiga *joke*
19:32 \o`         it might be a joke... but it's not impossible (well maybe not onto the 68k amigas)
19:32 \o`         to modern AmigaOS it's probably possible
19:33 nrzkt       \o`, and MT on TI68plus ? :)
19:33 \o`         lol maybe not that
19:34 \o`         nrzkt, are you in france?
19:35 nrzkt       ofc
19:35 \o`         nrzkt, can you introduce me to https://en.wikipedia.org/wiki/Aliz%C3%A9e ?
19:36 nrzkt       lol, she is older than the photo now, and we didn't talk about her since hir Moi Lolita in 2001 xD
19:36 \o`         heh
19:36 nrzkt       i think you should look at https://en.wikipedia.org/wiki/Mimie_Mathy
19:37 nrzkt       a very popular actress xD
19:37 \o`         umm. hmmm
19:38 \o`         I would look at her, but I am devoted to Alizée
19:39 nrzkt       :p
19:44 \o`         I should listen to French music more often. The songs not only have catchy tunes but they have deep and meaningful lyrics
19:44 \o`         like https://www.youtube.com/watch?v=bVDfmn_TMkI
19:44 est31       It makes me a bit sad that I couldn't find a cross platform library to make joysticks give force feedback on damage
19:45 est31       I think this would be really useful
19:45 est31       together with https://github.com/minetest/minetest_game/pull/1336
19:45 \o`         don't feel bad, est31, Ca Plane Pour Moi
19:45 sofar       #1336
19:45 ShadowBot   sofar: Error: Delimiter not found in "Page is too big or the server took too much time to answer the request."
19:45 sofar       even the bot can't read github
19:45 est31       lol, github works for me
19:46 sofar       lucky for you, your ISP has a cached github.com A record cached
19:46 sofar       the rest of us are fucked
19:46 est31       1336 has title "Default: Add damage sounds"
19:46 sofar       I've not seen a github page in 3 hours now
19:46 est31       "@paramat
19:46 est31       minetest member
19:46 est31       paramat commented 16 hours ago • edited
19:46 est31       Damage and falling damage sounds (one sound for both).
19:46 est31       No volume balancing or credits yet.
19:46 est31       "
19:46 \o`         github works for me
19:46 \o`         weird
19:46 est31       sofar: edit your host file
19:46 est31       and add ip address 192.30.253.113 for github
19:46 sofar       it doesn't work from my VPS in the netherlands, nor from my VPS in germany, nor from home
19:47 sfan5       sofar: why not
19:47 est31       its dns issue
19:47 sfan5       github has been fine all the time here
19:47 sfan5       unlike twitter
19:47 sofar       massive DDOS on DNS
19:47 est31       https://status.github.com/
19:47 sofar       ;; Got answer:
19:47 sofar       ;; ->>HEADER<<- opcode: QUERY, status: SERVFAIL, id: 43371
19:47 Krock       GitHub: ok, Twitter: ok
19:47 Krock       OpenDNS works fine right now
19:48 nrzkt       sofar, same in france
19:48 est31       github failed about two hours ago for me
19:48 est31       but now its okay again
19:48 nrzkt       google DNS, opendns, root servers, orange , illiad/free, all are empty
19:48 sofar       I can check from 3 countries, and yes google's dns also fails
19:49 nrzkt       maybe we can try south corea or china they have cheating DNS in these countries :p
19:49 sofar       actually, seems google now has it
19:49 sofar       dig github.com @8.8.4.4
19:49 sofar       but I don't use google dns or opendns
19:49 sofar       I run my own resolver
19:49 sofar       (I'm not trusting my ISP, nor would I rely on google)
19:50 nrzkt       sofar, same here, my @home resolver is bound on root servers :)
19:50 sfan5       you *are* already trusting your ISP
19:50 sfan5       whether you use their DNS or not
19:50 sofar       sure, I route my packets through them
19:50 nrzkt       sfan5, depend on dnssec :p
19:50 sofar       exactly
19:51 sofar       sfan5: they are more likely to poison their own DNS cache than intercept packets to port 53
19:51 sofar       if they start intercepting udp packets and modifying them, a lot of stuff will just break
19:51 sfan5       nah
19:52 sofar       sfan5: frontier isn't that capable :D
19:52 sofar       they're pretty bad if you ask me
19:53 sofar       technically
19:53 sofar       they can't even get ipv6 to work
19:53 sfan5       well it obviously depends on the equipment
19:57 est31       well in theory you could do dnssec
20:19 \o`         https://www.youtube.com/watch?v=14IRDDnEPR4
20:19 \o`         actually, this one is better: https://www.youtube.com/watch?v=Lur-SGl3uw8
20:32 sofar       WHOAH
20:33 sofar       ITS BAAAACK
20:33 * sofar     does happy dance
20:34 \o`         WHOA!
20:34 \o`         whoa whoa whoa: https://youtu.be/J_twh5alhxg?t=1249
20:35 \o`         puff the magic dragon, lived by the sea
20:37 est31       didnt know that frolic was an actual word
20:37 \o`         yep it is
20:38 \o`         I just wish they sang the whole song
20:39 sofar       never saw the short form before
20:39 est31       short?
20:40 sofar       always used frollic myself
20:40 sofar       maybe that's just bad spelling
20:40 est31       https://en.wiktionary.org/wiki/frollic
20:40 est31       doesnt exist
20:40 est31       https://en.wiktionary.org/wiki/frolic
20:40 est31       exists
20:41 * \o`       frolics
20:55 nrzkt       frollic is a trademark for dogs no ?
20:55 est31       yes
21:22 \o`         https://www.youtube.com/watch?v=v22SPtCFck8
21:22 \o`         night all
21:51 sofar       am I nuts? what methods are called when I press the `use` key?
21:52 sofar       is it even doing *anything* ?
21:55 est31       its updating the control fields I tihnk
21:55 est31       think*
21:55 est31       so you can use that
21:57 sofar       so that would work inside a vehicle
21:58 sofar       but not to enter/leave a vehicle
22:27 paramat     any more input on game#1285 ?
22:27 ShadowBot   https://github.com/minetest/minetest_game/issues/1285 -- Destroy flammable items when in fire or lava by Ferk
22:28 paramat     cool
23:21 PR4530lives paramat: will I be able to light thrown items on fire with flint and steel?
23:25 paramat     no