Minetest logo

IRC log for #minetest-dev, 2015-05-18

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

All times shown according to UTC.

Time Nick Message
00:31 Wayward_Tab joined #minetest-dev
00:58 hmmmm oh wow
00:59 hmmmm I never saw this until now - mapblock.h has this BlockGenerationStatus enum
00:59 hmmmm this sounds somewhat close to what I was planning
01:36 deltib joined #minetest-dev
01:41 Taoki joined #minetest-dev
01:42 paramat joined #minetest-dev
01:52 paramat serialize_schematic is still segfaulting for me, i would like to attempt some serious debugging to get a backtrace, i assume i need the debug build and a debugger program? any recommendations for a linux debugger to use?
01:53 paramat this is driving me nuts
01:54 * paramat drives his nuts around
01:55 paramat essentially how do i get more info on a bug when even verbose debug.txt shows nothing?
01:57 exio4 gdb?
01:57 exio4 + bt full / thread apply all bt full ?
01:59 paramat okay
02:15 Wayward_Tab joined #minetest-dev
02:16 hmmmm https://github.com/kwolekr/minetest/commit​/46684beec185d13f89c4a91aaa5dd2148ebb0273
02:17 hmmmm paramat, maybe give me the segfaulting schematic
02:18 paramat it segfaults without writing the mts file, same problem as before
02:19 paramat the mod that segfaults is here https://github.com/paramat/saveschems
02:20 paramat trying to serialize lua table to mts using format v4, uses per node force placement
02:20 paramat it's the bug you couldn't reproduce
02:22 hmmmm hmm
02:23 paramat i'll try a debug build next, but gdb reports the segfault is in '0x00000000006100ab in load_schematic_from_def(...)'
02:24 hmmmm ahhh
02:24 hmmmm if you were able to see the name of the crashing function, that means you do have a debug build
02:25 paramat weird, i used a standard build
02:31 hmmmm oops
02:33 hmmmm try now paramat
02:33 hmmmm thanks for the bug report.
02:34 paramat wow will try
02:42 paramat thanks no crash and files appear
02:43 paramat another step closer to mgv5/7 biomes
02:49 paramat per node force place confirmed working, no more tree trunks removed by a nearby tree's leaves
02:57 paramat left #minetest-dev
03:42 Hunterz joined #minetest-dev
05:05 jin_xi joined #minetest-dev
05:37 hmmmm hrmh
05:38 hmmmm the Player/PlayerSAO interaction seems kind of wonky to me.  I don't understand the reason why PlayerSAO is the one to assign a peer_id to a player
05:42 sfan5 joined #minetest-dev
05:42 Hunterz joined #minetest-dev
05:44 hmmmm celeron55:  explain to me why peer_id isn't set before adding the player to the environment
05:44 hmmmm there must be some kind of reason for this.
05:57 chchjesus joined #minetest-dev
06:06 crazyR joined #minetest-dev
06:06 crazyR joined #minetest-dev
07:20 kilbith joined #minetest-dev
07:26 Darcidride joined #minetest-dev
07:30 cib0 joined #minetest-dev
07:37 Calinou joined #minetest-dev
08:05 Yepoleb joined #minetest-dev
08:08 crazyR joined #minetest-dev
08:08 crazyR joined #minetest-dev
10:23 cib0 joined #minetest-dev
10:35 nicnixnic joined #minetest-dev
10:43 selat joined #minetest-dev
10:55 Gethiox joined #minetest-dev
11:13 OldCoder joined #minetest-dev
11:46 sfan5_ joined #minetest-dev
12:05 RealBadAngel joined #minetest-dev
12:33 RealBadAngel joined #minetest-dev
12:33 RealBadAngel kahrl, here?
12:54 book` joined #minetest-dev
13:05 RealBadAngel joined #minetest-dev
13:11 est31 joined #minetest-dev
13:12 est31 ~tell hmmm btw the memory layout of integers isn't set by the standard either. So its equally "wrong" to specify std::string and u32 inside the protocol.
13:12 ShadowBot est31: O.K.
13:17 kilbith :/ he missed one 'm'
13:23 est31 joined #minetest-dev
13:23 est31 ~tell hmmmm btw the memory layout of integers isn't set by the standard either. So its equally "wrong" to specify std::string and u32 inside the protocol.
13:23 ShadowBot est31: O.K.
13:23 est31 kilbith, better?
13:30 cib0 joined #minetest-dev
13:34 kilbith yes
14:12 hmmmm joined #minetest-dev
14:18 LittleJoe joined #minetest-dev
14:47 Amaz- joined #minetest-dev
14:51 Darcidride joined #minetest-dev
15:05 jin_xi joined #minetest-dev
15:06 nore joined #minetest-dev
15:07 AnotherBrick joined #minetest-dev
15:33 cib0 joined #minetest-dev
15:42 ElectronLibre joined #minetest-dev
15:54 ElectronLibre joined #minetest-dev
16:01 RealBadAngel joined #minetest-dev
16:05 Adimgar joined #minetest-dev
16:19 Krock joined #minetest-dev
16:21 Hunterz joined #minetest-dev
16:49 asl97 joined #minetest-dev
17:02 Hijiri joined #minetest-dev
17:06 rubenwardy joined #minetest-dev
17:07 RealBadAngel joined #minetest-dev
17:10 RealBadAngel hi, whats up guys?
17:14 rubenwardy nm
17:23 VanessaE hi
17:26 rubenwardy Reworking / replacing hud should be a top priority, IMO.
17:27 rubenwardy Is there any work / designs on this?
17:30 VanessaE joined #minetest-dev
17:30 RealBadAngel well, minimap is a part of hud :)
17:30 VanessaE nice
17:30 RealBadAngel im done with threading the minimap
17:31 RealBadAngel meanwhile i got rid of licensing problems (got no response from gsmanners, so i just opened new file and wrote it from scratch)
17:31 rubenwardy Is minimap entirely client side?
17:32 RealBadAngel yes
17:32 rubenwardy Awesome
17:32 VanessaE merge that sucker
17:32 * sofar wants to pull it so bad
17:32 RealBadAngel need a few days more
17:32 rubenwardy With proper testing, ofc
17:32 RealBadAngel have some problems with voxel area
17:33 RealBadAngel btw, reading the map from another thread without voxel area is not a problem
17:33 RealBadAngel but having large part of the map copied to it seems to be a problem
17:34 Adimgar joined #minetest-dev
17:34 rubenwardy Branch "minimap"?
17:34 RealBadAngel not in the repo yet, its too fresh n unstable
17:34 rubenwardy :(
17:34 rubenwardy I wanna test
17:34 cib_ joined #minetest-dev
17:34 rubenwardy oh well
17:34 rubenwardy :P
17:36 Adimgar joined #minetest-dev
17:47 blaze joined #minetest-dev
17:54 technomancy joined #minetest-dev
18:13 Hijiri joined #minetest-dev
18:25 VanessaE um... why are player names black now?
18:25 VanessaE (server's at git HEAD as of now, client is at 4c9a8a91c )
18:26 est31 joined #minetest-dev
18:27 VanessaE http://digitalaudioconcepts.com/vanessa/hob​bies/minetest/screenshots/random/Screenshot​%20-%2005182015%20-%2002%3a32%3a45%20PM.png
18:27 VanessaE i.e. that.
18:27 rubenwardy probably the set color tag patch
18:27 est31 yea
18:27 est31 there have been multiple changes though
18:27 est31 lemme check
18:28 est31 255 255 255 is that black or white?
18:28 VanessaE that *should* be white
18:28 Krock white
18:28 est31 ok then it probably has been one of hmmmm's commits
18:29 est31 but checking
18:29 hmmmm hrmm
18:30 hmmmm well, does your mod use set_nametag_attributes()?
18:30 hmmmm any of your mods
18:35 est31 hrmm
18:35 MinetestForFun joined #minetest-dev
18:35 est31 cant reproduce at fb36c471d7c35
18:35 VanessaE hmmmm: I doubt it.
18:36 kilbith i confirm : https://lut.im/GjebQYuI/Hsf3m8OA
18:36 VanessaE the only update I put through today was just homedecor, datastorage, and worldedit
18:36 * est31 checks dreambuilder to be sure its not because of the game
18:37 est31 nope
18:37 rubenwardy Krock? Is there a reason why carts has physical = false ?
18:38 est31 kilbith, can you confirm at singleplayer too?
18:38 VanessaE [05-18 14:42] <ThatGraemeGuy> VanessaE: player names are dark now
18:38 VanessaE [05-18 14:45] <ThatGraemeGuy> commit 20aa78cffb6594ff8fdea4291f18d1a8bd252529 Author: kwolekr <kwolekr@minetest.net> Date:   Sun May 17 22:32:35 2015 -0400
18:38 rubenwardy physical = true makes collisions. I'd imagine it would cause problems with cart logic. https://forum.minetest.net/v​iewtopic.php?f=9&amp;t=12185
18:38 Krock rubenwardy, I had troubles with cart movements while re-writing. Try to make it = true and see what happens
18:38 RealBadAngel hmmmm, i saw that you mentioned something bout client side modding. any details?
18:39 Calinou rubenwardy, if physical = true the carts may stop when there are players inside them
18:39 kilbith est31, i confirm at singleplayer as well : https://lut.im/EeztKEzz/N8Agm44G
18:39 est31 ok
18:40 Adimgar joined #minetest-dev
18:41 est31 it is really some commit between fb36c471d7c3 and 20aa78cffb6594ff8
18:41 est31 weird
18:41 est31 time for a bisect
18:41 est31 mini-bisect :)
18:42 TeTpaAka joined #minetest-dev
18:46 est31 aaaaah
18:46 est31 https://github.com/minetest/minetest/commit​/11a96e4901ce54c4646e47866f7a66cd9034c16b#d​iff-722710b5ecd9db42e7dd785dd517e186R1719
18:46 RealBadAngel Calinou, i wont make minimap showing more than 128x128, with more area visible you always can see unloaded blocks
18:46 Calinou ok, it's fine
18:47 VanessaE hmmmm: you bwoke it! :D
18:47 RealBadAngel and whole map (like 256*256) is visible only when you will stand down in one place for quite a while
18:47 est31 its understandable that he touched that line
18:47 hmmmm ?
18:47 est31 after all that one triggered a "unused int" warning
18:47 est31 still bad
18:48 est31 I'm not 100% sure yet whether thats the cause
18:48 est31 but its quite probabla
18:48 est31 e*
18:48 hmmmm ahhh...
18:49 hmmmm version was added in the 'generalized' version
18:49 hmmmm no wonder why
18:49 hmmmm yeah I broke it.  reading in a 0x00 for alpha can cause text to appear black
18:49 hmmmm :p
18:49 hmmmm sorry guys
18:49 VanessaE haha
18:50 hmmmm didn't think that was used anymore
18:50 est31 what I still wonder is why the compiler warns there
18:50 RealBadAngel hmmmm, what bout cilent side modding?
18:50 hmmmm what about it
18:50 TeTpaAka Because the version is never used.
18:50 RealBadAngel i saw that you mentioned it as your to-do
18:50 TeTpaAka I added it to allow future versions to transmit more information.
18:50 est31 but its sent by the protocol, isn't it?
18:50 TeTpaAka It is sent.
18:51 rubenwardy Do we have a wikipage/maillist/github issues discussing the technical details of client side modding, and discussing APIs?
18:51 hmmmm not yet
18:51 est31 sending serialisation versions is very good thing to do
18:51 hmmmm you're free to start one though
18:52 TeTpaAka https://github.com/minetest/minetest/b​lob/master/src/genericobject.cpp#L179
18:52 est31 ah ok
18:52 TeTpaAka But an alpha of 0x00 would make it transparent?
18:53 TeTpaAka Why is it black?
18:53 hmmmm because the base image it's being blended with it is all black
18:54 est31 why is it an alpha of 0x00? Isn't alpha read at the start?
18:54 Robert_Zenz joined #minetest-dev
18:54 hmmmm because the current version is 0
18:54 est31 and version is 1...
18:54 TeTpaAka Yes, but version is 1, so it would be 0x01
18:54 est31 ah
18:54 est31 yea
18:54 hmmmm so then it's almost completely transparent
18:55 TeTpaAka But it would still be white?
18:55 hmmmm the blend would cause the resultant pixel to be 1, 1, 1, 1
18:55 hmmmm so it's not completely black
18:55 hmmmm just... very, very, VERY dark gray
18:56 VanessaE gimp confirms.
18:57 VanessaE RGBA = 010101FF
18:57 VanessaE (well the final composited result, that is)
18:57 est31 so basically transparency is broken now?
18:57 est31 (for nametags)
18:57 VanessaE est31: kilbith just reported that Homedecor's frosted glass skylights are also rendering in black
18:57 hmmmm huh? no
18:58 kilbith whereas the texture is white-ish
18:58 hmmmm transparency works as expected
18:58 hmmmm it doesn't blend with the background z-node like you think it should
18:58 TeTpaAka The problem is, the client is reading something as color that isn't a color.
18:58 est31 but when I deliberately set an alpha for a nametag, does that work?
18:58 hmmmm it doesn't
18:58 hmmmm don't bother setting an alpha, it defaults to 255
18:59 hmmmm which is completely opaque
18:59 est31 then you can't disable nametags...
18:59 hmmmm that's true
18:59 TeTpaAka Wait. As I tested it in the first version, I could hide the nametags.
18:59 hmmmm by adjusting the alpha??
18:59 TeTpaAka Yes.
19:00 hmmmm that's odd
19:00 hmmmm does it work still if you set it to 0?
19:00 TeTpaAka I'm compiling right now.
19:00 hmmmm irrlicht has several transparency modes
19:00 TeTpaAka Have to adjust the source to respect the version, though.
19:00 hmmmm one is a blend, which is what I supposed this was
19:00 kilbith there : https://lut.im/rBB2MLaG/3dzYp3Um
19:00 hmmmm the second is with transparent pxiels
19:01 hmmmm kilbith, is that texture supposed to have alpha like that?
19:01 VanessaE hmmmm: it is, but it's supposed to be (and always has been) white as in the wield.
19:01 kilbith the texture is white is half-transparency
19:02 kilbith with*
19:02 VanessaE https://raw.githubusercontent.com/Vanes​saE/homedecor_modpack/master/homedecor/​textures/homedecor_skylight_frosted.png
19:03 VanessaE I can confirm the black skylight on my end also.
19:04 hmmmm hm
19:04 TeTpaAka http://i.imgur.com/CV73CTW.png
19:05 TeTpaAka Still can disable namtag after adjusting the source back.
19:05 hmmmm change script/common/c_converter.cpp on line 240 to color.setAlpha(lua_isnumber(L, -1) ? lua_tonumber(L, -1) : 0);
19:05 hmmmm I doubt this is the problem but try it anyway
19:05 hmmmm it's perhaps the most impactful change i made
19:06 * est31 pushes in 1 minute https://github.com/est31/minetest/commit/​21ab396a90215ac2df09a744dfa3f88a449f41af
19:06 rubenwardy if (m_textnode != NULL) {
19:07 rubenwardy { is unnecessary, doesn't the linux code style say to leave them out
19:07 hmmmm perhaps so, but adding braces around single line conditionals is an option
19:07 hmmmm either way is valid for us
19:07 est31 will push https://github.com/est31/minetest/commit/​41d5b99ca3640c808b0d7f2634777ee7b779a725
19:07 est31 (fixed commit msg)
19:08 est31 pushed
19:10 TeTpaAka But still, why was it black in the first place. For me, they were transparent.
19:11 est31 dammit
19:12 est31 amended the commit to remove the kate swp file
19:13 rubenwardy lol
19:13 Calinou you shouldn't have said it
19:13 VanessaE est31: skylights are still black :(
19:15 TeTpaAka Can someone with the black nametags test what setting the alpha of the nametag color to 0 would cause?
19:16 TeTpaAka Now, after the fix, of cause.
19:16 est31 moment
19:19 est31 yea it *is* still there
19:19 est31 in gray
19:20 TeTpaAka Strange. I can't reproduce at all.
19:21 TeTpaAka That ruins the main cause, why I implemented this thing. I wanted to hide the players completely, including nametags.
19:21 TeTpaAka But that doesn't seem to work reliable.
19:21 est31 http://pasteboard.co/yNQbOdd.png
19:21 est31 there is a flag in the engine which can control the transfer distance of SAOs
19:21 est31 its better to use that instead
19:22 est31 because that way the server doesn't send useless movement packets
19:22 est31 and users can't hack their clients to see players
19:24 TeTpaAka No, I meant some sort of an item (invisibility cloak/potion) that makes a temporary effect.
19:25 Sokomine hmmmm: where can i find more information regarding minetest.set_gen_notify ? especially how to compose the flags parameter? i just wanted to take a look at the output of minetest.get_mapgen_object('gennotify') for now. maybe it can be used in the future to write a small mod that adds nodes and mobs into dungeons. there are so many empty ones there
19:26 est31 https://github.com/minetest/minetest/commi​t/5b03857c625633e8daea3e0118bf701a93109127
19:27 TeTpaAka Why are the nametags black for some but not for me?
19:27 est31 can you post a screenshot?
19:28 TeTpaAka I already did. http://i.imgur.com/CV73CTW.png
19:29 Sokomine hmmmm: ah. found it through google. the dev wiki lists it in a diffrent way than lua_api.txt
19:30 Wayward_Tab joined #minetest-dev
19:32 TeTpaAka Closer: http://i.imgur.com/xp3wIHx.png
19:34 TeTpaAka Does at least setting the color work for you?
19:34 est31 yes
19:35 est31 can you try against a lighter background
19:35 est31 ?
19:35 Adimgar joined #minetest-dev
19:35 est31 (eg sky)
19:35 cib0 joined #minetest-dev
19:37 est31 "Error 503 Service Unavailable" thanks sourceforge
19:39 TeTpaAka est31: http://i.imgur.com/okgLVzi.png
19:41 kilbith skylight node with normal drawtype : https://lut.im/QAQmJhtD/urQrYvFt   same node with plantlike : https://lut.im/Fr1xiwDH/raQCgO1a
19:42 est31 can you bisect kilbith or VanessaE?
19:42 TeTpaAka In front of sky: http://i.imgur.com/nQwPLST.png
19:42 est31 hm, very weird
19:43 est31 perhaps your irrlicht build has other standard values
19:43 VanessaE est31: I could try I guess.
19:43 est31 I'm trying to solve that name issue, while you bisect, ok?
19:43 est31 multi-threading :)
19:43 TeTpaAka How can I find out?
19:44 est31 TeTpaAka, was a weird guess
19:45 VanessaE est31: bisecting now..
19:45 est31 TeTpaAka, which distro do you run on?
19:46 TeTpaAka Kubuntu 14.04
19:46 est31 hrmm
19:47 est31 (side note: dont switch to 15.10, its horribly buggy)
19:47 TeTpaAka Thanks.
19:47 Calinou 15.10 isn't released yet…
19:47 est31 ummm 15.04
19:47 TeTpaAka I tested shaders. Works with and without.
19:48 VanessaE wtf?
19:48 VanessaE 2015-05-18 15:55:04: ERROR[main]: ...ssa/.minetest/games/minetest_g​ame/mods/default/nodes.lua:1554: attempt to call a nil value
19:48 VanessaE minetest_game is at HEAD
19:49 VanessaE I'm in the middle of the bisect and it throws this error.  the connect-to-raillike patch apparently.
19:49 VanessaE use of it, I mean
19:50 est31 weird
19:50 est31 what I do on these occasions is firing up git bisect show
19:50 est31 and then checking out another commit in that list
19:50 VanessaE I just need to back minetest_game down to something old-ish
19:51 est31 or that
19:51 ElectronLibre VanessaE, I got this one or two days ago, you are running a version of minetest (engine)'s builtin without the rail-group system (that's what I assumed).
19:51 VanessaE ElectronLibre: yeah I know.
19:51 ElectronLibre So yes, comment (it's the only raillike line in the whole repo) or downgrade.
19:51 VanessaE there, rolled it back to dd34dffa
19:51 * VanessaE continues ..
19:52 Robert_Zenz joined #minetest-dev
20:02 Hijiri joined #minetest-dev
20:03 TeTpaAka The software renderer seems to ignore the alpha. Though the nametag is still white.
20:07 VanessaE *sigh*
20:07 VanessaE est31: the bisect failed.  All the way back to 0.4.12-release
20:09 est31 so its an old bug?
20:09 VanessaE idk, I don't remember seeing it do this
20:09 VanessaE I'm gonna run through it again
20:10 est31 it can be because of freetype
20:10 est31 TeTpaAka, do you use freetype?
20:10 VanessaE eh? I'm working on the thing with the glass skylight being black.
20:12 TeTpaAka Doesn't seem so. Now testing with it.
20:13 est31 lol was directed at TeTpaAka VanessaE :)
20:13 VanessaE oh ok :)
20:17 TeTpaAka est31: With freetype it is grey.
20:18 est31 then its freetype's fault
20:18 est31 hrmmm
20:20 TeTpaAka Hiding can be done by removing the nametag object. I'll code it in a few days. (Don't have time now).
20:20 TeTpaAka That should at least be reliable.
20:20 est31 yea
20:30 ElectronLibre left #minetest-dev
20:33 Calinou joined #minetest-dev
20:34 est31 ok setting the visibility is enough
20:43 est31 pushing in 10 minutes: https://github.com/est31/minetest/commit/​b5f41798eeb1bd28a0028889aaeaa422785e3a57
20:59 est31 kahrl, you there?
21:00 est31 I'm wondering about your code :)
21:01 jin_xi joined #minetest-dev
21:01 est31 there is #2710, and most likely the person runs a a bad mod, but I'm not 100% sure what we require from mods to do
21:01 ShadowBot https://github.com/minetest/minetest/issues/2710 -- Node Drop Errors
21:02 est31 I've traced it back to your commit : https://github.com/minetest/minetest/commi​t/157a4cf18cb9c098f465b8baecd7d2cd5705f2dd​#diff-494a8464a5335033fe2460099e805050R197
21:04 est31 I guess a mod just specified something wrongly
21:04 est31 nvm, I think I've understood the code.
21:04 est31 thanks though+
21:08 Hijiri joined #minetest-dev
21:10 VanessaE est31: bisect for the frosted skylight issue failed again.
21:11 est31 so 0.4.12 is noted as "bad"??
21:11 VanessaE if I bisect all the way back to 0.4.11, I get http://pastebin.com/UZNg1K7E (my minetest_game is still too new, all the skips are due to some mapgen function used).  if I mark the first good commit as 0.4.12, I get http://pastebin.com/VUeyg3kA
21:12 VanessaE neither of these makes sense
21:12 VanessaE and I made sure to dig and re-place the object at each test.
21:12 VanessaE here's the relevant code in the mod:  https://github.com/VanessaE/homedecor_modp​ack/blob/master/homedecor/shingles.lua#L14
21:13 est31 so 0.4.12 itself is good??
21:13 VanessaE idk.  In the above code, the only thing I noticed is that paramtype = "light" isn't specified ... but that shouldn't be needed should it?
21:14 est31 no idea
21:15 kilbith no, this doesnt affects anything here
21:24 VanessaE found the problem.  paramtype = "light" got dropped from that entry a few weeks ago, and it IS needed.
21:25 kilbith a new node has to be placed to see that difference
21:25 VanessaE that begs the question:  why isn't paramtype defaulted to "light"?
21:26 VanessaE seeing as how that's overwhelmingly the most common mode for param1
21:27 est31 changing defaults is dangerous
21:30 VanessaE perhaps
21:30 VanessaE but in this case I highly doubt it'll break anything
21:31 VanessaE seems to only be a problem with certain drawtypes
21:31 VanessaE "normal" cubes don't seem to need it (e.g. HD's bathroom tiles, which were missing that mode, worked fine despite)
21:32 VanessaE but raillike and plantlike do.
21:32 kilbith mesh as well
21:37 VanessaE in any case either it needs to be fixed to be consistent across all drawtypes, or the drawtype-specific behavior needs to be documented.
21:39 est31 yes
22:06 prozacgod joined #minetest-dev
22:18 Wayward_One joined #minetest-dev
22:26 RealBadAngel joined #minetest-dev
22:41 est31 argh
22:42 est31 gonna remove yet another errorstream
22:59 est31 10 minutes till I merge https://github.com/est31/minetest/commit/​497299afd62e2f9b18dbf2a366bb22996e73102b
23:11 kaeza_ joined #minetest-dev
23:12 hmmmm est31, why exactly??
23:12 hmmmm aren't those things errors?
23:16 est31 they can be triggered by users
23:16 hmmmm ahh
23:16 est31 errorstream is distributed to the whole server
23:17 hmmmm if anything i think we should have a separate logging level for that
23:17 hmmmm "a client is acting goofy stream"
23:18 est31 lol
23:18 hmmmm it could be helpful in identifying what are basically hack clients
23:18 est31 yea
23:18 hmmmm or maybe people making bots that try to crack accounts
23:18 est31 or try-to-hack ones
23:19 hmmmm or more likely is that it's a developer error
23:19 hmmmm well
23:19 hmmmm actionstream is logged by default so it's good
23:19 hmmmm what's your take on shadowninja's logging cleanup
23:19 est31 I like the warning loglevel
23:20 hmmmm yes
23:20 hmmmm i like the idea of a warning loglevel, and i'd like to get it committed before it version control rots any more
23:20 hmmmm but there's the whole dstream thing
23:20 est31 your only concern is the removed dstreamß
23:20 est31 ?
23:20 hmmmm basically yeah
23:21 hmmmm some things SN does are like wtf to me
23:21 est31 we most certainly need a way for 1. lua's print, and 2. the unittests
23:21 hmmmm dstream is purposely out-of-band because it's explicitly for debugging
23:21 hmmmm this rules out any issues with the regular logging facilities
23:21 est31 OTOH, one can use errorstream for debugging too
23:21 hmmmm yeah
23:24 troller joined #minetest-dev
23:24 est31 btw how did he fix print()?
23:26 hmmmm by using actionstream i thought
23:27 est31 I dont know whether I like that
23:28 est31 but mostly I think the situation is better than before
23:28 hmmmm I think we should set the modname global before each callback
23:29 est31 ?
23:29 hmmmm that way mods have their own names after initialization, and print could use it too for printing out the modname responsible
23:29 hmmmm as it stands right now, mods only know their own name during initialization
23:29 est31 yea thats the only point its well-defined
23:30 hmmmm if they don't store it somewhere else, they can't access it anymore
23:30 est31 the issue is that mods share a common lua context
23:30 est31 imagine intllib
23:30 est31 that library never directly interacts with the engine
23:30 hmmmm anything printed by intllib in lua would assume the modname that called it
23:30 est31 yea
23:30 hmmmm isn't that reasonable enough??
23:32 est31 its only the modname of the callback
23:33 est31 I dont like that
23:33 est31 I'm wondering about something else related to this
23:34 est31 there is an issue about a crash in builtin lua code
23:34 est31 it has been mostly a mod's fault
23:34 est31 now I wonder whether we should make code validate the input it got
23:34 est31 and make smarter erro messages
23:35 est31 not "this is nil, you wanted to dereference it"
23:35 est31 but "item definition of node 'modname:nodename' is missing recipe"
23:35 hmmmm hmm
23:35 hmmmm that's harder said than done
23:35 hmmmm you have to code defensively literally everywhere
23:35 est31 yea...
23:36 hmmmm so it'd be a big modification, but it would be worthwhile i think
23:36 hmmmm is this about the blank craft receipe string crash?
23:36 est31 no the other one
23:36 est31 2710
23:37 est31 blank craft recipe is perhaps a mod error too, I guess
23:40 est31 "goofystream"
23:40 hmmmm nah but seriously
23:40 hmmmm if we do add a goofystream it should be something like...
23:41 hmmmm auditstream
23:41 hmmmm ?
23:41 hmmmm because as a server owner, you'd want to audit for suspicious events
23:41 hmmmm actionstream works fine enough though
23:42 est31 yea
23:42 est31 its a bit crowded ...
23:43 hmmmm what would be a really worthwhile log enhancement is if you could split the different levels into different logfiles
23:43 hmmmm right now the log infrastructure is set up so that you could do that, but the way it's being used by minetest is that it uses the same file for everything
23:43 est31 yes
23:44 est31 there is also this ncurses thing
23:44 hmmmm lol
23:44 hmmmm what does ncurses have to do with this
23:45 est31 because it will modify log output too
23:45 est31 then there is also possible systemd integration
23:45 hmmmm kill it with fire
23:45 est31 all about how to bring logging to users
23:45 hmmmm we shouldn't integrate with something that disrespects our freedoms :/
23:46 hmmmm systemd is pretty evil
23:46 hmmmm in any case you realize what ncurses is right?
23:46 exio4 hmmmm: that's because you aren't hipster, if you were hipster, systemd would be perfect for you!
23:46 est31 hmmmm, yes I know
23:46 est31 for interactive server console
23:47 est31 exio4, the issue is that non-systemd is dying.
23:47 hmmmm oh
23:47 hmmmm I hope that systemd is just a fad
23:47 hmmmm it needs to die soon
23:47 hmmmm ncurses would be an interface with which to view the logs
23:48 hmmmm if there's an interactive server console, my opinion is that it shouldn't modify the logs at all
23:48 est31 systemd is the (pretty bad) continuation of where unix stopped. it defines much more APIs.
23:48 est31 sorry better expression would be "recieve" than modify
23:49 est31 pretty bad not just because it is glibc and linux only
23:50 proller joined #minetest-dev
23:52 hmmmm systemd is basically worse than windows
23:52 hmmmm same binary log file format and all that jazz, same cryptic BS, same bloaty API, same obscure stuff
23:52 Lunatrius` joined #minetest-dev
23:52 hmmmm except systemd takes on the roles of csrss.exe, smss.exe, svchost.exe, and all the services
23:52 est31 my hope is that its developers learn
23:53 hmmmm well honestly
23:53 est31 they are still different processes
23:53 hmmmm oh they are?
23:53 hmmmm I thought systemd had them all in the same pid 1 process
23:53 hmmmm just as loadable .sos
23:54 est31 ps -A -l | grep systemd | wc -l gives me 7
23:54 hmmmm i guess that's a little bit better... not by much though
23:54 est31 with pids of 1 319 350 665 890 1504 and 2300
23:54 hmmmm systemd isn't the result of developers forgetting design lessons from the past
23:54 hmmmm i think it's much more malicious
23:55 est31 systemd-journal systemd-udevd systemd-logind systemd-timesyn and three times "systemd"
23:55 est31 the linux world has one great weakness and problem
23:55 hmmmm poul henning kemp had a nice presentation about how easy it is for malicious actors to insert backdoors or whatever into open source
23:55 est31 its that its so heterogenous
23:56 hmmmm that being said, systemd has all the hallmarks of being placed there by the NSA
23:56 hmmmm i don't fucking trust systemd on my own computer
23:56 est31 ok thats possible
23:56 est31 you have bsd havent you?
23:56 hmmmm yeah
23:56 hmmmm it's like openrc
23:56 exio4 systemd is too bloated, it does too much, and its 'bugs' are things that were fixed 15 years ago
23:57 hmmmm if i recall, didn't arch linux switch from openrc to systemd??
23:57 exio4 it didn't use openrc, but its own minimal init system to systemd
23:57 hmmmm that's a damn shame
23:57 est31 thats precisely the issue
23:57 exio4 Archlinux's init system was really easy to modify and swap out, now it is systemd
23:57 est31 everybody wrote their own "little" init system
23:58 exio4 systemd is the worst of those little init systems, plus scope creep :)
23:58 est31 at the end, when you want to write a service, every distro has to maintain their own init scripts because they are different for every distro
23:58 hmmmm no i think systemd would be fine if it stuck to what it originally was, and it didn't get pushed into distros, and things didn't become dependent on it
23:58 est31 thats alot of work just wasted
23:58 hmmmm instead we have this hostile takeover thanks to red hat and the gnome foundation
23:58 hmmmm jerks..
23:59 exio4 hmmmm: yeah, the scope creep and hostile behavior is what put people off
23:59 hmmmm how can one organization have so much control over linux
23:59 proller joined #minetest-dev
23:59 hmmmm this goes against open source
23:59 est31 "open source" is for me "the right to merge"

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