Minetest logo

IRC log for #minetest-dev, 2020-05-26

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

All times shown according to UTC.

Time Nick Message
00:18 gd___ joined #minetest-dev
00:39 absurb joined #minetest-dev
01:04 troller joined #minetest-dev
01:50 bodqhrohro joined #minetest-dev
02:24 YuGiOhJCJ joined #minetest-dev
02:39 ANAND joined #minetest-dev
02:41 gh00p PGimeno: sorry for the delayed response. This is happening both on Android (5.1) and macOS (5.2) I can get specific versions if that helps.
02:43 ANAND joined #minetest-dev
02:44 gh00p OMG, I just deleted env_meta.txt (which was zero length) because I *hadn't* actually tried that .. and it works on macOS. Testing now on Android.
03:07 gh00p That solution worked on Android as well. Which I guess leaves me with three questions... what is that file supposed to do, why did its presence bork the world, and is this enough of an edge case that it's not worth the effort to come up with a way to handle this scenario automatically?
03:20 gh00p joined #minetest-dev
03:39 ghoti joined #minetest-dev
04:46 calcul0n joined #minetest-dev
06:17 Darcidride joined #minetest-dev
06:18 erlehmann rubenwardy, where can i download your minetest book for offline use?
06:18 erlehmann or the source code to render it myself?
06:32 ANAND ~tell erlehmann https://gitlab.com/rubenwardy/minetest_modding_book
06:32 ShadowBot ANAND: O.K.
06:42 erlehmann joined #minetest-dev
07:07 NetherEran joined #minetest-dev
07:39 mizux joined #minetest-dev
08:00 ShadowNinja joined #minetest-dev
08:04 Beton joined #minetest-dev
09:24 absurb joined #minetest-dev
10:12 sfan5 gh00p: 1) it stores variables like the time of day, day count and a few other internal things 2) it's presence didn't break it, what broke it that it was somehow corrupted 3) Minetest does everything to safely write the file (similar to how SQL databases ensure consistency) so it should beextremely rare to see it corrupted
10:13 troller joined #minetest-dev
10:18 PGimeno maybe Minetest can detect when it's empty, emit a warning and start anew
10:32 erlehmann joined #minetest-dev
11:29 Fixer joined #minetest-dev
11:50 gh00p sfan5: just to be clear, the file was zero-length, and removing it eliminated the crash.
11:51 sfan5 when you remove it the engine creates a new one with default values
11:54 gh00p But is being zero-length considered "corrupt"?
11:54 sfan5 yes
12:03 Lejo joined #minetest-dev
12:09 Lejo Feedback for Idea needed: I already had an discussion with Ravise on the Forum about Client-side chat-message handling. Our Idea was to pass the message throught all registred_chat_messages (expect one returns true) by just returning the changed value. This value will be passed than to the next mod handling or change the message, after the last mod the message is printed to the chat. The use case of this is that you can have independent
12:09 Lejo csms handling messages and not only one because they have to return true to change the message. Otherwise there would two messages... What do you think about that idea?
12:58 gh00p_ joined #minetest-dev
13:12 Taoki_1 joined #minetest-dev
13:24 Taoki joined #minetest-dev
13:25 PGimeno I've seen more cases of an empty env_meta.txt causing crashes, hence my suggestion
13:42 gh00p joined #minetest-dev
14:10 ANAND joined #minetest-dev
14:18 rubenwardy #9939
14:18 ShadowBot https://github.com/minetest/minetest/issues/9939 -- CSM: Pass ChatMessage table instead of message to callback by rubenwardy
14:21 sfan5 I believe the server doesn't actually populate those values in an useful way
14:21 sfan5 e.g. nick is always empty
14:22 rubenwardy you're correct, unfortunately
14:22 rubenwardy it's always a raw message
14:22 sfan5 the PR is still useful in principle of course
14:27 ANAND I was going to ask regarding the difference between the various types.
14:27 ANAND What are they *intended* to be used for?
14:29 rubenwardy I'm not sure on the difference between announce and system
14:29 rubenwardy I'm guessing announce = "New match begins"
14:29 rubenwardy and system = "Server is shutting down"
14:29 ANAND i.e. chat_send_all?
14:29 sfan5 my guess: system = from the engine, announce = from mods
14:29 ANAND Ohh
14:29 sfan5 I don't think it would actually be possible to tell those apart
14:29 ANAND Makes sense
14:29 sfan5 and then what is raw for?
14:29 ANAND ^
14:30 ANAND The first post shows player join messages using raw
14:31 rubenwardy yeah
14:31 rubenwardy raw is for anything I guess?
14:31 rubenwardy raw is currently everything
14:31 ANAND However, I assume they *should* be using 'system'?
14:31 rubenwardy perhaps it should be something like "unknown" or "none"
14:31 ANAND Tbh, all chat messages from C++ categorised as "system" doesn't make much sense
14:32 ANAND Does that type need to exist once the other types are properly implemented?
14:32 rubenwardy the selection does seem arbitrary
14:33 rubenwardy also, instead of normal is should be something like "player"
14:33 ANAND IMO, it's more intuitive to have the following modes: 'player', 'mod',
14:33 ANAND ..and 'engine'
14:34 ANAND or maybe just 'system' for the last one
14:34 rubenwardy maybe wait for nerzhul's import
14:34 rubenwardy *input
14:35 ANAND +1 for suggesting to use os.date instead of UNIX timestamp
14:36 ANAND The latter isn't of much use to mods, I think
14:38 sfan5 pretty sure mods can turn the timestamp into a date
14:38 rubenwardy yeah, the mod can do  `os.date("*t", msg.timestamp)`
14:38 rubenwardy but it's a nicer API for the engine to do it
14:43 Lunatrius joined #minetest-dev
14:44 sfan5 what if I want the timestamp for numerical comparisons?
14:44 rubenwardy could provide both timestamp and datetime I guess?
14:44 rubenwardy or just leave it for now
14:53 ANAND Good point about mods needing the raw timestamps for numerical comparisons
14:54 ANAND timestamp + os.date is unnecessary IMO, since the latter can already be obtained from the former.
15:13 Wuzzy joined #minetest-dev
15:45 Lejo joined #minetest-dev
15:45 sfan5 small and easy to review -> #9874
15:45 ShadowBot https://github.com/minetest/minetest/issues/9874 -- Fix liquids refusing to flow in X+ or Z+ in some cases by sfan5
15:48 sfan5 Wuzzy: https://a.uguu.se/g8hXthnirTEo_.png
15:49 Wuzzy i understand your bug report, but i just cannot reproduce it ?
15:49 sfan5 ah, works with shaders disabled
15:49 Wuzzy wtf
16:09 ANAND I get the following error when I try to open a chest:           AsyncErr: ServerThread::run Lua: Expecting itemstack, itemstring, table or nil. item=foo:bar
16:10 ANAND But there's no backtrace whatsoever
16:11 ANAND Could this be related to an itemdef callback by any chance?
16:11 Wuzzy which chest?
16:11 ANAND default:chest
16:11 ANAND I'm suspecting it's the wield item the causes this crash and not the right-clicked node
16:12 Wuzzy sfan5: I was able to reproduce the 2 bugs. But they seem to be unrelated to the PR
16:12 ANAND The right-clicked node needs to have an on_rightclick defined though
16:12 ANAND That, however, makes me more strongly  suspect a faulty itemdef callback
16:13 Wuzzy sfan5: Also, this seems to be a regression. It works fine in 5.2.0
16:13 ANAND joined #minetest-dev
16:13 ANAND The right-clicked node needs to have an on_rightclick defined though
16:13 Wuzzy sfan5: maybe the bug is caused by something entirely different?
16:13 ANAND This, however, makes me more strongly  suspect a faulty itemdef callback.
16:14 ANAND I'll check the itemdef callbacks in my mod again
16:18 rubenwardy Wuzzy: asserts only fail in debug builds
16:19 Wuzzy ah ok lol. I always build in Release ...
16:19 Wuzzy Also in RelWithDebInfo?
16:21 rubenwardy Just debug
16:21 ANAND I don't think RelWithDebInfo is considered to be debug mode
16:21 ANAND debug mode as in NDEBUG macro defined
16:31 sfan5 Wuzzy: might be caused by #9537
16:31 ShadowBot https://github.com/minetest/minetest/issues/9537 -- Reuse object_shader for "wielditem" and "item" entity drawtypes by dcbrwn
16:32 Wuzzy WTF?
16:32 Wuzzy I cannot find these commits in master ?
16:33 Wuzzy found it
16:46 Wuzzy sfan5: reverting #9537 seems to fix it ?
16:46 ShadowBot https://github.com/minetest/minetest/issues/9537 -- Reuse object_shader for "wielditem" and "item" entity drawtypes by dcbrwn
16:46 sfan5 would you mind creating an issue? if not I can do it
16:49 Wuzzy yes, i would mind. :P plz do it
17:00 gh00p joined #minetest-dev
17:02 sfan5 https://github.com/minetest/minetest/issues/9941
17:03 NetherEran joined #minetest-dev
17:09 Wuzzy Good we have that Devtest now. Without it, you wouldn't have seen the regression. ?
17:14 gh00p_ joined #minetest-dev
17:14 Darcidride joined #minetest-dev
17:28 Krock #9874  when flowing liquids are still changing around the source, liquid flow completely stops after the source node was replaced
17:28 ShadowBot https://github.com/minetest/minetest/issues/9874 -- Fix liquids refusing to flow in X+ or Z+ in some cases by sfan5
17:28 sfan5 that's not a new bug is it?
17:28 Krock probably not
17:28 Krock just something I observed. otherwise looks good
17:29 Krock there's probably some cache that isn't erased (or overwritten) by liquidflow
17:30 gh00p joined #minetest-dev
17:31 Krock nasty serialization. good catches.
17:31 Krock will merge #9937 in 10 minutes
17:31 ShadowBot https://github.com/minetest/minetest/issues/9937 -- Devtest: Fix crash in player unittest by Wuzzy2
17:44 gh00p joined #minetest-dev
17:48 Krock merging
17:54 gh00p joined #minetest-dev
18:38 gh00p joined #minetest-dev
18:59 appguru joined #minetest-dev
19:06 appguru Hey, someone mind throwing an eye on #9907 ? Seems to be a relatively highly demanded feature (judging by the reactions) which breaks nothing & passes all checks
19:06 ShadowBot https://github.com/minetest/minetest/issues/9907 -- Shaders as media files (server-sent shaders) by appgurueu
19:06 appguru I'd really like to get it merged for 5.3
19:06 appguru Ideally together with the second rendering stage
19:09 Wuzzy About the devtest crash #9938: Should I just remove the offending nodes? There was a reason i added those (see issue).
19:09 ShadowBot https://github.com/minetest/minetest/issues/9938 -- Crash loading devtest world (Assertion `liquid_type == LIQUID_SOURCE' failed.)
19:13 Krock there are liquid flow varieties, so I think those nodes are not necessary?
19:55 appguru joined #minetest-dev
19:57 appguru rubenwardy: thanks for noticing
19:57 Wuzzy Krock: waiting  for more core dev opinions
19:59 appguru rubenwardy: what were you trying to say with "I'd prefer requiring the shaders to be named like world.vs.glsl in order to replace opengl_vertex.glsl, to allow more shaders files to be used later."
19:59 appguru it's already working like this
20:05 rubenwardy appguru: responded in issue
20:05 rubenwardy didn't notice the messages here
20:06 appguru rubenwardy: it doesn't ignore the name
20:06 appguru basically world.vs.glsl would be shaders/world/opengl_vertex.glsl
20:11 Wuzzy How do you rightclick in Android?
20:20 appguru Wuzzy: double-tap I think
20:24 rubenwardy what happens if I made a shader called foobar.vs.glsl?
20:26 rubenwardy oh wait, I see
20:29 appguru on a different note, https://github.com/iponweb/luavela seems to be sort of a maintained LuaJIT
20:29 appguru (and is Lua 5.1)
20:30 sfan5 https://github.com/LuaJIT/LuaJIT/tree/v2.1
20:30 sfan5 luajit is maintained too, there's just no releases
20:32 skyliner_369 joined #minetest-dev
20:33 rubenwardy Pushing trivial bug fix in 5: https://github.com/rubenwardy/minetest/commit/ca5a865c1d1e36c5361292ecf79a9bc8f14d5296
20:33 sfan5 s/being ignored/not being ignored/
20:33 rubenwardy it shouldn't be ignored
20:33 rubenwardy notice the !
20:34 sfan5 oh
20:34 sfan5 yea
20:34 skyliner_369 so I'm making a texture pack... and I have a rather simple question... does the "crack_anylength" texture have to be 5 states? or can I, say, put 10 states in the texture?
20:37 proller joined #minetest-dev
20:38 sfan5 any amount works
20:42 skyliner_369 Awesome... gonna bump the 5 break-states up to 10 (like minecraft) because I'm doing something relatively simple... reusing textures from the Painterly Pack in a new pack.
20:45 gh00p joined #minetest-dev
20:48 skyliner_369 also do I want to remove the "default_" prefix to the textures? or is that supposed to stay?
20:49 rubenwardy that's supposed to stay
20:49 rubenwardy default_ is from the mod
20:50 rubenwardy textures will be named   modname_texturename.png
20:50 rubenwardy *should
20:53 skyliner_369 thanks. Had to doublecheck, in case, the prefix was, like, for denoting the default texture pack
21:03 gh00p joined #minetest-dev
21:05 PGimeno wow @ 9874 - I researched 2399 for a while but I didn't find anything outstanding. Good job sfan5!
21:06 sfan5 actively debugging code tends to be more successful I'd say...
21:08 gh00p joined #minetest-dev
21:22 PGimeno I wonder how many worlds will it break :)
21:24 PGimeno but that's a necessary evil
21:41 NetherEran joined #minetest-dev
22:07 rubenwardy Decisions: regarding adding support for a ContentDB CLI in minetest, does it make sense for it do initiate a Lua environment? The main menu one, currently
22:07 rubenwardy It won't even be compiled into the server...
22:07 rubenwardy In which, I'll need to move the pkgmgr.lua to C++ to allow that
22:08 sfan5 keeping it in lua sounds useful to me
22:09 rubenwardy it's easier to edit
22:09 rubenwardy in which case, I should move it from   mainmenu/  to a dedicated folder, and have a special environment for it
22:09 rubenwardy eh
22:10 paramat joined #minetest-dev
22:10 paramat merging #9936
22:10 ShadowBot https://github.com/minetest/minetest/issues/9936 -- Devtest: Update biome registrations by paramat
22:11 rubenwardy The reason I'm thinking about this now is because I'm planning to look into adding a download manager soon - ie: some code that allows multiple files to download at once, and install individually
22:11 rubenwardy so the choice of whether to do it in Lua or C++ comes up
22:11 rubenwardy the download manager will have the ability to query for existing downloads, etc
22:14 rubenwardy I'd really like to have nice progress bars for downloads, formspec updates interrupting focus and text input kinda destroy that
22:16 paramat merging #9874
22:16 ShadowBot https://github.com/minetest/minetest/issues/9874 -- Fix liquids refusing to flow in X+ or Z+ in some cases by sfan5
22:36 Fixer_ joined #minetest-dev
22:40 rubenwardy #9944
22:40 ShadowBot https://github.com/minetest/minetest/issues/9944 -- Add button to ContentDB in game buttonbar by rubenwardy
22:41 sfan5 can you make it more colorful
22:41 sfan5 >table.indexof does not work on tables that contain `nil`
22:41 sfan5 it uh what
22:42 rubenwardy it uses ipairs
22:42 rubenwardy I could fix that, but I'm not sure it's entirely correct behaviour....
22:42 rubenwardy as the table isn't a valid array table
22:43 sfan5 ah yes it doesn't work on sparse arrays
22:44 rubenwardy also, it may be worth adding a link in the Configure dialog too
22:50 sfan5 couldn't filter_types_type[1] as well be "" instead of nil?
22:53 sfan5 I was also thinking how you could make users actually click on it
22:53 sfan5 maybe add an annoying animation :D
22:54 rubenwardy snippy for Minetest
22:54 rubenwardy Hey! It looks like you haven't got any games installed
22:55 rubenwardy also, adding a small piece of text saying "Games" would help the button bar quite a lot
22:56 rubenwardy Should we feature freeze this weekend?
22:57 rubenwardy In time for 2 months since 5.2.0
22:58 sfan5 hm there's a few PRs that'd be good to have before 5.3
22:58 rubenwardy ok
22:58 sfan5 but yes I agree a freeze should happen soon
22:58 rubenwardy make sure to add those to the milestone
22:59 rubenwardy https://github.com/minetest/minetest/milestone/16
23:00 rubenwardy I really need to start dog fooding the Android app
23:02 sfan5 rubenwardy: added the feature PRs I consider important
23:02 proller joined #minetest-dev
23:03 sfan5 paramat: do you remember what "night sky colour" in the milestone description refers to and if it's been solved already?
23:04 sfan5 there's also https://github.com/minetest/minetest_game/milestone/5 where the last two are most important (IMO)
23:07 rubenwardy ah yes
23:07 skyliner_369 repeatedly having to change the color mode for many of the textures from indexed to RGB is kinda sluggish
23:08 rubenwardy yeah, is there an issue about automatically converting to RGB when using a texture modifier?
23:08 sfan5 1) why do you have to do that? 2) I am 100% sure a script could do it for you
23:08 rubenwardy I have a feeling that ^[colorize doesn't work with indexed textures
23:09 rubenwardy I'll check
23:23 skyliner_369 I just have to go to Image>Mode>RGB in Gimp each time.
23:24 Lone_Wolf joined #minetest-dev
23:35 paramat oh ...
23:36 paramat sfan5 'night sky colour' is the part of set_sky API likely to change, that is, we are considering the RGB colour is used as-is instead of being made much darker
23:37 paramat so that has not been solved yet
23:38 paramat so issue #9508
23:38 ShadowBot https://github.com/minetest/minetest/issues/9508 -- Use actual colors passed to set_sky() for night sky and dawn sky
23:42 paramat however i do not consider that essential for 5.3.0
23:42 paramat no PR for it yet either
23:43 paramat i removed 'floatlands' and 'night sky colour'

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