Minetest logo

IRC log for #minetest-dev, 2015-04-18

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

All times shown according to UTC.

Time Nick Message
00:06 Hijiri is river water going to be a flowy finite liquid?
00:07 paramat no way =)
00:07 Hijiri o-ok
00:08 paramat the whole river would drain into the ocean if it was heh
00:14 * Sokomine can't resist to think about liquid mobs that'd move around the map like liquids...flowing from a to b...
00:14 * Sokomine picks up a bucket and removes the liquid before anything happens
00:17 Hijiri like the lava monster mod for infiniminer?
00:18 Sokomine i don't know them. there are some decorative lava mobs in i believe zeg9s mob modpack. they wobble around a bit
00:20 Hijiri the lavamonster thing was a popular plugin for a modded version of infiniminer, the game would create a "snake" of lava that would follow the nearest player
00:20 Hijiri I think you could kill it by detonating explosives when it was near, but I don't remember
00:25 paramat i might push the river water PR later since i have simplified it on PilzAdam's request, have hmmmmm's approval and since it's partly a mapgen issue
00:39 hmmmm i wouldn't be too pushy about things like this
00:39 hmmmm that's a great way to gain distrust
00:40 paramat it's risky, i guess i can wait a day
00:54 paramat ugh my crazy scatter-brained judgement =/
01:40 paramat left #minetest-dev
02:52 Zeno` joined #minetest-dev
03:34 Yepoleb joined #minetest-dev
03:36 Zeno` wth... the damage tilt/flash bug isn't fixed yet?
03:38 VanessaE nope.
03:38 Zeno` it's a single line (if that) fix
03:39 Zeno` I'll give it one more day and just do it myself :/
03:39 Zeno` I thought the person who introduced the bug would have fixed it by now TBH
03:39 VanessaE you're talking damage-disabled/creative mode right?
03:40 Zeno` yeah
03:40 VanessaE (I can stand in a swarm of bees all day long and watch the flash/tilt effect from the bees trying to injure me)
03:40 Zeno` yes that one.
03:40 Zeno` it's a very very simple fix
03:44 Zeno` Is he here now?
03:44 Zeno` (I know his nick isn't hehe)
03:45 Zeno` That should have been fixed as soon as I gave the bisect. Lucky it's not critical
03:49 leat joined #minetest-dev
04:23 paramat joined #minetest-dev
04:25 paramat http://i.imgur.com/ZZe6a5b.png slice of a 4D mandelbrot. surprising thing is the shape improves (smoother, more continuous) with fewer iterations, here only 12, makes mapgen faster
04:29 Zeno` looks like a flying fish
04:30 Zeno` fractals are supposed to get more complexity (less smooth) with more iterations :P
04:30 Zeno` So I don't think it's surprising :)
04:30 paramat exactly
04:30 paramat i was about to use blur, then remembered that
04:31 Zeno` what is that square thing on the fish's tail?
04:31 Zeno` a dungeon?
04:40 paramat slice of dungeon
04:43 leat joined #minetest-dev
04:51 hmmmm is this really what we want?? https://github.com/minetest/minetest/commi​t/ca5df44edb1963d4c634390d33229ec928e284dc​#diff-59f99d02fe5b328bb9a57b023497b5edR390
04:52 hmmmm it seems to me like this totally breaks the concept behind RUN_IN_PLACE for Mac OS X
05:11 leat joined #minetest-dev
05:19 nore joined #minetest-dev
05:21 Zeno` odd. I'm not sure if --world and --worldname are handled correctly
05:21 Zeno` hmm
05:22 Zeno` well maybe there are
05:22 Zeno` what would be the way to start a simple single player game and skip the main menu?
05:24 Zeno` /home/gruehunter/minetest/bin/minetest --config ~/.minetest/minetest.conf --worldname ram-test --go --random-input --address '' works
05:24 Zeno` :/
05:24 * Zeno` trembles in fear
05:29 Zeno` The comments given by minetest --help are kind of misleading
05:30 Zeno` "--world <value>       Set world path (implies local game)" and "--address <value>     Address to connect to. ('' = local game)" are referring to two different things when they say "local game"
05:46 est31 joined #minetest-dev
06:04 hmmmm wth
06:04 hmmmm too many things require an overhaul
06:04 hmmmm i say we just scrap it and start over from scratch =]
06:05 hmmmm total rewrite of minetest
06:05 VanessaE O_o
06:05 * VanessaE dies
06:05 hmmmm lol
06:05 VanessaE appropriate you should say this now:
06:05 VanessaE you're joking of course, about flushing the existing codebase down the toilet...
06:06 hmmmm dead serious
06:06 VanessaE I'm in the middle of modeling a toilet to replace the one in homedecor right now :)
06:06 hmmmm I'm about to blank all the source files and commit it
06:06 hmmmm somebody can then make a github issue that the game doesn't exist any longer
06:06 hmmmm "it sucks.  we blew it up"
06:10 hmmmm it's sorta like this video:  https://www.youtube.com/watch?feature=p​layer_detailpage&amp;v=KR2cjZJQoww#t=57
06:15 Miner_48er joined #minetest-dev
06:16 paramat left #minetest-dev
06:17 leat joined #minetest-dev
06:23 hmmmm whoops, turns out that Irrlicht was a dependency of the server the whole while thanks to a typo in porting.cpp
06:23 Hunterz joined #minetest-dev
06:24 est31 O-o
06:28 leat joined #minetest-dev
06:32 neoascetic joined #minetest-dev
06:40 Zeno` let's do it
06:40 * Zeno` presses the big red button
06:45 OldCoder joined #minetest-dev
06:48 Zeno` https://www.youtube.com/watch?v=RMXpnTJ1Lx4
06:49 est31 "Ends as expected."
06:49 est31 man why is a spoiler there?
06:53 hmmmm I guess those guys had a blast
06:57 Zeno` the cow pad kind of does look like a mine
06:58 Zeno` well, I'm sick of input handler now... and all I've done is move some stuff around, lol
06:59 hmmmm i know that feel bro
07:00 hmmmm it started out with me wanting to add solaris support for initializePaths
07:00 hmmmm and it ends up with me basically refactoring the entire porting.cpp because i find so many problems and messiness that i can't work in
07:00 hmmmm you really need to put blinders on to the messiness if you want to stay focused
07:01 hmmmm i told myself that i'd have this patch done by 1:30 or else, and it's already 3
07:02 Zeno` well I almost ended up changing stuff in main.cpp... kind of hard to remain focused sometimes when looking at stuff leads you to having to look at other stuff and realising the other stuff needs fixing/changing/cleaning also
07:02 est31 yea I know that
07:02 hmmmm minetest is a full time job
07:02 hmmmm i wish sane, efficient abstractions were chosen the first time around though
07:02 est31 yesterday I had to fix the wstring stuff for translatePassword, because I didn't like to add wstring to my code
07:03 est31 (thanks to ShadowNinja for pointing this out)
07:08 Zeno` Is there a fallback mechanism for your SRP, est31?
07:08 Zeno` i.e. for existing users using an insecure password?
07:09 hmmmm hrmm
07:10 hmmmm is defined(__SVR4) really needed to determine if something is solaris?
07:10 Zeno` something like if (!srp_login(name, password) && insecure_login(name, password)) { // migrate password or tell them to change it }
07:10 Zeno` There is no __SOLARIS ?
07:10 hmmmm nope
07:10 hmmmm there's __sun
07:11 hmmmm I am assuming that __sun && __SVR4 == solaris, otherwise SunOS
07:11 Zeno` I don't know
07:11 est31 Zeno`, yes
07:11 hmmmm SunOS was originally based off of BSD
07:11 Zeno` est31, cool :) I still don't like the MP lib being in the src tree though
07:11 hmmmm when they switched the name to Solaris they also switched the base to SVR4
07:11 est31 the server looks up the password entry from the auth handler
07:12 Zeno` SunOS is a derivative of BSD?
07:12 est31 then it looks whether it is an srp hash
07:12 Zeno` I didn't know that
07:12 hmmmm yes
07:12 Krock joined #minetest-dev
07:12 est31 if its an srp hash, its doing the srp, if its a traditional password, its still doing srp, but over the traditional has
07:12 est31 kind of matroska here :p
07:13 Zeno` it'd be nice to tell the user to update their password. Maybe that can be done via lua
07:13 est31 user interaction isn't neccessary
07:13 est31 We can directly update the password as part of the protocol
07:13 Zeno` oh
07:13 Zeno` I see
07:13 Zeno` yeah
07:13 Zeno` magic
07:14 est31 the srp over the traditional password will get that mechanism
07:14 est31 yea :)
07:15 est31 so, to summarize: ShadowNinja and Zeno` are against mini-gmp, hmmmm and me are for it.
07:16 Zeno` Why can't the person compiling minetest download it (e.g. we don't include irrlicht in the source tree)
07:16 Zeno` sfan5 would have to update his mingw pre-compiled libs of course
07:17 est31 what about first adding them, then letting everybody migrate, then removing them again?
07:17 est31 eg 1 month of migration time
07:17 Zeno` What advantage would that have?
07:17 sfan5 updating the pre-compiled libs is no problem
07:18 hmmmm well
07:18 hmmmm libgmp is a GNU library
07:18 sfan5 actually
07:18 hmmmm typically GNU libraries give you as much trouble as possible on windows
07:18 sfan5 does it even work out with licenses?
07:18 Zeno` it's LGPL sfan5
07:18 sfan5 i see
07:18 Krock I could try to compile libgmp on here
07:19 hmmmm I dunno
07:19 hmmmm the rationale was to make things more broken by requiring another dependency
07:19 hmmmm more dependencies == more of a chance for things to go wrong
07:19 Zeno` well, that's possible sure
07:20 hmmmm like, I can't even compile minetest using gcc with leveldb enabled
07:20 hmmmm thank goodness it's not a required dependency
07:20 sfan5 are you using ubuntu 12.04, wait no
07:20 sfan5 the leveldb package on 12.04 is completly broken
07:21 Zeno` I've seen your stackoverflow question, hmmmm. You can't even compile int main(void) { return 0; }
07:21 hmmmm yeah man
07:21 hmmmm well no I can compile it, but it doesn't execute
07:21 hmmmm irrlicht is a library with C++ linkage, I'm 99% sure it's caused by an ABI mismatch between clang and gcc
07:22 hmmmm I really cannot advocate using C++ for a library interface to anybody
07:22 est31 @ licenses: it seems I will have to add another copy of mini-gmp (if we decide to add mini-gmp at all), one that isn't LGPLv3+, but one that is LGPLv2.1, like minetest
07:22 hmmmm if you want an object oriented library, make your base API in C and add the OO bits in an include file as a thin wrapper instead
07:23 hmmmm I thought it was LGPL3 or GPL2
07:23 sfan5 tesseract-ocr does it the other way; the base api is c++ and there is some c -> c++ wrapper or something
07:23 hmmmm like, you have a choice
07:23 hmmmm yeah well tesseract-ocr is a piece of shit library just asking for an ABI incompatibility and incredible amounts of frustration when you mix and match compilers for whichever reason
07:25 Zeno` I guess it's not an optional part of the client
07:25 Zeno` how big is it?
07:25 hmmmm libgmp itself is incredibly huge
07:26 hmmmm minigmp is 4000 lines of code in a single .c file
07:26 Zeno` mini-gmp is just one file?
07:26 est31 140k lines
07:26 est31 yes mini-gmp only has two files: header and c
07:27 sfan5 this seems to be exactly like sqlite3; it's going to be a required dependency sooner or later
07:27 sfan5 why not make it like that from the start?
07:27 Zeno` well I guess the first thing to make sure of is that it compiles with mingw and msvc
07:27 Zeno` yeah I thought est31 was including the entire libgmp source
07:27 Krock It's possible to compile a .lib and .dll with mini-gmp
07:28 Zeno` this mini-gmp is more palatable
07:28 hmmmm he did at first until i pointed out there is a mini version
07:28 est31 yup
07:28 hmmmm this comes at a cost.... the mini version is much slower
07:29 hmmmm but yeah generally it's a good idea to bundle libraries because then things simply work
07:29 est31 original gmp takes 40 ms total computation time (client+server), mini-gmp takes 1200ms
07:29 hmmmm it might not be nice from an organizational standpoint though
07:30 Zeno` I don't think performance is an issue
07:30 hmmmm it's okay for clients
07:30 Zeno` I've changed my mind. Now it's not the full-blown library I'm more inclined to vote towards inclusion
07:31 Zeno` I assume the prototypes are the same though, so those *with* libgmp can use it?
07:31 hmmmm yes
07:31 Zeno` well, everyone is happy. Those wanting performance can use the "unsupported" libgmp :D
07:32 hmmmm oh btw est31 did you ever figure out that undefined reference problem
07:32 est31 ?
07:32 est31 what do you mean
07:33 hmmmm i thought you were having an issue compiling on linux because of some undefined reference error
07:33 est31 ah yes
07:33 est31 I have fixed it
07:33 * est31 tries to remember how
07:34 sfan5 why do people keep messing with shit
07:35 sfan5 the windows builds are now names Minetest-something instead of minetest-something
07:35 est31 it has been my own mistake
07:35 est31 sfan5, its most likely SN's build system cleanup patch
07:36 est31 so hmmmm the issue was that there were some functions that I have added during porting
07:37 est31 mostly thin wrappers around functions that are more powerful in gmp than in openssl
07:37 est31 so that I don't have to copy-paste all those invocations
07:38 est31 now the problem has been that I've declared them but forgotten to add "static".
07:38 sfan5 SN also made some but not all endif(USE_CURL) be endif()
07:38 sfan5 it's less readable
07:38 hmmmm ahh okay
07:38 hmmmm makes sense
07:39 est31 somehow gcc compiled them anyway, but clang didnt.
07:40 Zeno` I didn't think inline static made sense... but I guess it does :3
07:41 est31 ?
07:42 Zeno` https://github.com/minetest/minetest/pull/2620/​files#diff-765e4fc5d8245871d9a5a4db834e4631R217
07:42 hmmmm haha
07:42 hmmmm oh gosh
07:42 Zeno` I didn't know that was possible. *shrug*
07:42 hmmmm "* sizeof(char)" and "if (ptr) free(ptr)" are sure signs of a C novice =]
07:43 sfan5 hm
07:43 Zeno` I commented two minutes ago on if (ptr) free(ptr) :P
07:43 sfan5 iirc free does not need to take a null ptr
07:43 sfan5 anyway
07:43 est31 seen that comment
07:43 sfan5 i told SN that it's stupid to have project(Minetest) and use lowercase everywhere
07:43 sfan5 a lowercase function*
07:44 err404 joined #minetest-dev
07:45 Zeno` I'm not sure about these casts, est31
07:45 hmmmm those aren't his casts
07:46 Zeno` e.g. the one on line 228 of string.cpp
07:46 Zeno` oh?
07:46 hmmmm oh
07:46 hmmmm i thought you meant the casting-const-away casts in srp.c
07:46 Zeno` ah, no
07:47 hmmmm those look weird but they're the CSRP author's attempts to make SRPUser a completely opaque data type
07:47 err404 joined #minetest-dev
07:48 hmmmm in any case I feel like mini-gmp shouldn't be in util/
07:48 hmmmm maybe its own directory huh
07:48 Zeno` yeah
07:48 est31 good idea
07:48 kilbith joined #minetest-dev
07:49 hmmmm eventually i'd like to move these all into a third_party/ directory or something
07:49 Zeno` yep
07:49 est31 what about the sha function files I added?
07:49 Zeno` nice work btw est31
07:49 hmmmm that's okay because there's already a sha1 in util
07:49 hmmmm yeah nice work
07:49 Zeno` ok, I'll bbl
07:50 hmmmm i need to look at it better
07:50 hmmmm it looks great so far though
07:51 Taoki[mobile] joined #minetest-dev
07:51 est31 I have to do some polishing still
07:52 est31 eg freeing some buffers given to csrp
07:52 hmmmm i don't think the SRP things should be in util/string.cpp at all
07:52 hmmmm that's definitely out of place
07:52 est31 its where translatepassword is
07:52 hmmmm translatepassword shouldn't be there either
07:53 hmmmm like wtf that has nothing to do with strings
07:53 hmmmm that has to do with auth
07:53 est31 also those functions only do the verification key calculation
07:53 est31 so make new file util/auth.cpp?
07:53 hmmmm maybe there should be a new file, like util/
07:53 hmmmm yeah.
07:53 est31 ok
07:53 hmmmm that's exactly what i was in the process of typing
07:53 est31 :D
07:58 cib0 joined #minetest-dev
08:01 Yepoleb_ joined #minetest-dev
08:16 Taoki[mobile] joined #minetest-dev
08:37 kilbith joined #minetest-dev
08:37 chchjesus joined #minetest-dev
08:44 ElectronLibre joined #minetest-dev
08:49 jin_xi joined #minetest-dev
08:58 nrzkt joined #minetest-dev
09:26 VargaD joined #minetest-dev
09:36 cib0 joined #minetest-dev
09:41 MinetestForFun joined #minetest-dev
09:42 MinetestForFun_ joined #minetest-dev
09:57 Taoki joined #minetest-dev
09:59 selat joined #minetest-dev
10:30 err404 joined #minetest-dev
10:34 blaze joined #minetest-dev
10:40 est31 now why is this happening: https://travis-ci.org/minet​est/minetest/jobs/59017470
10:40 est31 nothing changed
10:40 est31 ok I've moved mini-gmp into its own directory
10:40 est31 but still
10:41 Fritigern joined #minetest-dev
10:41 est31 only theory I have right now is that it didn't work all the time
10:41 est31 and always used mini-gmp
10:46 est31 I mean the invocation of the compiler *does* get a -lgmp
10:47 est31 and the compilers man pages sais the compiler then searches for .a files in standard paths
11:14 leat joined #minetest-dev
11:29 est31 ok travis successfully builds now .. USING REAL LIBGMP FOR THE FIRST TIME
11:51 SudoAptGetPlay joined #minetest-dev
12:02 cd2 joined #minetest-dev
12:02 cd2 hi
12:04 cd2 whats the type of client->getNodeDefManager()->get(n).name? (game.cpp - class Game)
12:05 nrzkt std::string (struct ContentFeatures)
12:06 cd2 thx
12:07 cd2 but, Why dont that works? if(nodedef->get(n).name == (std::string)"default:stone")
12:08 nrzkt strcmp ?
12:08 est31 first you dont need the cast
12:08 est31 in c++ you can directly compare using ==
12:09 cd2 est31 : i know
12:10 cd2 oh, right - I didnt mind strcmp :)
12:11 est31 strcmp isnt c++ style
12:13 nrzkt it's libc, C not C++
12:23 cd2 yay, strcmp works
12:24 cd2 It works very very very great XD, thank you for helping
12:26 est31 strcmp is good for 0 terminated strings
12:26 est31 but not when the string holds binary data
12:26 est31 then == is better
12:26 est31 I mean the only working one
12:29 MinetestForFun joined #minetest-dev
12:50 Zeno` joined #minetest-dev
12:54 ElectronLibre joined #minetest-dev
13:01 Zeno` nrzkt, the tilt/damage flash bug isn't fixed yet? :(
13:01 nrzkt *i was in holiday in a foreign country
13:02 Zeno` I don't recall you asking permission to have a holiday away from minetest
13:02 Zeno` :P
13:02 nrzkt tsss :)
13:28 cib0 joined #minetest-dev
13:30 SopaXorzTaker joined #minetest-dev
13:55 kilbith joined #minetest-dev
14:09 selat joined #minetest-dev
14:45 blaze joined #minetest-dev
14:45 ElectronLibre Hello, I need opinions on game#494 which I crafted from Ombridride's one (game#492) and improved as PilzAdam told Ombridride to, please (I'm working on the PR because he is occupied).
14:45 ShadowBot https://github.com/minetes​t/minetest_game/issues/494 -- Simplified flowers registration by LeMagnesium
14:45 MinetestForFun joined #minetest-dev
14:47 jin_xi joined #minetest-dev
15:08 SudoAptGetPlay joined #minetest-dev
15:14 Calinou joined #minetest-dev
15:33 kilbith joined #minetest-dev
15:50 hmmmm joined #minetest-dev
15:53 neoascetic joined #minetest-dev
16:04 ElectronLibre joined #minetest-dev
16:18 MinetestForFun joined #minetest-dev
16:34 SopaXorzTaker joined #minetest-dev
17:02 Miner_48er joined #minetest-dev
17:22 err404 joined #minetest-dev
17:28 nrzkt est31: deployed with u16 could be great. u8 is too low, if we are doing a new protocol it's better i think
17:28 nrzkt to prevent a very very future incrementation
17:29 est31 I'm only doing the srp
17:30 nrzkt okay
17:30 est31 I also make the client send the v25 packet, but only if v25 is activated
17:31 est31 so we can freely change v25
17:31 est31 until we think its stable
17:31 est31 then we increment the number
17:31 est31 and everybody can test it
17:31 est31 then we wait until everybody has tested
17:31 est31 then we release :)
17:32 nrzkt i will add more network changes with this protocol next weeks, low layer modifications
17:42 jin_xi #2605 is a 2 line change to fix a bad bug with particle spawners. can someone please review?
17:42 ShadowBot https://github.com/minetest/minetest/issues/2605 -- fix particles oversight by obneq
17:43 leat joined #minetest-dev
17:44 est31 can you provide some code to reproduce the issue?
17:44 est31 ShadowNinja has labeled it unconfirmed
17:45 jin_xi http://paste.ubuntu.com/10845979/
17:45 VanessaE https://github.com/VanessaE/homedecor_modpack/c​ommit/fb768d039dd7277e9f8ad56014485618f67d8eb7
17:45 VanessaE at line 185, set the time to zero
17:46 VanessaE (this commit is waiting for some code by me, and for jin's fix so that I can set the time to 0 for real)
17:47 jin_xi try my test mod, set time to 0 to see this bug
17:47 jin_xi also try some expiring spawners and see server ids never freed
17:47 jin_xi #2602 adresses this issue if you want to confirm that as well
17:47 ShadowBot https://github.com/minetest/minetest/issues/2602 -- move particle spawners to env by obneq
17:51 nrzkt the code seems correct, because the particle is created but never used... also this is a memleak because we affect particle without deleting it
17:57 est31 no there is no leak
17:57 est31 look at particles.cpp line 376
17:58 jin_xi non expiring spawners new particles are never added to update queue atm so they are never expired
17:58 jin_xi so placing spawner with time 0 results in growing blob of non moving particles
18:00 jin_xi just compare lines 271 and 311 and immediately after new Particle()
18:08 jin_xi http://i.imgur.com/mcA6cMi.png here is what i get from a single non expiring spawners after half a minute
18:08 jin_xi note the fps
18:09 est31 I can confirm now
18:12 Krock joined #minetest-dev
18:13 est31 fine
18:13 est31 Have tested it and works
18:13 est31 nrzkt, can I push it?
18:13 nrzkt you can, no problem for me
18:14 VanessaE jin_xi: *looks at your FPS*  wat?
18:14 VanessaE ouch
18:15 jin_xi well particles keep accumulating
18:16 leat joined #minetest-dev
18:16 est31 pushed
18:16 jin_xi thank you
18:16 kilbith i tested your 1st attempt for the irrlicht particles...
18:16 kilbith 1st punch, *boom* : -30 FPS
18:18 nrzkt irrlicht particles are very slow on intel GPU. With my AMD GPU it's a little bit better but i think needs a little bit more optimization
18:18 jin_xi weird its fine on my integrated intel
18:19 est31 next time: more explanations, and perhaps sample code would get it merged faster
18:20 nrzkt before merge i will test it on some GPU to be sure
18:22 VanessaE when I tested the original irrlicht particles on my HD6870 GPU at the time (if I remember right), I had to put several thousand particles on-screen before my FPS started to drop below 60
18:22 VanessaE of course that's intel integrated, but you get the point.
18:22 jin_xi est31: bug was not reported for 6 months, fix was merged plenty fast
18:23 jin_xi but you are right. im gonna attach that code to the other issue and also my trivial pathfinder mod to that issue
18:24 leat joined #minetest-dev
18:55 leat joined #minetest-dev
19:09 SudoAptGetPlay joined #minetest-dev
19:28 OldCoder joined #minetest-dev
19:50 est31 joined #minetest-dev
20:05 est31 hmmmm, found out why wikipedia states that "equivalence" wrongly
20:06 est31 or at least a possible explanation
20:06 est31 there is a website by srp creator thomas wu: http://srp.stanford.edu/ndss.html
20:06 est31 it sais "user passwords are stored in a form that is not plaintext-equivalent to the password itself, so an attacker who captures the password database cannot use it directly to compromise security and gain immediate access to the host"
20:07 est31 That is right, and I guess he has written it because at the time, other protocols were used, where there was a similar requirement
20:07 est31 for example there is one protocol, where the password is directly hashed with the challenge
20:07 est31 pure stupidity, as you can't even store the hash, but have to store the password in clear
20:08 est31 people came up with an ugly hack, not storing the hash, but the internal SHA-1 state
20:08 est31 before finalisation and so
20:08 est31 thats even worse then SHA
20:08 est31 just wanted to share.
20:17 est31 btw, somehow this srp requires vast amount of space
20:17 est31 so you get this in auth.txt:
20:17 est31 est31:$1$eEpP+cmXlMWWza05zJbQZW4AAABkAAAAIQAAAA​AAAAA$LX4/UNn38ZWacODx82U3ZecqSa1SXQF+mKtyEDHJB​qDc7mnnkl0OVoVQ9HNAfq52iryMI4eV3+liaR4K2COMD3WL​oy60AxQQigBxBh4TB0E2Z+PGKqIvsjKpFuCRaT88URqqZzc​I8SK+n33sqQ5LenABPmDvIUCoxF24xLow2p627HOECmkRy2​MVq+CcJbsm5ufOAzK4y1qcocJcUaMIn4HbedO9aY+bOy41Y​HPMny35RTh5czSUKbiUDVcJgMPHHfm1KcdIDnRG7Cn3xEzw​f6TVKejFS8Enf3XRvj6TigTfgDU69K2ggLldai6xgPSSQ0S​l3FfhBykjESHzzkfUOw:interact,shout:1429388138
20:17 VanessaE O_O
20:24 paramat joined #minetest-dev
20:25 est31 so it *can* be lowered, but I don't know how secure it would be then
20:25 est31 usual tradeoff
20:25 est31 perhaps we could make a shadow.txt containing the auth data instead?
20:26 est31 but I guess thats another PR
20:26 est31 because the code is quite well separated
20:26 est31 inside lua
20:28 paramat sfan5, ShadowNinja, please can i push game#493 later?
20:28 ShadowBot https://github.com/minetes​t/minetest_game/issues/493 -- Default: Add river water by paramat
20:31 luizrpgluiz joined #minetest-dev
20:33 sfan5 paramat: aside from having a different water for rivers being inconvenient, seems ok
20:39 MinetestForFun joined #minetest-dev
20:45 ShadowNinja paramat: I don't like the seperate water node.  I'd prefer a flag in param2 or something like that.
20:46 Calinou yeah, keep it to 1 water node
20:46 Calinou I'd rather see acid/slime
20:49 paramat okay i'll try to figure out how to do that with param2
20:49 paramat if that's possible
20:50 paramat although it may need to be a separate node because the biome API needs to recognise it and replace it
20:51 ShadowNinja All my cleanups are starting to conflict with each other :-(
20:52 hmmmm shadowninja, they're too large
20:52 hmmmm i'm sorry'
20:53 ShadowNinja Yeah.
20:53 hmmmm maybe if they were smaller easily reviewable chunks
20:53 hmmmm and they'd have less of a chance to conflict too
20:55 paramat param1 and param2 are in use for water, i think it will have to be a separate node, since the biome API uses a voxelmanip, i think this will be simpler
20:55 paramat and this will only be used in new mapgen, not mg5/6/7
20:55 leat joined #minetest-dev
20:56 ShadowNinja hmmmm: Have you checked https://github.com/minetest/minetest/pull/2628 yet?
20:57 hmmmm for like 5 minutes
20:57 hmmmm 64 files changed, lol
20:57 ShadowNinja (fixing Android build now, aparently the page that said that __android_log_text was a thing lied)
20:57 hmmmm even if that were mergable, it'd break everybody elses' pull requests
20:57 paramat i think a separate node will be faster to process during mapgen, due to having a different content id
20:58 hmmmm shadowninja, maybe it would be a smarter idea to keep these __FUNCTION_NAME and infostreams as aliases and make gradual changes
20:58 ShadowNinja hmmmm: It doesn't break much, just debug.*, log.* and parts of main.cpp mostly.
20:59 hmmmm but you've got to ask; why change infostream to warningstream to begin with?  it doesn't seem like it's really helping minetest that much
20:59 ShadowNinja hmmmm: infostream and friends are still around.  I don't think I've seen __FUNCTION_NAME used in any PRs, and it's a trivial fix if it is.
21:00 hmmmm yeah but when PRs go to merge they're going to have a merge conflict
21:00 hmmmm because it's going to expect infostream
21:00 est31 my PR has
21:00 hmmmm and maybe not used directly
21:00 ShadowNinja hmmmm: I just *added* warning stream.  It's used instred of *stream<<"WARNING: ...
21:00 hmmmm and totally fine
21:00 ShadowNinja infostream is still around.
21:00 est31 yes
21:00 hmmmm changes need to be gradual though
21:01 hmmmm this big change-everything strategy simply does not work
21:01 hmmmm it worked when you had a small project
21:01 ShadowNinja You can't really fix the worst parts without breaking other things.
21:02 hmmmm i'm concerned about breaking PRs
21:02 ShadowNinja My logging PR is a minimal-conflict PR though.
21:02 hmmmm how about we clear out the PR backlog and then we'll see what we can do with large, sweeping changes
21:03 hmmmm man, 64 changed files and you say it's minimal conflict
21:03 hmmmm but how can you really be sure about that
21:03 hmmmm 64 changed files!
21:03 hmmmm 64!
21:03 hmmmm my largest commits never touched more than 16 files
21:03 ShadowNinja Mostly just __FUNCTION_NAME and *stream<<"WARNING: ->warningstream<<" changes.
21:04 hmmmm so let's say we did merge this right away
21:04 hmmmm what do we gain by having it all at once
21:04 hmmmm and what happens when people start complaining about conflicts
21:05 hmmmm and how about for src/socket.cpp, you changed the dstream to a warningstream
21:05 hmmmm that's really not a warning at all
21:06 hmmmm I'm worrying this might make things too verbose in general
21:06 ShadowNinja hmmmm: Well, the __FUNCTION_NAME change isn't essential, and i could leave the warning thing.  socket.cpp states it's a warning, if it isn't it shouldn't have said that.
21:06 sfan5 what's with that face_position_cache.cpp?
21:06 sfan5 how does it relate to logging?
21:07 kahrl_ I was about to ask that, same for the util/numeric.h changes?
21:07 hmmmm you removed CONTAINS()?
21:07 hmmmm :(
21:07 hmmmm that could be useful to other source files
21:07 ShadowNinja sfan5: I moved it out of numeric.h because it was pulling in some more headers, it's only used in two places, and it's rather Minetest-specific.
21:08 sfan5 you should name your PRs better
21:08 sfan5 what you've done is clean up 2 or 3 unrelated things
21:08 hmmmm but I mean other than those things the Pr is fine
21:08 sfan5 but your PR is called "Clean up logging"
21:08 proller joined #minetest-dev
21:08 ShadowNinja hmmmm: It pulled in <algorithm> in every file that used numeric and it's only used in two files.
21:08 hmmmm it shouldn't need <algorithm> included unless the file uses it
21:08 ShadowNinja sfan5: I sometimes get a bit carried away, yes.
21:09 hmmmm so keep the macro in the file and move <algorithm> to the files which do use it
21:09 hmmmm that was my bad
21:09 ShadowNinja hmmmm: Seems fair enough.
21:09 sfan5 <hmmmm> maybe if they were smaller easily reviewable chunks
21:09 est31 cleanups are good, but commits should focus on separate things.
21:09 sfan5 try not to put too much into a PR
21:09 hmmmm will changing the meaning of debug_log_level values mess up peoples' config files?
21:11 ShadowNinja hmmmm: For anyone that uses that setting, yes.  I think that it should be changed to a string setting so that levels can be added without breaking everything.
21:11 ShadowNinja I didn't get to that yet though.
21:12 * ShadowNinja will do that now.
21:12 sfan5 ShadowNinja: why s/ARRLEN/ARRAY_SIZE/?
21:12 ShadowNinja sfan5: Readability.
21:13 sfan5 really?
21:13 sfan5 IMO both are equally readable.
21:13 ShadowNinja *shrug*
21:15 sfan5 ShadowNinja: why remove the log-level examples (log.h)?
21:16 ShadowNinja sfan5: Most of them seemed obvious.  I added more concise comments for the others.
21:17 sfan5 ShadowNinja: the only additionall thing is "// Special level that is always printed"
21:18 hmmmm yeah
21:18 hmmmm ARRLEN is just as readable
21:18 ShadowNinja sfan5: And "// In-game actions"
21:18 kahrl_ oh btw, "grep -R 'sizeof.*sizeof' src" reveals a bunch of places that could benefit from using ARRLEN/ARRAY_SIZE
21:18 hmmmm and i need to point out it's not always a global string replace.  changing identifier names with longer ones might push some pieces of code onto new lines
21:18 sfan5 ShadowNinja: that ones was already there before
21:19 sfan5 ShadowNinja: "" /* In-game actions ("celeron55 placed block at (12,4,-5)") */ ""
21:19 hmmmm i vote for changing back to ARRLEN.
21:19 paramat yeah i'm sure river water has to be a separate node, mapgen will be slower and more complex having to set or read a special parameter, the separate content id helps a lot. please can i push? =)
21:19 hmmmm paramat: sure..
21:19 sfan5 <hmmmm> i vote for changing back to ARRLEN.
21:19 * sfan5 too
21:19 est31 *too
21:20 sfan5 est31: /me <text>
21:20 sfan5 (if your irc client does not have that you are probably using netcat ;) )
21:20 est31 lol
21:20 sfan5 https://github.com/minetest/minetest/pull/2628/​files#diff-7ec3c68a81efff79b6ca22ac1f1eabbaR193
21:20 sfan5 i think this is less readable
21:20 est31 yes I know that just didnt use it that time
21:20 sfan5 having an #ifdef in the middle of an expression
21:21 sfan5 just to save some bytes or something
21:23 ShadowNinja O.K.
21:24 hmmmm agreed I hate that
21:24 hmmmm i saw a rather extreme example in porting.h and i want to change it there
21:29 est31 man this is stupid
21:29 est31 I have srp login set up and everythin works then it sais: supplied wrong (SRP) password from address 127.0.0.1
21:30 est31 it means I have to find a way to debug srp itself
21:31 est31 I'll let both hash the verifier first to make sure
21:31 est31 then print that to stdout
21:58 paramat left #minetest-dev
22:04 hmmmm wow I look at all this crap in porting.cpp and I wonder how on earth it actually works
22:04 hmmmm need to resist the urge to make radical changes
22:48 est31 hmmmm, you dislike passing arguments as references, don't you?
22:48 est31 now I have following code: https://github.com/minetest/minetest/pull/2620/​files#diff-2d67c2347848dbd1d1d74e4b8d608694R112
22:48 est31 Its shit
22:48 est31 obviously
22:49 est31 I'm passing pointers to a stack string's c string representation
22:49 est31 now I want to "return" both salt_str and bytes_v_str
22:50 est31 If you know a way thats nicer than references please tell
22:50 est31 (in order to fix)
22:53 est31 oh, having strings makes things more complicated
22:53 hmmmm est31, references are okay
22:54 * est31 sighs and does malloc and free
22:54 hmmmm the rule is to not pass references as arguments that are non-const
22:54 hmmmm const std::string &<foobar> is fine
22:54 est31 yes
22:54 hmmmm std::string &foobar is not
22:54 est31 my question was about the latter one
22:55 hmmmm you want to make it clear that if a variable could be modified, it's going to be a pointer
22:56 est31 either way I'm coping
22:56 est31 either with the = operator
22:57 est31 or ... with the = operator
22:57 est31 :)
22:58 proller joined #minetest-dev
23:08 ShadowNinja I've changed ARRAY_SIZE to ARRLEN, changed debug_log_level to a string setting (with a compatability wrapper for the old format), and changed the game_params.is_dedicated_server thing.
23:09 hmmmm well
23:09 ShadowNinja And changed the socket.cpp windows warning ignoring thing to infostream.
23:09 hmmmm I don't know if the second item there is worth it.  debug_log_level being consistent for people is sort of irrelevant since the majority of the stream levels have been modified
23:10 hmmmm and the number format is good because it's simpler to implement and has a well-defined order to it
23:10 hmmmm my previous statements about reverse compatibility, now that i think about it, are wrong
23:10 ShadowNinja hmmmm: Well, I like it better this way, and at this point it's more work to change it back.
23:10 hmmmm okay then.
23:11 Miner_48er joined #minetest-dev
23:11 ShadowNinja Should I undo the __FUNCTION_NAME thing?
23:12 ShadowNinja I don't think it has any conflicts, but if it'll get this merged faster...
23:12 hmmmm no that's definitely a good change
23:12 ShadowNinja Alright, good.
23:13 ShadowNinja I think I'll add a string_to_level helper or similar, since minetest.log uses pretty much the same code as init_log_streams now.
23:24 Wayward_Tab joined #minetest-dev
23:47 ShadowNinja Should the main thread's logging thread name be changed to "Main" or "MainThread" to be consistent with the casing of the other thread names?

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