Time Nick Message 03:26 ThomasMonroe i gtg to bed, cya tomorrow 03:32 paramat will re-merge #6026 in a few mins as is now approved by sfan5 03:32 ShadowBot https://github.com/minetest/minetest/issues/6026 -- Ores: Make 'absheight' flag non-functional by paramat 03:39 paramat merging 03:42 paramat done 04:26 Natechip If im needed talk to my mobile when its here. 06:36 burli Hi 06:37 burli Is it correct that Minetest currently uses UDP for network communication? 06:39 VanessaE yes mostly 06:39 VanessaE TCP is used for some stuff, like the cURL code that fetches media from a http server 06:39 VanessaE an* 06:39 VanessaE but in-game it's all UDP 06:39 burli Hm, so UDP and TCP is mixed? 06:40 VanessaE UDP and TCP during sign-in, and pretty much UDP-only while playing 06:40 burli I learned right now that it is not a good idea to mix both in the same app 06:40 VanessaE it's fine in this case. 06:40 burli ah, ok 06:42 burli found a nice article about that http://gafferongames.com/networking-for-game-programmers/udp-vs-tcp/ 06:42 nerzhul UDP is used everywhere except media downloading & server list which use TCP/HTTP 06:42 burli The problem is that since TCP and UDP are both built on top of IP, the underlying packets sent by each protocol will affect each other. Exactly how they affect each other is quite complicated and relates to how TCP performs reliability and flow control, but fundamentally you should remember that TCP tends to induce packet loss in UDP packets 06:43 nerzhul burli, yes it was especially right in 56k, with modern bandwidth this problem disappear 06:43 VanessaE nerzhul: one exception to that btw 06:43 nerzhul you have same problem if you listen stream music on youtube when playing MT 06:43 VanessaE http used in-game if you have a mod installed that uses that feature. 06:47 burli However. Was just curious. As far as I remember there is a bottleneck in networking. 08:43 sfan5 burli: like nerzh​ul already said this is not a problem at all on modern connections 08:44 sfan5 also tcp is only used during sign-in and the amount of traffic used by mt (on the client) isn't enough to fill up all bandwidth anyway 08:45 sfan5 the exception VanessaE mentioned only applies to the server, which should be running on a good connection anyway 08:45 VanessaE right. 08:45 sfan5 i don't know from which year that article you linked it but if my connection started dropping packets just because i have some tcp connection under mild load i'd call my isp to complain 08:51 jomat The oldest comments are from 2008 :-) 08:52 burli That's what I want to say right now 08:52 burli So it's not that old I think 08:56 jomat No, Internet wasn't that bad back then :-) It's just the architecure of udp that allows packets beeing dropped... 14:38 red-001 can someone review #5965? 14:38 ShadowBot https://github.com/minetest/minetest/issues/5965 -- Create a filesystem abstraction layer for CSM and only allow accessing files that are scanned into it. by red-001 17:59 nerzhul merging #6041 in ~5 mins 18:00 ShadowBot https://github.com/minetest/minetest/issues/6041 -- Isolate irrlicht references and use a singleton by nerzhul 18:00 nerzhul (if GH permits it, else doing it manually) 18:04 nerzhul oops merged the wrong PR, forcepush to solve that 18:05 lisac nerzhul, No need this PR is also nice 18:05 lisac :P 18:05 nerzhul oh seems we only need to rename one function for Krock 18:05 nerzhul will push the fix, it's better 18:05 nerzhul ok Krock if i push the fix ? gh doesn't show the kilbith branch fix and doesn't merged it too 18:06 Krock yes, LGTM 18:07 nerzhul okay doing that, i compile it locally and if okay push it with kilbith name 18:07 kilbith nah, no need 18:07 kilbith I mean for the name 18:07 nerzhul okay 18:09 Krock nerzhul, uhm.. wrong commit? 18:09 nerzhul Krock, ? 18:09 Krock whatever, this trivial function name change can be omitted 18:09 nerzhul i would merge 6041 and i merged 6050 18:09 nerzhul too late :p 18:10 nerzhul thanks kilbith for that feature, i think players will like it :) 18:11 nerzhul then now merging 6041 18:11 kilbith I expect $500 from the community now 18:11 nerzhul thanks Krock for your time 18:11 nerzhul kilbith, sorry no github bounty 18:12 kilbith $1500 if you want the god rays and deferred rendering 18:12 nerzhul i think it's a lgtm price in private software, it's 2 days of senior dev development 18:12 nerzhul then it's not very expensive :p 18:13 Krock $1337 for a kingdom 18:15 red-001 https://www.openhub.net/p/minetest-c55/estimated_cost 18:18 nerzhul wow :p 18:20 Krock heh, Linux: $ 287,679,508 18:21 red-001 #6054 18:21 ShadowBot https://github.com/minetest/minetest/issues/6054 -- Hand isn't visible with arm inertia enabled. 18:21 kilbith nerzhul has the same issue afaik 18:21 kilbith *had 18:22 kilbith you set the setting to true in-game? 18:23 kilbith nevermind 18:23 red-001 no, it was enabled when starting the game 18:23 red-001 disabling it fixes the issue 18:23 nerzhul red-001, same pb, i have removed my cmake-build-debug and rebuild 18:23 kilbith are you compiling in debug mode? 18:23 nerzhul i don't understand why i have that problem 18:23 nerzhul but cleanup your cmake env and rebuild fix the problem 18:24 Krock did a release build and it worked alright 18:24 kilbith it shouldn't cause that even without cleaning cmake 18:24 nerzhul debug and release works 18:24 nerzhul kilbith, agreed 18:24 red-001 huh 18:25 kilbith it's the compiler doing some things in your back when using fabsf or std::fabs 18:27 kilbith at least it's my theory 18:27 kilbith does your CPU supports SSE/AVX? 18:28 red-001 I think so let me check 18:31 red-001 SSE4.1/4.2, AVX 2.0 18:31 kilbith right 18:32 sfan5 does gcc even generate avx instructions by default? 18:32 sofar with the right -march yes 18:32 sfan5 by default 18:32 sofar depends on the distro default gcc flags, most of the times 18:35 nerzhul oh sofar hi 18:35 nerzhul in debug mode march is not defined 18:35 nerzhul for me only cleaning up cmake build dir fix the problem, and i got it on two machines, home + work 18:35 nerzhul very strange 18:36 nerzhul (one is AMD FX 8350 and the other core i3) 18:36 red-001 testing it on a freshly cloned copy of the repo 18:36 nerzhul red-001, you don't need to clone 18:36 kilbith add this flag if necessary: -march=native 18:36 nerzhul just remove your cmake build dir 18:36 nerzhul (i hope you have a specific cmake dir and don't use source tree :p 18:37 red-001 alright it worked now 18:37 kilbith very strange 18:38 nerzhul yeah, nothing in cmake config changed between builds it's very strange 18:40 red-001 huh this looks a lot nicer in person then in the example video 18:41 kilbith yeah, I agree 18:43 jcalve I have the arm disappearing bug too in a debug build, freshly cloned 18:43 red-001 huh 18:44 * red-001 smells undefined behavior 18:45 kilbith well I don't see how 18:46 nerzhul sounds like a underfined variable init ? 18:46 nerzhul oh 18:46 nerzhul bool m_arm_inertia; is not inited 18:46 nerzhul but yes inited in constructor 18:46 nerzhul no sense 18:47 red-001 some werid bug in incremental linking? 18:49 nerzhul seems jcalve said he has the bug on fresh build 18:49 jcalve I do 18:50 jcalve I cleaned the folder and I'm rebuilding to see if it goes away 18:58 jcalve still there 19:09 nerzhul i can reproduce it and i know where it is 19:09 nerzhul i found where is the problem 19:09 nerzhul there is division per zero 19:12 nerzhul if i forbid 0.0 division bug disappear 19:17 bigfoot547 I'm compiling new minetest! I'm excited! :D 19:25 kilbith red-001: #6055 19:25 ShadowBot https://github.com/minetest/minetest/issues/6055 -- Camera: Fix possible division per zero by kilbith 19:33 bigfoot547 bigfoot547> Ehm, compiling minetest is giving me a linking error 19:33 bigfoot547 I'll get the paste 19:33 bigfoot547 https://pastebin.com/6pHgqd4b 19:33 bigfoot547 There's the error 19:33 bigfoot547 (#minetest ) 19:36 jcalve without knowing anything about c++, looks like your json lib has a problem 19:37 bigfoot547 Yeeees 19:37 bigfoot547 Look: Linking CXX static library libjsoncpp.a 19:37 bigfoot547 It compiled 19:37 bigfoot547 Then, [ 12%] Built target jsoncpp 19:38 bigfoot547 But, the linker said stuff :( 19:38 sfan5 are you linking to system jsoncpp? 19:38 bigfoot547 Also, .text+0x3cfe is assembly stuff 19:39 bigfoot547 sfan5: How do you mean? 19:39 sfan5 do you have -DENABLE_SYSTEM_JSONCPP=TRUE 19:39 bigfoot547 No 19:40 bigfoot547 I'm building with it now 19:51 Fixer compiled newest mt just fine via msys2/mingw64 19:52 bigfoot547 I'm having problems with jsoncpp 19:55 bigfoot547 I'm installing libjsoncpp-dev' 19:55 bigfoot547 I'm installing `libjsoncpp-dev' 19:56 bigfoot547 How do I use the system jsoncpp? 19:56 sfan5 -DENABLE_SYSTEM_JSONCPP=TRUE 19:57 bigfoot547 Ok, I'm using that flag 19:57 bigfoot547 Let's just wait for the comp 20:06 Shara dra :P 20:09 bigfoot547 It works! Thank you sfan5, Fixer, jcalve, and Shara! 20:09 bigfoot547 :D 20:09 bigfoot547 Credit where credit is due ;) 20:11 nerzhul no problem i'm here :p 20:57 Shara old sneak* 20:58 Shara Eww, wrong tab sorry 21:10 nerzhul jcalve, are you there , 21:10 jcalve nerzhul: yes 21:10 nerzhul can you edit camera.cpp L199 to 203 to replace the two fabs with 21:10 nerzhul https://pastebin.com/MRBAMNzv 21:11 nerzhul and tell me if it works (for me okay) 21:11 jcalve sure 21:12 nerzhul red-001, ^ 21:13 jcalve nerzhul: that fixed it for me too 21:19 nerzhul jcalve, nice 21:19 nerzhul red-001, ? 21:19 nerzhul okay pushing he fix to master 21:19 red-001 sorry I was afk 21:19 red-001 testing now 21:19 nerzhul pj 21:19 nerzhul nice i'm preparing the patch 21:23 nerzhul red-001, waiting you 21:25 red-001 yeah that worked 21:31 nerzhul nice 21:31 nerzhul pushing the fix 21:31 nerzhul + optim reinit += operation 21:32 red-001 any idea why it worked in release builds? 21:32 nerzhul i don't have idea 21:34 nerzhul i don't know why compiler let division by zero, if i remember i crash programs genrally 21:34 nerzhul it's not an exception in C++ hmmm 21:35 red-001 * red-001 smells undefined behavior 21:35 nerzhul red-001, yes 21:35 red-001 I guess it optimized out the crash 21:35 nerzhul https://stackoverflow.com/questions/3004095/division-by-zero-undefined-behavior-or-implementation-defined-in-c-and-or-c 21:35 kilbith well, that was a good lesson 21:35 nerzhul no crash, just undefined behaviour as it seems 21:36 nerzhul whenever you do a division ensure you never have zero, if you have, handle 21:40 red-001 at least it's not JS 21:41 red-001 which seems to think it's infinaty 21:41 red-001 infinity* 21:42 nerzhul :p 21:44 red-001 huh 21:44 red-001 I should check if it returns negative infinite when dividing a negative number 22:09 nerzhul sfan5, when you get time can you review #5963 ? i'm waiting for your approval to merge, it looks nice but i need your expertise :p 22:09 nerzhul i should go 22:09 ShadowBot https://github.com/minetest/minetest/issues/5963 -- Fix msvc annoyances by adrido