Minetest logo

IRC log for #minetest-dev, 2015-10-26

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

All times shown according to UTC.

Time Nick Message
00:05 Supertan1er2 joined #minetest-dev
00:10 proller joined #minetest-dev
00:21 est31 hmmmm, what commits are still open for getting merged?
00:42 Player2 joined #minetest-dev
00:42 paramat left #minetest-dev
01:50 Supertan1er2 joined #minetest-dev
01:59 Taoki joined #minetest-dev
02:26 est31 joined #minetest-dev
02:36 Supertan1er2 joined #minetest-dev
02:48 est31 cheapie, it works for me. Can you re-apply the newest version of the patch and try again?
02:48 est31 regarding your comment https://github.com/minetest/minetes​t/pull/3292#issuecomment-151005935
02:48 cheapie est31: OK, one moment.
02:56 cheapie est31: Seems to be working now.
02:58 hmmmm est31: ??
02:59 est31 in the past days, you had "PTAL"ed some commits of yours
02:59 est31 I wonder which still are "open"
02:59 est31 and unmerged
02:59 est31 etc
02:59 hmmmm uhh, this is the only one that's pushed to github so far:  https://github.com/kwolekr/minetest/commit​/18e0f1f24fb78a228afa837c7358dcb3737b0d03
03:02 est31 why have you only removed the first line
03:02 est31 https://github.com/kwolekr/minetest/commit​/18e0f1f24fb78a228afa837c7358dcb3737b0d03#​diff-9444313da8be4290e9304a4b1faed804L135
03:02 est31 and not also modified the second
03:02 hmmmm because NO_MAP_LOCK_REQUIRED and GET_ENV_PTR is an oxymoron
03:03 hmmmm except for very special cases, in which GET_ENV_PTR_NO_MAP_LOCK_REQUIRED is used
03:03 hmmmm i.e. atomic env access
03:03 est31 yeah, why you not do that
03:03 hmmmm because a map lock is indeed required there
03:03 est31 ah you say it was wrong
03:03 est31 ok then
03:04 est31 wow that one call really did get the environment just for the nodedef?
03:04 est31 things exist...
03:05 est31 well, the commit looks good
03:05 est31 +1
03:05 est31 and congrats to commit 5.1k
03:06 hmmmm wow it was commit 5k not too long ago
03:13 hmmmm how about this one:  https://github.com/kwolekr/minetest/commit​/fb3ee2d50023435904af157343d2e1b5f14223c3   ??
03:14 est31 there is another return 0 just above
03:14 hmmmm so?
03:14 est31 ahh
03:15 hmmmm oops, I see a really bad mistake there
03:15 est31 hrmm, doesn't LuaError miss to print a stacktrace?
03:15 hmmmm no, LuaError does print a backtrace
03:15 est31 only the error thats it
03:15 est31 also must be thrown ofc :)
03:16 est31 guess thats the error you saw
03:16 est31 mistake*
03:16 hmmmm yup
03:16 hmmmm https://github.com/kwolekr/minetest/commit​/d69ef6acd386ccfd5ff456002591b97b92a26956
03:16 est31 I'll first test what LuaError does
03:18 est31 in the meantime can you look at https://github.com/est31/minetest/commit/​454923c1b198abd98c6d8f95d2cae6eb79fe0869
03:19 hmmmm does any compiler throw a "forced boolean conversion" warning for "return num_modules_failed;"?
03:20 hmmmm if so you can (bool) or !!
03:20 est31 ok will cast it
03:20 hmmmm well no
03:20 hmmmm I'm just asking if you've seen that warning, because sometimes that happens for MSVC
03:20 hmmmm err you didn't make a PR for this did you
03:21 hmmmm http://stackoverflow.com/questions​/1847860/why-is-there-a-performanc​e-warning-on-cast-pointer-to-bool
03:21 hmmmm seems like you can't do (bool)
03:22 est31 travis makes no warning
03:22 est31 ermm
03:22 est31 travis makes no msvc compiles
03:23 hmmmm damn
03:23 est31 and no didnt make a pr
03:23 est31 lets wait till krock comes along and complains :)
03:23 hmmmm ok :)
03:23 est31 how can you provocate that LuaError?
03:23 est31 the one you added
03:24 hmmmm by calling get_mapgen_object() in a non-mapgen thread :)
03:24 est31 doesnt work
03:24 hmmmm hmmm?
03:24 hmmmm what do you mean
03:24 est31 doesnt crash
03:24 est31 it works if it crashes :)
03:24 hmmmm can you paste the code?
03:25 est31 print("The object is: " .. dump(minetest.get_mapgen_object()))
03:25 est31 and that carelessly put into the init.lua of a mod
03:26 hmmmm oh that's because it's failing first
03:26 hmmmm you have an invalid mapgen object specified
03:26 hmmmm so you're not even querying anything
03:26 hmmmm try minetest.get_mapgen_object("voxelmanip")
03:26 est31 yeah
03:26 est31 no stacktrace
03:26 est31 so it works
03:26 est31 but is broken
03:27 hmmmm hang on let me see this
03:27 est31 I think only lua_error makes a stacktrace
03:27 est31 or luaL_error cant remember
03:27 est31 lemme see the doc
03:28 hmmmm works for me
03:28 est31 in the sense that you have a stacktrace??
03:28 hmmmm are you using vanilla Lua, perhaps?
03:29 est31 yes
03:29 hmmmm hmmmmm.
03:29 hmmmm so LuaError is broken in non-LuaJIT?
03:29 hmmmm great...
03:29 est31 retrying with jit
03:30 est31 but first I try lua_error
03:30 hmmmm oh
03:30 hmmmm that's because regular Lua doesn't have the script_execution_wrapper()
03:31 hmmmm this has been broken for quite some time i guess
03:31 est31 okay luaL_error prints a stacktrace
03:33 est31 whats that
03:33 est31 script_execution_wrapper()
03:33 est31 grepping brings no results
03:33 hmmmm s_base.cpp
03:34 est31 and googling points to a discussion between us on this channel from august
03:35 est31 yeah its exception not execution.
03:39 est31 +1 to your second commit as well
03:44 est31 ok now to my commit
03:45 est31 https://github.com/est31/minetest/commit/​5f342aa0154e59503fbb22e46ac95ef655aaad1a
03:45 hmmmm +1
03:46 est31 I've done it because I modify fatal_error_fn in the ncurses pr
03:46 est31 and abort() isn't enough then anymore, because it can leave the terminal in an inconsistent state
03:48 est31 I had to make the terminal object global because of this
03:48 est31 so that it can ask it politely to stop
03:49 est31 I could have simply included ncurses.h in debug.cpp as well and manually called endwin(), there even was a function provided by ncurses that tells one whether one has to call endwin()
03:49 est31 but that would lead to race conditions
03:50 est31 the terminal runs in its own thread, and even if it ran in e.g. the server thread, there are still other threads that can call FATAL_ERROR
03:53 est31 it now even prits the error of FATAL_ERROR in the console after the terminal terminated
03:56 est31 that could have been solved as well by e.g. printing out the scroll back buffer of the terminal, but as you do already need to wait for the thread to call endwin(), its not hard to ask it to restore the old situation with errorstream being logged to stderr, and to simply print the msg to errorstream like before
03:56 est31 that was the simplest setup I could come up with
03:57 est31 I think only details remain now
04:06 est31 yeah, pr looks ok for a review
04:08 est31 #3292 is open for review!
04:08 ShadowBot https://github.com/minetest/minetest/issues/3292 -- Add server side ncurses terminal by est31
05:32 OldCoder joined #minetest-dev
06:43 nrzkt joined #minetest-dev
06:47 Hunterz joined #minetest-dev
07:00 cib0 joined #minetest-dev
07:22 nrzkt joined #minetest-dev
07:27 Calinou joined #minetest-dev
07:42 OldCoder joined #minetest-dev
07:47 hmmmm https://github.com/kwolekr/minetest/commit​/306b0670916fee8d0316a69f161c7a9b14e9516d
07:47 hmmmm trivial reorganization - pushing in 20 minutes
07:48 nrzkt LGTM
08:05 hmmmm https://github.com/kwolekr/minetest/commit​/bc0318d2fc0d435e20b1186203ce2908162241ca
08:05 hmmmm PTAL
08:06 nrzkt hmmmm, +1
08:33 cib0 joined #minetest-dev
08:44 nrzkt joined #minetest-dev
08:57 PilzAdam joined #minetest-dev
09:18 DFeniks joined #minetest-dev
09:22 Gael-de-Sailly joined #minetest-dev
09:23 Gael-de-Sailly joined #minetest-dev
09:26 Megaf joined #minetest-dev
09:55 cib0 joined #minetest-dev
09:56 ElectronLibre joined #minetest-dev
10:08 proller joined #minetest-dev
10:10 proller joined #minetest-dev
10:43 Darcidride joined #minetest-dev
11:18 Lunatrius` joined #minetest-dev
12:40 CraigyDavi joined #minetest-dev
13:24 VanessaE joined #minetest-dev
13:51 Lunatrius joined #minetest-dev
14:10 BlockMen joined #minetest-dev
14:15 BlockMen pushing in 10 minutes if noone objects https://github.com/BlockMen/minetest/commi​t/2c25107c31c0b0e72fe4a48b0dc3f5b09a0b46aa
14:17 kilbith joined #minetest-dev
14:18 kilbith https://github.com/minetest/minetest/blob/​master/builtin/settingtypes.txt#L407-L411
14:18 kilbith ^ fail in descriptions
14:18 kilbith width is horizontal size
14:18 kilbith height is vertical size
14:30 zupoman joined #minetest-dev
14:30 zupoman joined #minetest-dev
14:32 kilbith also i'm getting lotsa of truncated strings on Weblate, here's a sample : https://lut.im/kUcJAZzE9u/c0dfkiteWz8S2jnt.png
14:32 zat joined #minetest-dev
14:33 PilzAdam kilbith, 1) click on "comments" 2) read 3) ... 4) Profit!
14:34 kilbith ah, i see
14:35 kilbith but weblate did not updated automatically
14:37 PilzAdam yes, we need to run updatepo.sh to update weblate, but that has some problems currently
14:37 PilzAdam est31 is working on this
14:50 est31 joined #minetest-dev
14:56 est31 BlockMen, how does it print the chat msg twice for you?
14:56 est31 is it really like
14:56 est31 <ACTION> 2015-10-26 15:56:17: [Server] CHAT: <est_test> hi!
14:56 est31 <ACTION> 2015-10-26 15:56:17: [Server] CHAT: <est_test> hi!
14:57 est31 or somehow different?
14:57 est31 does it also happen when you send chat msgs from the client?
14:57 BlockMen est, the ingame chat, so
14:57 BlockMen <test1> hi
14:58 BlockMen <test1> hi
14:58 BlockMen the console output is correct
14:59 est31 ah I see what you mean
14:59 BlockMen for me it looks like either it is printed twice on client OR the server sends the msg back to sender aswell
14:59 est31 didnt change the client
15:00 est31 it is bc the server sends it back
15:00 est31 i have changed that piece of code and it seems I forgot the negated check
15:01 hmmmm joined #minetest-dev
15:02 nrzkt est31: the main problem is that we show the client message on the client whereas server didn't receive it. I think this should be changed in next protocol version, showing it only for 0.4.13 servers and less and not for upper, and add the properly handling serverside
15:03 BlockMen nrzkt, what is wrong with showing it always on client?
15:04 est31 ^
15:06 est31 okay BlockMen bug is fixed
15:07 nrzkt BlockMen, the main problem is if server timeouts, client can test the connection with chat for some minetest, continuing to dig and thinking server is up. I saw that many times with my players or palyers on other servers
15:08 nrzkt I think server should broadcast messages itself and client should show the message in this case only
15:08 est31 I always use /status to check for a server to be up
15:08 nrzkt yes you can do many things
15:08 nrzkt but if you are really talking
15:08 est31 I think irc does it that way too
15:08 nrzkt you didn't see that the server is disconnected
15:08 * est31 disconnecting internet to see whether hes right
15:08 nrzkt a player don't use server commands
15:09 est31 zzzzz
15:09 est31 zzz
15:09 est31 z
15:09 est31 z
15:09 est31 zzzzzz
15:09 nrzkt a player will talk and lost messages without seeing that they are not received on the server
15:09 est31 yup, hexchat shows it
15:11 BlockMen nrzkt, other way you loose it aswell, and cant even reread or make a snap of
15:11 BlockMen since its only printed in case it is received. so we need a better way of notifying players they are disconnected, not hiding chat msg IMO
15:13 BlockMen or we add a timeout for msg response of server and print an information if its not received within the timeout
15:15 nrzkt BlockMen, on every MMORPG game when you are disconnecting due to server timeout (loosing internet connection) your messages are not shown into chat
15:18 BlockMen nrzkt, so? how is it worse to see what you have typed and get notifyed that its lost than just seeing nothing?
15:20 nrzkt BlockMen, fundamentaly it's same, but it's more complicated to add a hack to timeout handling to verify if packet is a chat message to notify the player than not showing the message :)
15:22 BlockMen nrzkt, well, i disagree. IMO it better this way, hiding it is not user friendly
15:27 nrzkt BlockMen, did you play other multiplayer RPGs ? :) changing user feeling on problems is not user friendly
15:27 Hunterz joined #minetest-dev
15:28 BlockMen yes, and i marked it clearly as my opinion
15:30 BlockMen and just because a (claimed) majority like to jump from a bridge, do i have to follow them because it the (claimed) majority?
15:30 BlockMen +s
15:30 nrzkt Don't use those stupid analogies because they are broken. :)
15:30 nrzkt We are talking about gaming long time implemented things
15:31 nrzkt if you play with a platform game did you want to change the pause button to button left because it's better ?
15:31 BlockMen is it "the best" just because it is long time used?
15:31 BlockMen most PC are using windows, why don't you use it?
15:32 est31 yeah, thats the more drastic example than the bridge one
15:32 est31 xD
15:32 BlockMen lel
15:36 kilbith left #minetest-dev
15:37 nrzkt BlockMen, i'm not using it because i don't need it, simple :p i need some tools only present on linux :)
15:39 BlockMen nrzkt, ok, its not the best for you. but is it the best in general?
15:43 nrzkt if you ask me 10 years ago i would tell you no. But for the majority of end users, i think using a good distrubtion for new users like fedora/ubuntu/mint is enough for them. But if we talk about the marketing of Linux, yes it's the nether :p
15:48 Player2 joined #minetest-dev
15:48 BlockMen wait, i ment is windows the best in general. and we are not talking about marketing. the question is: is win the best OS since it is long time used and by a majority?
15:49 BlockMen *PC only^
15:57 CraigyDavi joined #minetest-dev
16:03 iqualfragile joined #minetest-dev
16:10 CraigyDavi joined #minetest-dev
16:15 Calinou maybe stop calling it "win" if you believe it's bad
16:25 rom1504 losedows
16:26 rom1504 http://www.urbandictionary.c​om/define.php?term=Losedows ah that's really a thing
16:30 BlockMen Calinou where did i say that i believe its bad?
16:36 Calinou nowhere :)
16:38 BlockMen hehe
16:38 Robby joined #minetest-dev
16:42 Amaz joined #minetest-dev
16:50 neti_netwalker left #minetest-dev
16:50 nrzkt joined #minetest-dev
16:52 iqualfragile_ joined #minetest-dev
17:08 Hunterz joined #minetest-dev
17:14 iqualfragile joined #minetest-dev
17:18 Krock joined #minetest-dev
17:40 PilzAdam joined #minetest-dev
17:41 Nemin joined #minetest-dev
17:52 iqualfragile joined #minetest-dev
18:01 rubenwardy joined #minetest-dev
18:12 ElectronLibre joined #minetest-dev
18:25 proller joined #minetest-dev
18:28 BlockMen can i merge #3040? it adds "standing_on_object" variable to collision check to stop playing footstep sound when stading on an object instead of the ground.
18:29 ShadowBot https://github.com/minetest/minetest/issues/3040 -- Fix jittering sounds on entities (fixes #2974) by BlockMen
18:31 est31 joined #minetest-dev
18:32 est31 anybody wants to review #3292
18:32 ShadowBot https://github.com/minetest/minetest/issues/3292 -- Add server side ncurses terminal by est31
18:34 est31 idk about 3040
18:34 est31 PilzAdam says its not proper.
18:35 BlockMen est31, https://github.com/minetest/minetes​t/pull/3040#issuecomment-130403541
18:36 nrzkt est31: can you fix the password change in SRP please ?
18:36 est31 well, better one bug fixed than none fixed.
18:36 BlockMen you will need this even if collision with objects is fixed since collision currently returns "standing on object == standing on ground"
18:36 est31 nrzkt, what do you mean
18:37 nrzkt https://github.com/minetest/minetest/blob/mas​ter/src/network/serverpackethandler.cpp#L1211
18:37 est31 well thats the legacy way to change the password
18:38 est31 its disabled so that you don't set an insecure hash
18:38 nrzkt i understand but what about changing the password now ?
18:38 est31 it works via another mechanism
18:39 est31 lemme look
18:39 nrzkt ty
18:39 est31 the basic concept is that the client just authenticates again
18:40 est31 and then sends a TOSERVER_FIRST_SRP
18:40 est31 here: https://github.com/minetest/minetes​t/blob/master/src/client.cpp#L1191
18:40 est31 its called sudo mode
18:41 est31 named after a similar mode in github.
18:42 PilzAdam BlockMen, I guess #3040 is good (haven't looked at the code, though)
18:42 nrzkt okay, but how do we change our password in reality ?
18:42 ShadowBot https://github.com/minetest/minetest/issues/3040 -- Fix jittering sounds on entities (fixes #2974) by BlockMen
18:42 est31 nrzkt, what do you mean?
18:42 PilzAdam BlockMen, but the other thing needs to be fixed, too, in a different PR
18:42 nrzkt now i understand the protocol concept to change the password. But how can i do  this with my client ? :)
18:42 BlockMen PilzAdam, which one?
18:43 PilzAdam BlockMen, that touching_ground jitters when standing on an object
18:44 est31 okay, you are connected to the server, and logged in, player is spawned etc, everything is normal
18:44 BlockMen this is unrelated to this pull, it an issue in collision at all.
18:44 BlockMen the sneaking thing suffers under the same
18:44 BlockMen PilzAdam: ^
18:44 est31 now you have got a set of flags in TOCLIENT_AUTH_ACCEPT : https://github.com/minetest/minetest/blob/​master/src/network/networkprotocol.h#L177
18:44 est31 you store it somewhere in the client
18:45 PilzAdam BlockMen, that's why I said "in a different PR"
18:45 est31 its the same format as the set of flags sent in TOCLIENT_HELLO https://github.com/minetest/minetest/blob/​master/src/network/networkprotocol.h#L167
18:45 est31 now the client choses one of the supported auth methods
18:45 BlockMen PilzAdam ah, i thought you ment an existing one
18:46 nrzkt est31, okay mais where is the trigger on my client, not in the code :)
18:46 est31 as the user already exists, its only AUTH_MECHANISM_SRP, and AUTH_MECHANISM_LEGACY_PASSWORD
18:46 BlockMen ok, i will wait until #3040 passes build and push it then
18:46 ShadowBot https://github.com/minetest/minetest/issues/3040 -- Fix jittering sounds on entities (fixes #2974) by BlockMen
18:46 nrzkt the client change the method in the send password form ?
18:46 PilzAdam BlockMen, someone needs to review the code first
18:46 est31 no, once that authentication is done, the client is in "sudo mode"
18:47 est31 the server sends a packet
18:47 est31 TOCLIENT_ACCEPT_SUDO_MODE
18:47 est31 or, if auth fails, it sends TOCLIENT_DENY_SUDO_MODE
18:47 nrzkt okay
18:47 nrzkt but
18:47 nrzkt how can i change my password as a player, not as a C++ programmer ? :p
18:47 BlockMen PilzAdam this https://github.com/minetest/minetest/pull/3040/​files#diff-746d6155906a987734997d4cad358153R542 and https://github.com/minetest/minetest/pull/3040/​files#diff-9810d3c46db625806881548efdc5fa47R60 and https://github.com/minetest/minetest/pull/3040/​files#diff-efb51cde4bb3c805a95dc31b2b1781d0R347
18:48 est31 ah
18:48 BlockMen just these few lines are relevant
18:49 est31 nrzkt, it is a button if you press esc
18:49 PilzAdam BlockMen, seems good, then
18:49 PilzAdam BlockMen, but don't mix style changes with actual functionality
18:49 est31 second one
18:49 PilzAdam BlockMen, can you split this into 2 commits?
18:49 nrzkt est31, it's in the change password menu then ? :)
18:50 est31 well yeah there you change your password as a player
18:50 BlockMen PilzAdam, wont do.
18:50 est31 BlockMen, you have added the rule yourself
18:50 est31 havent you
18:51 PilzAdam but http://dev.minetest.net/Git_Guidelin​es#Upstream_pull_requests_and_issues
18:51 BlockMen est31, we decided to add. dont you remember
18:51 nrzkt est31, okay thanks :p somes reported they can't change their password
18:51 BlockMen and if anybody want to do that, have fun. i wont fiddle with it again
18:51 BlockMen also, PilzAdam, you didnt split you commit yesterday either
18:52 PilzAdam it didn't contain style changes; and it was arguably one coherent thing
18:52 PilzAdam meh, dunno
18:52 PilzAdam I would +1 if it it's spitted; I'm neutral if not
18:52 PilzAdam *splitted
18:53 est31 we decided to add what?
18:53 BlockMen est31, the rule to wiki.
18:54 est31 yes I know
18:55 BlockMen and i told you then already that im not going to fiddle with this commit again ;)
18:55 BlockMen so, since it passed the build i want merge it now, finally
18:57 est31 PilzAdam, you want to split it up :)
19:01 PilzAdam est31, are there any news on the msgmerge front?
19:02 est31 no
19:03 est31 https://lists.gnu.org/archive/html​/bug-gettext/2015-10/msg00025.html
19:03 BlockMen good, mergin now :)
19:03 Darcidride joined #minetest-dev
19:14 PilzAdam Ihr wollt doch etwa nicht die Franzosen gewinnen lassen, oder? https://forum.minetest.net/v​iewtopic.php?f=3&amp;t=13485
19:14 PilzAdam eh, wrong channel
19:16 BlockMen PilzAdam :D
19:30 Tank joined #minetest-dev
19:36 VanessaE ok I don't know what it is but I'm getting a lot crashes lately, SIGABRT
19:37 est31 every time I hear the word "lately" I'm reminded of a certain game.
19:38 est31 do you have stacktraces?
19:38 VanessaE sure, moment
19:39 VanessaE http://pastebin.ubuntu.com/12973154/
19:40 VanessaE that's at HEAD; the trouble started in the past couple of days, so some commit after Oct 10 best guess
19:40 VanessaE (if I read 'git log -g' correctly)
19:41 VanessaE and that last crash did not save-out the map, either, apparently.
19:41 VanessaE (well, the changed blocks)
19:42 PilzAdam https://github.com/minetest/minete​st/blob/master/src/server.cpp#L505 the line that causes this
19:42 PilzAdam (according to the stacktrace)
19:42 est31 well, that means its an async error
19:42 PilzAdam VanessaE, anything like that in the logs?
19:42 est31 so thrown ... somewhere
19:43 PilzAdam it should log it to errorstream
19:43 est31 Lua: Runtime error from mod 'colormachine' in callback node_on_receive_fields()
19:44 est31 ...etest/games/dreambuilder_game/​mods/colormachine/init.lua:1124: attempt to perform arithmetic on local 'page' (a nil
19:44 est31 rest is truncated
19:44 PilzAdam ah, its in the stacktrace too
19:44 VanessaE huh.  that one was a colormachine error
19:44 VanessaE but that's impossible
19:44 VanessaE that mod hasn't been touched in months
19:44 PilzAdam why does it async?
19:44 VanessaE (for the record:  est31 has access to my servers'
19:44 PilzAdam who added multithreaded mods behind our backs?
19:44 VanessaE (for the record:  est31 has access to my servers' logs when needed)
19:44 est31 I think all lua errors are async errors , no?
19:44 zat joined #minetest-dev
19:45 PilzAdam dunno, but this doesn't look like the standard Lua error handler
19:45 PilzAdam ShadowNinja, ping
19:45 VanessaE I need to fix my "start server" script so that it properly dumps ALL of the servers' output into the moredebug file
19:45 est31 hmmmm has touched the code as well lately
19:45 ShadowNinja Pong.
19:46 VanessaE (only crashdumps are ending up there, regular debug output is in debug.txt)
19:46 PilzAdam ShadowNinja, weird Lua stuff, thats your business
19:46 est31 shortly before the 0.4.13 release
19:52 ShadowNinja "<VanessaE> but that's impossible that mod hasn't been touched in months"  --  A engine change can break mods too.  Also, some bugs seem to like to hide a while and then suddenly pop up a lot for no reason.
19:53 est31 like nazguls
19:53 ShadowNinja est31: Yes, all Lua errors are set with server->setAsyncFatalError() (unless something's changed recently.
19:54 ShadowNinja https://xkcd.com/859/
19:55 BlockMen "Yes, all Lua errors are set with server->setAsyncFatalError()" <- has this a special reason?
19:55 ShadowNinja PilzAdam: Looks like someone did if (async_err) FATAL_ERROR(async_arr);
19:56 ShadowNinja BlockMen: No, that's just the error API that Lua uses.
19:56 ShadowNinja (AFAIK)
19:57 BlockMen ShadowNinja, ok thx
19:57 ShadowNinja PilzAdam: Yep, that's exactly what happened.
19:58 ShadowNinja Someone misunderstood what FATAL_ERROR's for I guess.  Should be throw ServerError(async_err) or so.
19:58 est31 either way will go
19:58 est31 bye all
19:59 ShadowNinja I GTG too.  The fix for ^ should be pretty trivial.
20:16 Gael-de-Sailly joined #minetest-dev
20:24 BlockMen This ok? https://github.com/BlockMen/minetest/commi​t/d3a567cf52fcaff57a7a8a9d1611133ec1489bc3
20:24 BlockMen PilzAdam ^
20:24 VanessaE ok, fixed the bad redirect on the start-server script.
20:24 PilzAdam I guess
20:25 BlockMen means that i should push?
20:28 PilzAdam if it works
20:30 BlockMen it should. you can give me a test case to proove?
20:30 * PilzAdam hands VanessaE over to BlockMen
20:32 * VanessaE adds BlockMen's commit and builds...
20:34 VanessaE deployed to my "Basic" and "Building" servers.
20:34 BlockMen thx
20:34 VanessaE np
20:36 VanessaE and...deployed to the others also.
20:47 rubenwardy joined #minetest-dev
21:19 hmmmm BlockMen, looks good to me
21:20 hmmmm yeah I remember est31 recently changed that to fix something totally different
21:22 hmmmm no nevermind, it's nerzhul:  https://github.com/minetest/minetest/commi​t/768596927458d4d1d4ae7914526311471d242555
21:23 hmmmm I'm pretty sure I didn't sign off on this...
21:23 VanessaE March 17?!
21:23 BlockMen hmmmm, so should i merge?
21:23 hmmmm I don't know let me think more about it
21:25 hmmmm so the original reason for that commit was because a SIGABRT (uncaught exception) propogates all the way to the top of minetestserver
21:27 hmmmm but why is this a problem...
21:28 hmmmm does it not print out the error?
21:29 hmmmm if that is the case, then BlockMen's patch should work fine to resolve the problem
21:29 hmmmm but what is the current problem exactly?
21:30 hmmmm that FATAL_ERROR() only happens on a dedicated minetest server
21:31 hmmmm VanessaE:  what was the problem again??
21:38 VanessaE hmmmm: crashes (SIGABRT) out of nowhere in mods that have been stable for months
21:39 VanessaE http://pastebin.ubuntu.com/12973154/ is an example
21:39 hmmmm maybe there's been a bug introduced in one of the APIs?
21:39 VanessaE earlier today, "mg" mod stopped working
21:39 VanessaE that one hasn't been touched in like 2 years
21:39 hmmmm I dunno, perhaps try debugging colormachine and see how "page" could possibly become nil
21:40 VanessaE yesterday, I was getting errors citing ridiculous player coords
21:40 VanessaE (like X = -2 million )
21:40 hmmmm colormachine is stable, yes?
21:40 VanessaE yep.
21:40 VanessaE Sokomine: *poke*
21:41 VanessaE if another crash happens somewhere, I'll post the backtrace (it'll be with Blockmen's patch)
21:43 VanessaE (ok well mg was last updated in august, but it wasn't related to the crash)
21:43 hmmmm the backtrace is completely cut out but I'm going to assume it happens in the call at init.lua:2090
21:43 hmmmm the parameter fields.page is apparently nil there
21:45 VanessaE well, the lua backtrace got dropped from the paste because of a brain-o on my part (bad redirect on one of my scripts, I've since fixed that)
21:45 VanessaE but,
21:45 VanessaE the thing that bothers me is that it's dying with SIGABRT
21:46 hmmmm why though?
21:46 hmmmm it's always going to die with SIGABRT
21:47 hmmmm if not with FATAL_ERROR, the ServerError is going to propogate to the top
21:47 VanessaE hm.  just seems a bit... harsh
21:47 VanessaE it doesn't even write the map out when it dies like that
21:47 hmmmm well what would you prefer the server do
21:47 Taoki joined #minetest-dev
21:47 VanessaE it's no big deal -- you understand the engine far better than me
21:47 VanessaE so I'll wait for the next crash and see what it does.
21:48 hmmmm just to be clear, this happens when running a dedicated server, yes?
21:48 VanessaE yep
21:49 hmmmm right
21:49 hmmmm this has been the same behavior all the time
21:49 VanessaE ok
21:49 VanessaE it just threw me is all
21:55 PilzAdam when I set max_block_send_distance = max_block_generate_distance = 40, why doesn't generate / load the whole area? my view range is fixed at 500 and there are massive holes in the world
22:13 Taoki joined #minetest-dev
22:53 ElectronLibre left #minetest-dev
22:56 paramat joined #minetest-dev
23:01 paramat PilzAdam, maybe because of 'client_mapblock_limit = 5000', that's quite low, it affects viewing a large area
23:19 rubenwardy joined #minetest-dev

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