Minetest logo

IRC log for #minetest-dev, 2015-04-12

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

All times shown according to UTC.

Time Nick Message
00:02 paramat i'm happy with #2598 now and will push in a few hours. Code cleanup, some changes, some bugfixes, lots of checking terrain against master branch mgv6, and have added some comments in the thread
00:02 ShadowBot https://github.com/minetest/minetest/issues/2598 -- Mgv6: Add optional snow biomes by paramat
00:04 johnnyjoy1 joined #minetest-dev
00:09 johnnyjoy joined #minetest-dev
00:42 paramat left #minetest-dev
01:55 luizrpgluiz joined #minetest-dev
01:58 luizrpgluiz left #minetest-dev
02:57 decimalguy left #minetest-dev
03:02 est31 hmmmm, what do you know about shebangs on freebsd?
03:03 est31 because there is a "#! /bin/sh"
03:03 est31 and jenkins doesnt want to build because of "configure: error: cannot run /bin/sh /usr/local/jenkins/jobs/minetest - FreeBSD - Official/workspace/src/gmp/src/config.sub"
03:04 est31 according to internet, this points to problems with either the shebang or +x permissions.
03:04 est31 but +x is there
03:07 paramat joined #minetest-dev
03:08 paramat will push 2598 in a few minutes
03:09 est31 what does taiga look like below?
03:10 paramat below surface?
03:10 est31 snowblock, then dirt with snow I guess
03:10 paramat yes
03:10 est31 ok
03:11 paramat snowblock at surface level, dirt_with_snow below
03:17 hmmmm est31, i'm pretty sure you can have a space between the shebang and the shell command
03:18 hmmmm just tested, the space makes no difference
03:19 est31 ok
03:19 hmmmm I'll download gmp 5.1.3 and try to compile it myself and see what happens
03:19 hmmmm it's possible that the port version has a makefile patch
03:19 est31 yes
03:20 hmmmm where'd you download it from?
03:20 hmmmm ah nevermind found the mailing list announcement
03:20 est31 https://gmplib.org/download/gmp/
03:20 est31 https://gmplib.org/download/gmp/gmp-5.1.3.tar.bz2
03:21 est31 The according PR is #2620
03:21 ShadowBot https://github.com/minetest/minetest/issues/2620 -- SRP based login by est31
03:22 hmmmm I just did ./configure && make
03:22 hmmmm no problems here
03:23 hmmmm wha... 1774 files...
03:23 hmmmm oh my
03:25 hmmmm maybe we could use mini-gmp
03:25 paramat now pushing 2598, will notify when done
03:25 hmmmm it's not as high performance as the entire libgmp, but it's one single file that we can include just like that
03:26 hmmmm and then we'll have an option for the 'full' version of libgmp... sound good??
03:26 est31 yea
03:26 hmmmm we can totally sidestep this horrifying autoconf build system and the incredible amount of bloat
03:27 est31 "The performance target for mini-gmp is to be at most 10 times slower
03:27 est31 than the real GMP library, for numbers of size up to a few hundred bits"
03:28 est31 but yes rather something slow than this kind of bloat
03:28 hmmmm in practice people are gonna want libgmp
03:28 hmmmm but in a pinch this works fine
03:28 est31 its analogous to the lua and luajit
03:28 hmmmm right
03:31 est31 aaaaaand it compiles
03:31 est31 wow
03:31 est31 now lets test the performance
03:32 est31 normal libgmp is ~40 ms
03:32 hmmmm we're definitely going to want the full libgmp for servers, but it's not such a necessity for clients
03:32 est31 yes
03:32 est31 that should be documented somewhere
03:33 est31 (my PR currently misses a README change for libgmp)
03:33 paramat push complete
03:33 crazyR_ joined #minetest-dev
03:34 est31 wow 1815
03:34 est31 1815 ms
03:34 est31 thats not quite fast
03:34 est31 also, its not factor 10
03:36 est31 so the "performance target" isn't reached
03:39 hmmmm maybe it's not such a good idea to bundle libgmp
03:39 hmmmm this is pretty horrifying
03:40 est31 I think mini-gmp is enough for the start
03:41 est31 (and permanent too)
03:41 hmmmm well
03:41 hmmmm you wanted 'slow' for auth
03:41 hmmmm there you have it
03:41 est31 lol
03:43 est31 I'll add mini-gmp.{c,h} into util/ and adjust the cmake.
03:44 hmmmm okay =]
03:48 VanessaE I hope that extra 1800 ms is compensated for by figuring out why some people have trouble with long sign-on times for servers that ought to be fast (e.g. lightly-modded). :)
03:48 est31 VanessaE, they just compile minetest with real libgmp, and its fast as hell
03:49 VanessaE (not trying to pour cold water on y'all's work here, just warning based on users' complaints we've all heard of before)
03:50 est31 They will complain, but what should we do?
03:50 VanessaE for mini-gmp, nothing.
03:50 VanessaE I'm talking about other parts of the sign-in process.
03:50 VanessaE you know, post-auth stuff
03:50 est31 ah
03:51 VanessaE you've already made it past auth, now you hypothetically wait like 10 seconds on some servers for reason that's not clearly spelled out in the progress display, that sorta thing.
03:51 est31 ah
03:52 VanessaE (I've been lucky not to get hit by that very often, and when it does happen, it's more like only a few seconds usually)
03:56 selat joined #minetest-dev
04:02 cheapie joined #minetest-dev
04:25 paramat updated, compiled and tested, taiga was missing most of it's pinetrees, will push the fix soon
04:41 paramat will push the fix #2621 when checks are done
04:41 ShadowBot https://github.com/minetest/minetest/issues/2621 -- Mgv6: Fix taiga, allow pine tree spawning on snowblocks by paramat
04:50 paramat now pushing 2621
04:54 paramat done
05:08 paramat left #minetest-dev
05:13 est31 very weird
05:13 est31 the build is somehow unreliable
05:14 est31 once it works https://travis-ci.org/minet​est/minetest/jobs/58130277
05:14 est31 once it doesnt https://travis-ci.org/minet​est/minetest/jobs/58130278
05:15 est31 so gcc likes it clang doesnt
05:40 hmmmm you got minetest working with freebsd/gcc?
05:40 hmmmm jeez
05:40 hmmmm well, rather should i ask, did it merely compile or is the resulting object file able to be executed
05:40 est31 https://travis-ci.org/minete​st/minetest/builds/58130274
05:41 est31 windows claims to have not found the .h file I have no idea
05:41 est31 perhaps we have to link against a .lib file
05:41 est31 hrm I think that dependency isnt needed at all
05:42 hmmmm for Linux, you mean the 'undefined reference to <thing>'?
05:42 est31 yes
05:43 hmmmm i know what the problem is
05:43 est31 ?
05:43 hmmmm you need to add extern "C" { } around the definitions in the mini-libgmp.h
05:44 est31 there is
05:44 est31 #if defined (__cplusplus)
05:44 est31 extern "C" {
05:44 hmmmm oh, hrmm
05:44 hmmmm well the object file is clearly being made
05:44 hmmmm generated rather
05:45 hmmmm is there some sort of __declspec__((export)) or equivalent
05:45 est31 wut?
05:45 hmmmm like some kind of compiler directive for external linkage
05:45 hmmmm whatever the gcc equivalent is
05:45 est31 regarding the windows failure?
05:45 hmmmm hmm well i'm looking at it right now
05:46 est31 win failure should be fixed now
05:46 est31 (removed the include, we dont need it anymore)
05:46 hmmmm yeah I don't know
05:47 hmmmm I wonder if __cplusplus is being defined as it should be
05:47 hmmmm add an #error "yep, __cplusplus is defined here" inside that if directive
05:49 Kray_ joined #minetest-dev
05:49 est31 ok windows works
05:52 VargaD_ joined #minetest-dev
05:55 est31 hmmmm, it doesnt trigger the #error
05:56 hmmmm welp there's your problem
05:56 est31 so, what happened?
05:56 hmmmm uhhh
05:56 hmmmm i don't know yet :(
05:58 hmmmm oh duh
05:58 hmmmm srp.c is the thing that includes mini-gmp.h which is also .c
05:59 hmmmm of course it's not going to have that defined
06:00 Hijiri joined #minetest-dev
06:00 est31 it doesnt like the forced extern at all: https://travis-ci.org/minetes​t/minetest/jobs/58132367#L660
06:01 hmmmm right, as expected
06:01 hmmmm i don't know, this got a whole lot tougher
06:55 Krock joined #minetest-dev
06:55 Calinou joined #minetest-dev
06:58 cib0 joined #minetest-dev
07:01 Hunterz joined #minetest-dev
07:36 kilbith joined #minetest-dev
08:31 err404 joined #minetest-dev
08:58 Krock Updated #2618
08:58 ShadowBot https://github.com/minetest/minetest/issues/2618 -- 20s timeout when connecting to server by SmallJoker
09:19 Hijiri joined #minetest-dev
09:48 jin_xi joined #minetest-dev
10:01 est31 *****FACEPALM****
10:01 est31 *****FACEPALM****
10:01 est31 Those functions the linker can't link are my own
10:01 est31 I have defined them
10:01 est31 in that file
10:02 est31 I just wonder what the compiler didnt like about them
10:05 Hijiri joined #minetest-dev
10:06 Anchakor_ joined #minetest-dev
10:13 MinetestForFun joined #minetest-dev
10:30 est31 added a "static" now it compiles
10:30 est31 very fine
10:31 AnotherBrick joined #minetest-dev
10:40 leat joined #minetest-dev
10:44 err404 joined #minetest-dev
11:03 Calinou joined #minetest-dev
12:09 err404 joined #minetest-dev
12:45 leat joined #minetest-dev
12:57 leat joined #minetest-dev
13:49 leat joined #minetest-dev
14:45 Player_2 joined #minetest-dev
15:02 Hijiri joined #minetest-dev
15:03 blaze joined #minetest-dev
15:10 Hijiri joined #minetest-dev
15:20 nrzkt joined #minetest-dev
15:20 nrzkt left #minetest-dev
15:21 Hijiri joined #minetest-dev
15:22 selat joined #minetest-dev
15:27 leat joined #minetest-dev
15:30 Hijiri joined #minetest-dev
15:37 err404 joined #minetest-dev
15:44 Anchakor_ joined #minetest-dev
15:48 est31 joined #minetest-dev
15:51 Hijiri joined #minetest-dev
15:57 selat joined #minetest-dev
16:10 hmmmm joined #minetest-dev
16:15 Krock hmmmm and est31: Are there any comments for #2618 ?
16:15 ShadowBot https://github.com/minetest/minetest/issues/2618 -- 20s timeout when connecting to server by SmallJoker
16:25 hmmmm i have a general comment on Game::connectToServer)
16:25 hmmmm the assumption that things are going on in a separate thread in a certain order is mostly invalid unless special precautions are being taken
16:26 hmmmm everybody should read this:  http://www.cs.ucla.edu/~todd/research/snapl15.pdf
16:28 Krock > Java is an unsafe language
16:29 hmmmm i kind of agree with the comments in that PR that a configurable setting isn't all too useful
16:29 hmmmm under what circumstances is somebody ever going to want to modify it?
16:30 hmmmm I can't believe the client currently just hangs there and freezes up if the server never times out
16:30 Krock It wasn't meant serious. I wondered if they also would request a minetest.conf setting for something simple like this
16:30 johnnyjoy joined #minetest-dev
16:31 hmmmm 20 seconds is kind of long too, maybe 10 would be better *shrug*
16:31 hmmmm why did you choose 20
16:31 Krock because 30 was too long
16:31 Krock 10s is fine
16:32 VanessaE I'd have suggested 10 also.
16:36 est31 quicky scanned that paper, and I think the issues are not that serious
16:37 est31 its mostly complaining that low level programmers have to insert fences
16:37 VanessaE that paper makes me wanna go back to BASIC. :P
16:37 est31 doesnt help either
16:38 est31 I'm not sure which kind of reordering the paper targets: the hw one or the compiler one
16:38 VanessaE looks like both.
16:38 est31 basic would only target the compiler one
16:38 est31 :)
16:39 VanessaE I Was joking, and besides BASIC isn't multi-thread capable :)
16:39 est31 and it doesnt have float typed line numbers
16:39 VanessaE haha
16:39 VanessaE of course the whole paper starts off from a false premise - that all languages are surely as bad as Java
16:40 VanessaE C++ can't possibly be THIS bad can it?
16:40 est31 Java has advantages c++ has its advantages
16:40 est31 for example, build systems for C are like real shit
16:41 est31 with .dll .so .al .a and all that stuff
16:41 est31 java only has class files, and then it zips them to jar files
16:41 est31 simple and clean
16:41 VanessaE I'll agree with you there, to a degree - but dll, .so, etc is OS-level stuff, not language-level
16:41 sfan5 what is an .al file
16:42 est31 .al is for autolinker or so
16:43 sfan5 you mean .la probably
16:43 sfan5 that belongs to pkg-config and is optional
16:43 est31 perhaps yes
16:44 est31 VanessaE, portability is one of java's advantages
16:45 VanessaE write one, pray everywhere :P
16:45 VanessaE once*
16:45 est31 lol
16:45 est31 also, c++'s API is very basic and trimmed down
16:46 Krock est31, ah and c++ ain't portable?
16:46 est31 Krock, basic c++ programs yes
16:47 est31 but already simple things as fetching the current time lead to a port mess
16:47 est31 "mess"
16:47 Krock That's because everybody wants his own C++ version/API and no-one cares about Java
16:48 est31 ?
16:50 Krock nvm. This theme is too deep to start with.
16:58 Calinou if we used C++11, C++ would be better
16:58 Krock so, what's the status of that PR? Is there any rewrite required?
16:58 Calinou else, Java beats C++
16:58 MinetestForFun joined #minetest-dev
16:59 Krock Calinou, afaik, there are still supported OS's which don't support it.
16:59 err404 joined #minetest-dev
16:59 Krock s/it/C++11/
16:59 Calinou yeah, like Ubuntu 12.04
16:59 Calinou noone should use that anymore
17:00 Krock Same goes with Windows XP :P
17:00 Calinou Windows XP is not a supported OS
17:00 Krock oh well, it's not suportted anymore
17:01 Krock I feel like I've shot myself into the knees.
17:56 rubenwardy joined #minetest-dev
18:10 hmmmm Windows XP is still supported
18:10 hmmmm I think we dropped windows 2000 support
18:19 err404 joined #minetest-dev
18:24 Krock hmmmm, we were talking about the official support by developers, not about Minetest.
19:17 hmmmm oh
19:17 hmmmm my bad, didn't read the whole thing
19:17 hmmmm yeah.. the price you pay for having such an expansive base API such as Java's is that it's extremely difficult to port to other architectures
19:17 hmmmm and you do pay
19:18 hmmmm if you want to run Java on INTEGRITY OS, it costs $200k a pop
19:18 hmmmm and it doesn't even support the full SE library
19:19 hmmmm i ran up against this problem at my last job, and when we discovered the incredible cost and deficiencies, we were happy we went with C instead
19:23 Krock C was already enough?
19:57 err404 joined #minetest-dev
20:00 prozacgod joined #minetest-dev
20:25 hmmmm it might be a good idea to show a stack backtrace on printing to the errorstream from within the lua api
20:25 hmmmm there's no other way to identify which call in your mod was failing
20:41 hmmmm https://github.com/kwolekr/minetest/commit​/de60b340b06ef07f2656979cf59bbb898da25a6c
20:41 hmmmm PTAL^
20:48 MinetestForFun joined #minetest-dev
20:59 prozacgod joined #minetest-dev
21:35 kahrl_ PTAL = ?
21:35 kahrl_ Paramat, Take A Look?
21:45 kilbith s/Paramat/Please
22:23 Vexyl joined #minetest-dev
22:26 Megal joined #minetest-dev
22:51 paramat joined #minetest-dev
22:54 paramat hmmmm perhaps you should look at this PR #2561 i don't see any comments from you so just letting you know
22:54 ShadowBot https://github.com/minetest/minetest/issues/2561 -- Add mapgen specific flags to create world dialog by srifqi
23:08 paramat thanks for the serialize schematic commit
23:11 paramat also, there is a request for mapgen flags to be the same in all subgames #2585 and i agree. i would like to move the jungles and dungeons flags out of mtgame and make them engine defaults
23:11 ShadowBot https://github.com/minetest/minetest/issues/2585 -- Add the default settings of mg_flags by Rui914
23:11 paramat bbiab
23:11 paramat left #minetest-dev

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