| Time |
Nick |
Message |
| 00:01 |
fluxionary |
makes sense, just never done it that way and i don't see any explicit documentation |
| 00:01 |
erle |
fluxionary what do you use it for? |
| 00:01 |
fluxionary |
currently, just going to keep track of a constantly incrementing ID |
| 00:01 |
erle |
for what? |
| 00:02 |
fluxionary |
and wanted to initialize the value while the mod was loading cuz that's easier than writing "register_on_mobs_loaded" or something |
| 00:02 |
erle |
but what is the id for? |
| 00:02 |
erle |
i am asking because i have seen this use case ;) |
| 00:02 |
fluxionary |
tentatively starting a new mob API (because of course we need yet another one) |
| 00:02 |
erle |
and why does it need a counter per world? |
| 00:02 |
fluxionary |
and part of what it's going to do is not store huge quantities of metadata in the mob object itself |
| 00:02 |
MTDiscord |
<Jonathon> obligatory https://xkcd.com/927/ |
| 00:03 |
erle |
uh-oh |
| 00:03 |
fluxionary |
i haven't quite decided on a backend, but probably will use mod_storage |
| 00:03 |
erle |
i want no part in this |
| 00:03 |
erle |
but good luck |
| 00:03 |
fluxionary |
but an entity and its metadata will be identified w/ a unique numeric identifier |
| 00:03 |
fluxionary |
well, the metadata that isn't actually used by the engine directly |
| 00:04 |
erle |
i think this is the inner-platform effect? |
| 00:04 |
erle |
https://en.wikipedia.org/wiki/Inner-platform_effect |
| 00:04 |
fluxionary |
basically, the result of `get_staticdata()` will just be `{id = something}` |
| 00:04 |
erle |
what tangible benefit does this give? |
| 00:05 |
erle |
i mean mod storage is less reliable than basically every other way to store data in a world |
| 00:05 |
erle |
so you have to make up for that |
| 00:05 |
erle |
somehow |
| 00:05 |
erle |
so where is the advantage? |
| 00:05 |
fluxionary |
oh, not having to store all that metadata when you e.g. pick up the mob as a tamed pet |
| 00:05 |
erle |
oh i see, you want to make world downloads useless |
| 00:06 |
fluxionary |
well, in terms of stored pets, sure |
| 00:06 |
erle |
i am not sure how much metadata you are talking about |
| 00:06 |
erle |
is this a performance issue? |
| 00:06 |
rubenwardy |
using an incrementing id may be problematic - after a crash, you could end up with the incrementing id being lower than a mob in the world |
| 00:06 |
rubenwardy |
using a random guuid is probably a better idea |
| 00:06 |
fluxionary |
ask Sokomine, the two of use discovered that having a huge warehouse full of chests w/ 90 petz items in each, caused *huge* lag for the client when loading those areas |
| 00:07 |
fluxionary |
and also for some reason even affected FPS |
| 00:07 |
erle |
trivially true, yes |
| 00:07 |
fluxionary |
in any event, i just started on the project today, not sure where it will end up |
| 00:07 |
erle |
you get the same thing by putting shulkers in shulkers |
| 00:07 |
erle |
or making a warehouse full of books |
| 00:07 |
fluxionary |
erle: yup, too much metadata in the item |
| 00:08 |
erle |
no, it's too much metadata in the area |
| 00:08 |
fluxionary |
i suppose? but that's because there's too much metadata in the items? |
| 00:08 |
erle |
rubenwardy is correct btw. you wan get an almost-monotonously increasing value by using seconds since epoch and a random value after the decimal separator |
| 00:08 |
erle |
i do that in some code i wrote |
| 00:09 |
erle |
and unless your clock goes bonkers, it is guaranteed to have no collisions outside of a second |
| 00:09 |
erle |
just look at the different types of uuid |
| 00:10 |
erle |
fluxionary how much metadata does a single tamed pet have? |
| 00:10 |
fluxionary |
rubenwardy, possible, but i'm doing a call to `mod_storage:set_int()` before returning the next value. there's probably a race condition w/ saving to the db, but it seems unlikely and effectively unavoidable w/out some sort of actual locks |
| 00:10 |
fluxionary |
erle: for petz? around 10k i think? |
| 00:10 |
erle |
what |
| 00:10 |
fluxionary |
i'd have to go back and check |
| 00:10 |
erle |
are you saving an uncompressed bitmap of the pet or what |
| 00:10 |
fluxionary |
i think most mobkit mobs store way too much metadata, petz is pretty egregious |
| 00:11 |
fluxionary |
in that *every* mob stores *every possible key* for *every possible pet*, even when it doesn't make sense |
| 00:11 |
erle |
ohhhh |
| 00:11 |
erle |
look, the solution is to work smarter, not harder |
| 00:11 |
erle |
the enchantments in mineclone2 had a similar problem |
| 00:11 |
|
est31 joined #minetest |
| 00:11 |
erle |
each enchanted tool was storing every digtime |
| 00:11 |
erle |
so it had 21kb of metadata |
| 00:12 |
erle |
in mineclone2 0.71 this is the most accessible source of lag |
| 00:12 |
erle |
anyways, it was reduced to like 2kb |
| 00:12 |
fluxionary |
O_O |
| 00:12 |
fluxionary |
like the digtime for *every node*?! |
| 00:12 |
erle |
i'm not saying it was smart |
| 00:12 |
fluxionary |
that's hilarious |
| 00:13 |
erle |
in fact, almost all code i see, including my own, is super stupid if you look at it too long |
| 00:13 |
erle |
stare into the abyss for too long and the abyss stares back |
| 00:13 |
fluxionary |
in any event, what i want to do really isn't that hard to implement, and the users of the mod (other mod writers) won't have to think about it |
| 00:13 |
erle |
yeah, but you would have a better time and your users too |
| 00:14 |
erle |
if you would look at the “adventures” that kay27 had with his mod storage ideas |
| 00:14 |
erle |
you know, the griefing-proof mapgen that works fine as long as mod storage works perfectly |
| 00:14 |
erle |
he posted the failures of it himself |
| 00:14 |
erle |
if you crash at exactly the right time, it just is totally bonkers |
| 00:15 |
erle |
holes in the map and so on |
| 00:15 |
erle |
anyway, what's stopping you from doing the UUID suggestion? |
| 00:15 |
erle |
that's a very common pattern |
| 00:16 |
erle |
birthday paradoxon? |
| 00:17 |
erle |
fluxionary could you point me at the metadata code in question? maybe i can see a quick fix to drive the data down |
| 00:17 |
erle |
or have you fixed it already? has Sokomine? |
| 00:19 |
fluxionary |
sorry was afk for a moment, give me a moment to respond |
| 00:20 |
fluxionary |
erle: what storage backend was kay27 using? |
| 00:20 |
erle |
mod_storage? |
| 00:20 |
erle |
it doesn't really matter, the semantics are the same |
| 00:20 |
erle |
and i do not know |
| 00:20 |
fluxionary |
but which backend for that? |
| 00:20 |
fluxionary |
like, `minetest.serialize`, sqlite, postgres? |
| 00:21 |
fluxionary |
the semantics have changed |
| 00:21 |
erle |
in what way |
| 00:21 |
fluxionary |
(does it even do postgres) |
| 00:21 |
erle |
is it *in any way* more reliable? |
| 00:21 |
fluxionary |
in that writes are written to the backend immediately |
| 00:21 |
erle |
as if |
| 00:21 |
fluxionary |
really? |
| 00:21 |
erle |
well i don't believe it without proof, show pls |
| 00:21 |
fluxionary |
are the sql backends not doing what i think they do? |
| 00:21 |
erle |
the semantics of mod storage as i remember are “you have to save stuff manually” |
| 00:22 |
erle |
so where and when did they change? |
| 00:22 |
fluxionary |
let me find that |
| 00:22 |
erle |
i remember also another thing like that |
| 00:22 |
erle |
the nether portals |
| 00:22 |
erle |
they used mod storage |
| 00:22 |
rubenwardy |
fluxionary: mod storage only saves every 15 seconds or so, and I don't think it's synced with world saves(?). Also, the mod storage file could be individually corrupted or lost |
| 00:23 |
erle |
if you crashed the server in just the right way you could make a portal that would teleport the user into a lava cube |
| 00:23 |
rubenwardy |
oh do we have sqlite mod storage now |
| 00:23 |
erle |
what's the guarantee of sqlite? |
| 00:23 |
rubenwardy |
slightly better than spitting everything out to json every X seconds |
| 00:23 |
fluxionary |
it used to just not save until you quit (or after not-serious crashes) |
| 00:23 |
erle |
tbh if you have 10kbb metadata that's okay. but if you have them and they should not be 10kbb, maybe clean them up? |
| 00:24 |
erle |
i can help you probably doing that |
| 00:24 |
erle |
“not-serious crashes” lol |
| 00:25 |
erle |
you mean the once where you still get a trace? |
| 00:26 |
fluxionary |
erle: entities are the hardest thing for the client to deal w/. if you've got just a few hundred entities in an area, low-power machines drop to single-digit FPS |
| 00:27 |
erle |
rubenwardy can i do anything about that modpack i made which used to only hang the game and then i was told that's not useful and then i pivoted to a mod that fixes the hang by filtering the input (it is trivially triggerable as soon as users can trigger arbitrary location emerges) and then everyone seemed to think it was still the old mod or something. it is marked with action/change needed, but i got no advice what i should change |
| 00:27 |
erle |
IIRC. |
| 00:27 |
fluxionary |
erle: i'm not really sure what all the causes are, but i know that the size of the metadata that's sent to clients is at least correlated |
| 00:27 |
erle |
rubenwardy i remember us talking about it some time back and you seemed to have some opinions on testing or so? |
| 00:27 |
fluxionary |
sqlite isn't just the available, it's the default, and it'll toss out your old mod_storage if you aren't paying attention (or at least it used to?) |
| 00:28 |
erle |
i wonder why that was the case |
| 00:28 |
erle |
like why would it not be the default to migrate the data? |
| 00:28 |
fluxionary |
that, i don't know |
| 00:29 |
erle |
fluxionary i know about entity problems. but storing a lot of stuff in a mapblock is rarely an entity problem. |
| 00:29 |
fluxionary |
it might have been because i was using dev versions, i never looked into it |
| 00:29 |
fluxionary |
and none of my local worlds mattered |
| 00:30 |
erle |
well either there is a good reason, or someone was impossibly lazy |
| 00:30 |
fluxionary |
erle: or it was an "early adopters" bug or something |
| 00:30 |
fluxionary |
not that i was trying to adopt that feature specifically |
| 00:30 |
erle |
well, i have encountered a lot of people who have the mindeset ”it's done, i don't care about the legacy stuff” |
| 00:31 |
erle |
one of the more toxic programmer memes IMO |
| 00:31 |
fluxionary |
erle: i care about the legacy stuff, but at the same time, there's only a finite amount of attention in the world =D |
| 00:31 |
erle |
compare minetest and minecraft |
| 00:31 |
erle |
migrating worlds is a breeze in minetest |
| 00:31 |
erle |
in minecraft … i sometimes read the mojang bugtracker just to be amused |
| 00:31 |
erle |
it's much funnier than the minetest one |
| 00:32 |
fluxionary |
erle: i've been working on re-doing a number of major mods, and i've tried very hard not to break any legacy stuff that was actually functional |
| 00:32 |
erle |
for example, they have a bot that autocloses issues on trigger words |
| 00:32 |
erle |
IIRC if a user comments “this works for me” or so hehe |
| 00:32 |
fluxionary |
erle: i'm surprised they have a public-facing bugtracker |
| 00:32 |
erle |
it's a flaming pile of garbage though |
| 00:33 |
erle |
the funniest thing is when the devs themselves confidently assert something about the mechanics and users then point out it is wrong |
| 00:33 |
erle |
also a lot of people post stuff like ”my 2FA is not working” hehe |
| 00:33 |
erle |
closing that is legit |
| 00:34 |
fluxionary |
i haven't used minecraft since i found this game, i'm pretty sure given all the weird things they've done to it, i probably won't be able to get it to work for me |
| 00:34 |
erle |
back to the mob metadata thing |
| 00:34 |
erle |
i have an old laptop and i know the pains |
| 00:34 |
fluxionary |
back to that |
| 00:34 |
fluxionary |
i feel that mod_storage is a reliable backend now |
| 00:35 |
fluxionary |
but i've also left what i've written so far open to adopting some other backend |
| 00:35 |
erle |
cool then you can also use it for chests hehe |
| 00:35 |
erle |
breaking world downloads completely |
| 00:35 |
fluxionary |
erle: that doesn't break world downloads completely |
| 00:36 |
fluxionary |
erle: already, your local world shouldn't be getting info about the contents of other player's locked chests |
| 00:36 |
erle |
if you save the inventory in mod_storage? |
| 00:36 |
erle |
i disagree on principle |
| 00:36 |
erle |
about the lag, there are several things |
| 00:36 |
fluxionary |
erle: it breaks chest contents, for sure. but there's literally a `metadata:mark_as_private()` feature already |
| 00:36 |
erle |
people are not using it ;) |
| 00:37 |
fluxionary |
erle: i am |
| 00:37 |
erle |
oh nice |
| 00:37 |
erle |
i believe it reduces traffic |
| 00:37 |
fluxionary |
most explicitly in my smartshop fork |
| 00:37 |
erle |
one simple thing that you can do is look at the amount of traffic that the mapblock generates btw |
| 00:37 |
erle |
and look if there are any ghost inventories around the area |
| 00:37 |
fluxionary |
erle: yup, i know how to use wireshark :) |
| 00:37 |
erle |
do you have an up to date dissector for the protocol? |
| 00:38 |
fluxionary |
i don't have any better disector for the protocol than what ships currently |
| 00:38 |
erle |
:/ |
| 00:38 |
erle |
there was an area on clamity that had issues like the ones you had |
| 00:38 |
fluxionary |
though i wrote a couple ad-hoc scripts |
| 00:38 |
erle |
because of a lot of chests |
| 00:38 |
erle |
turns out that a big issue was leftover unused metadata in the world |
| 00:38 |
fluxionary |
though that was a couple years ago |
| 00:39 |
fluxionary |
erle: interesting. left-over after... what? worldedit? |
| 00:39 |
erle |
after duping |
| 00:39 |
fluxionary |
like a physical node duping? |
| 00:40 |
fluxionary |
using a piston or something? |
| 00:40 |
erle |
oh tell i do not know piston duping hehe |
| 00:40 |
erle |
i only know how to create the elusive standalone piston head |
| 00:40 |
erle |
which can not be mined |
| 00:40 |
fluxionary |
there was an ... interesting bug related to pistons and machines from the terumet mod |
| 00:40 |
erle |
but blown up |
| 00:40 |
erle |
tell |
| 00:40 |
erle |
the duping made it so that you could have inaccessible inventories not associated with chests |
| 00:41 |
erle |
those were full of shulkers, or even shulkers full of shulkers |
| 00:41 |
erle |
a peculiar feature of item metadata is that you can stuff A LOT more data into an item than you can into an entity (and games rely on it lol) |
| 00:42 |
erle |
and recursive shulkers are one of the worst cases i know |
| 00:42 |
erle |
anyways, the area was full of it |
| 00:43 |
erle |
upside: if you visit such an area and have the appropriate CSM, you can literally pick items out of thin air |
| 00:43 |
erle |
even on vanilla client |
| 00:44 |
erle |
fluxionary if you do not show me the 10kb pets metadata code, i am not going to try to figure out how to crush it |
| 00:44 |
fluxionary |
oh, i can show you petz? |
| 00:45 |
fluxionary |
https://github.com/runsy/petz/ |
| 00:45 |
erle |
well i am pretty bad in lua, but good enough to have made some mods and sped up some code |
| 00:45 |
erle |
so where is the part about the pet meta |
| 00:46 |
erle |
https://github.com/runsy/petz/blob/master/petz/petz/bunny_mobkit.lua take this |
| 00:46 |
erle |
how does this generate 10kb of meta |
| 00:46 |
erle |
that would be like 3 times that file |
| 00:46 |
erle |
which is incidentally, 3.33kb hehe |
| 00:47 |
erle |
fluxionary tell me about big chungus |
| 00:47 |
erle |
ples |
| 00:47 |
erle |
please |
| 00:48 |
erle |
so far i have not found anything |
| 00:49 |
erle |
i remember using this mod i think |
| 00:49 |
erle |
werewolves |
| 00:50 |
erle |
https://github.com/runsy/petz/blob/096d2d4cd806ee8e58f12d0c529a89a604c5bef2/petz/api/api_capture.lua |
| 00:51 |
erle |
local stack_meta = new_stack:get_meta() |
| 00:51 |
erle |
local ent = self.object:get_luaentity() |
| 00:51 |
erle |
local staticdata = ent:get_staticdata(self) |
| 00:51 |
erle |
stack_meta:set_string("staticdata", staticdata) |
| 00:53 |
erle |
github code search is so incredibly dumb |
| 00:53 |
erle |
no results for “stat”, but 46 for “staticdata” |
| 00:54 |
erle |
https://github.com/runsy/petz/blob/68eb2bed092ff1687991b13e163031fd13f00ac8/kitz/engine.lua#L736 |
| 00:54 |
erle |
oh lol |
| 00:54 |
erle |
capturing an animal serializes its memory |
| 00:55 |
erle |
funny |
| 00:55 |
erle |
fluxionary, i still do not see how this gets to 10kb |
| 00:56 |
erle |
or how it is remotely bad |
| 00:56 |
erle |
it sounds a lot of unncessary features but quite cleanly implmeented |
| 01:07 |
|
fling joined #minetest |
| 01:08 |
|
YuGiOhJCJ joined #minetest |
| 01:10 |
|
YuGiOhJCJ joined #minetest |
| 01:13 |
|
sys4 joined #minetest |
| 01:15 |
|
ecraven joined #minetest |
| 01:18 |
|
asdflkj_sh joined #minetest |
| 01:18 |
fluxionary |
it's apparently not 10k, a single test on a single mob shows 3.5k |
| 01:26 |
|
M34c164e3-3084-4 joined #minetest |
| 02:03 |
|
Verticen joined #minetest |
| 03:04 |
|
Verticen joined #minetest |
| 03:13 |
|
sys4 joined #minetest |
| 04:00 |
|
MTDiscord joined #minetest |
| 04:44 |
MTDiscord |
<luatic> erle: petz isn't cleanly implemented at all |
| 04:48 |
|
lemonzest joined #minetest |
| 04:54 |
|
Yad joined #minetest |
| 05:19 |
|
calcul0n joined #minetest |
| 05:41 |
|
Verticen joined #minetest |
| 05:58 |
|
Lesha_Vel_ joined #minetest |
| 06:00 |
|
TomTom joined #minetest |
| 06:14 |
|
sys4 joined #minetest |
| 06:50 |
|
appguru joined #minetest |
| 07:09 |
|
Lunatrius joined #minetest |
| 07:20 |
|
Lunatrius joined #minetest |
| 08:18 |
MinetestBot |
[git] Wuzzy2 -> minetest/minetest: Fix lua_api typo: animated_images f0703f3 https://github.com/minetest/minetest/commit/f0703f3c5b8ed51d2c7caf83ffe547bf54bc1eb1 (2022-07-29T08:18:30Z) |
| 08:20 |
|
Lunatrius joined #minetest |
| 08:20 |
|
Talkless joined #minetest |
| 08:21 |
MinetestBot |
[git] sfan5 -> minetest/minetest: Allow direction keys with autoforward again 6611d7e https://github.com/minetest/minetest/commit/6611d7e1ee3b078b5c3414bf57833755d8b43620 (2022-07-29T08:19:23Z) |
| 08:21 |
MinetestBot |
[git] appgurueu -> minetest/minetest: Fix media overriding regression (#12602) 99c8295 https://github.com/minetest/minetest/commit/99c8295e71eb86333ae21b1083b6f55fb9479565 (2022-07-29T08:19:36Z) |
| 08:31 |
|
Talkless joined #minetest |
| 09:01 |
|
specing_ joined #minetest |
| 09:03 |
|
mrgreymatter joined #minetest |
| 10:01 |
|
appguru joined #minetest |
| 10:23 |
|
YuGiOhJCJ2 joined #minetest |
| 10:30 |
|
erle joined #minetest |
| 11:28 |
|
jonadab joined #minetest |
| 11:37 |
|
YuGiOhJCJ2 joined #minetest |
| 11:37 |
|
fling joined #minetest |
| 11:37 |
|
est31 joined #minetest |
| 11:45 |
|
appguru joined #minetest |
| 12:03 |
|
ronoaldo joined #minetest |
| 12:31 |
|
YuGiOhJCJ2 joined #minetest |
| 12:36 |
|
definitelya joined #minetest |
| 12:41 |
|
YuGiOhJCJ2 joined #minetest |
| 13:04 |
|
appguru joined #minetest |
| 13:10 |
|
skunky3 joined #minetest |
| 13:10 |
MinetestBot |
[git] RichardTry -> minetest/minetest_game: Update Russian translation for mtg_craftguide (#2873) c744829 https://github.com/minetest/minetest_game/commit/c744829ce5e8b8a67439e5e33571569a25a3ad47 (2022-07-29T13:10:24Z) |
| 13:13 |
MTDiscord |
<luatic> https://github.com/minetest/minetest/milestone/19 is 5.6 supposed to release in two days and there are still 5+ issues to be resolved (half the things on the list are pulls to fix issues tho), so I suggest considering to kick #12131 off the milestone if it isn't fixed swiftly |
| 13:13 |
ShadowBot |
https://github.com/minetest/minetest/issues/12131 -- Add `minetest.settings` to CSM API and allow CSMs to provide `settingtypes.txt` by AFCMS |
| 13:13 |
skunky3 |
hey, im wondering: is there a similar system to minecraft ticks/random ticks in minetest? basically a global timer that calls a callback on nodes every x seconds or so? |
| 13:13 |
rubenwardy |
See https://rubenwardy.com/minetest_modding_book//en/map/timers.html |
| 13:13 |
rubenwardy |
!title |
| 13:13 |
MinetestBot |
rubenwardy: Node Timers and ABMs - Minetest Modding Book |
| 13:14 |
sfan5 |
!book Node Timers |
| 13:14 |
MinetestBot |
sfan5: Node Timers and ABMs - https://rubenwardy.com/minetest_modding_book//en/map/timers.html |
| 13:14 |
sfan5 |
hey that works |
| 13:14 |
rubenwardy |
oh cool. When was that added? |
| 13:14 |
rubenwardy |
oh wait, did I do that? |
| 13:14 |
sfan5 |
didn't you add it yourself |
| 13:14 |
rubenwardy |
probably |
| 13:14 |
rubenwardy |
I must be thinking of cdb that doesn't have a command |
| 13:15 |
skunky3 |
thanks, i was more looking in the other official refference, but ill check it out |
| 13:17 |
rubenwardy |
the official reference is that good to learn from if you don't know what to ctrl+f |
| 13:17 |
rubenwardy |
*isn't |
| 13:17 |
rubenwardy |
the official reference isn't good to learn from. You need to know what to ctrl+f |
| 13:26 |
skunky3 |
i have been figuring out some things from there but interesting |
| 13:28 |
|
fluxionary joined #minetest |
| 13:33 |
skunky3 |
on abms, is it possible to check if the node is covered by another node on top, or does that have to be checked in the callback? |
| 13:34 |
sfan5 |
if you don't see it in the definition in lua_api.txt then you need to check manually in the callback |
| 13:34 |
sfan5 |
@luatic I commented a fix there, this took me 10 minutes; someone unfamiliar with the code might need up to an hour but it's not impossible, so I'm wondering if just nobody cares |
| 13:35 |
|
appguru joined #minetest |
| 13:36 |
skunky3 |
for example i want to implement grass turning into dirt if its covered by another normal type node |
| 13:37 |
erle |
skunky3 for everything that does not happen very rarely, you will probably almost always want to use an ABM. each node timer uses memory per node to keep state. ABMs do not. |
| 13:38 |
erle |
skunky3 i think you can define an abm with adjacent nodes as a constraint and then can check in the abm handler |
| 13:38 |
erle |
that is, in my experience, fast enough |
| 13:38 |
sfan5 |
short of very elaborate workarounds (like implementing in-world node tracking yourself) this is currently the most efficient way: https://github.com/sfan5/minetest_classic/blob/c05f2690d36e814bea03ae6a8194a3fdee852183/mods/default/abm.lua#L24-L37 |
| 13:39 |
sfan5 |
and also simplest |
| 13:39 |
erle |
about ”very rarely”, for an oven you can use a timer. the advantage of a timer is that it will keep state even if the game is restarting. you don't want your smelter to restart once you reload the game. |
| 13:41 |
skunky3 |
i see, so abms seem seem to be the way to go for crops growing, and similar |
| 13:41 |
erle |
i think this is not exactly the most efficient way, even if it is the canonical way. IIRC you can save a tiny tiny amount of time by not using the vector function here but constructing the table yourself. |
| 13:42 |
sfan5 |
abs are more "you have often occurring nodes and semi-random stuff happens to them once in a while" |
| 13:42 |
sfan5 |
nodetimes are "you have specific nodes where something happens on an exact schedule" |
| 13:42 |
sfan5 |
you could implement crop growing with both |
| 13:42 |
sfan5 |
but for e.g. furnace you'd go with nodetimers |
| 13:43 |
sfan5 |
and natural grass growth is a definite usecase for ABMs |
| 13:43 |
erle |
there are indeed several approaches to growing crops. depending on what you want, you can use timers, but you will have a very uniform growth schedule then in a naive implementation. |
| 13:44 |
skunky3 |
thanks for the info, i will now be messing around with these things on my own again! |
| 13:44 |
erle |
kay27 also has a weird thing where he tries to track the time elapsed outside of the game in mineclone5 i think. please don't do it, but maybe look at it as an example of what not do to. it is barely achieving his goals, i bet it will not achieve any of yours. |
| 13:45 |
erle |
IIRC the idea was something like “my crops will grow the same if i let the game run for an hour or if i quit it and restart it an hour later” or so? i do not remember. |
| 13:45 |
erle |
but certainly a niche goal that *could* be useful, but most likely is not |
| 13:46 |
erle |
if you, for example, have a game that syncs its day/night cycle to the local time |
| 13:46 |
erle |
skunky3 is there any place where you publish your code? i am curious, having done a bit of crop growing stuff myself |
| 13:47 |
erle |
maybe i learn of a new approach or new gameplay way (like the kay27 way, i did not even think of that before i sawo it) |
| 13:48 |
skunky3 |
not yet, im still very much in the learning phase, i will publish my code once im away from experimenting, and actually making something playable |
| 13:49 |
erle |
skunky3 a warning: for every process that can go exponential, do not use node timers. i.e. if some plant is spawning two other plants. literally every fire mod that uses node timers leads to lag if the fire becomes large enough. |
| 13:49 |
erle |
even if you have 16G RAM instead of 4G like me, the difference is merely a few generations until it is full of node timers |
| 13:50 |
skunky3 |
i can imagine, i plan on going very much on the efficent route instead of that |
| 13:51 |
erle |
i am just saying that because it changes your outlook on how the logic has to work |
| 13:51 |
erle |
you can have node-timer-based fires and ABM-chance-based-fires that are the same in gameplay terms |
| 13:51 |
erle |
but they are implemented very differently |
| 13:51 |
erle |
and the ABM fire never lags the computer ;) |
| 13:51 |
erle |
except, maybe, if you have issues looking at many flames lol |
| 13:52 |
skunky3 |
i basically thought of almost always using abms, except for when timers are really better |
| 13:52 |
erle |
oh, also don't forget about ABM catch-up |
| 13:52 |
skunky3 |
i saw |
| 13:52 |
erle |
you might or might not want your stuff to depend on if the area is loaded |
| 13:53 |
erle |
also LBMs are unreliable unless you set them to always execute, so just forget them for this kind of purpose, they are not the right tool for the job anyway (but i think i have seen LBM-based growing code) |
| 13:54 |
erle |
skunky3 so what is your idea, still the biome blending thing? |
| 13:54 |
erle |
just like, more colors? |
| 14:02 |
skunky3 |
ill see, just more colours for now |
| 14:23 |
|
cynthiune joined #minetest |
| 14:23 |
|
sparky4 joined #minetest |
| 14:29 |
|
YuGiOhJCJ2 joined #minetest |
| 14:36 |
|
kilbith joined #minetest |
| 14:36 |
kilbith |
the Black Mesa mainmenu: https://i.imgur.com/DwockxN.png |
| 14:57 |
erle |
i believe this is again a console main menu, with no clear indications what exactly is a click target, it also suffers from small fonts (again) and everything being in french. ;) |
| 15:42 |
skunky3 |
on abms: is it possible to check which neighbouring block made the condition true, if it is set? |
| 15:43 |
skunky3 |
i could check all neighbouring blocks manually, but im unsure about its performance |
| 15:44 |
erle |
skunky3 you are very unlikely to run into performance issues using ABMs and even if you do, the ABM evaluation is just going to skip a bunch of things |
| 15:44 |
erle |
just don't overdo it |
| 15:45 |
erle |
like, grass spread was never an issue |
| 15:45 |
erle |
but in mineclone2 for example, i think the vine growing ABM did something bad |
| 15:45 |
erle |
it was consistently the one that took the most time when i measured it |
| 15:45 |
erle |
bubble column ABM also programmed really badly |
| 15:46 |
erle |
skunky3 the most important thing about any ABM solution is that you need to be able to afford the ABM not firing |
| 15:46 |
|
proller joined #minetest |
| 15:46 |
erle |
for example, IIRC in mineclone5, nether portals are despawned using ABMs |
| 15:46 |
erle |
so they might hang around for a time if there is too much ABM lag around |
| 15:47 |
erle |
the game will notlag |
| 15:47 |
erle |
but the ABM evaluation will be inhibited |
| 15:47 |
sfan5 |
if you have to use find_nodes_in_area |
| 15:47 |
sfan5 |
or was it find_node_in_area? |
| 15:48 |
erle |
for the portal despawning case there exist multiple algorithms featuring timers, most of them recursive |
| 15:48 |
erle |
and one that is the simplest, just bulk set the nodes |
| 15:48 |
erle |
an ABM is just wrong there |
| 15:48 |
skunky3 |
ive been thinking about making abms that spread grass, over other grass, and then smoothly interpolate between their 2 colours |
| 15:48 |
erle |
as it also affects portals that do not need to despawn (a portal despawns if its edge meets a non-obsidian-and-non-portal node) |
| 15:49 |
erle |
you can do that, but |
| 15:49 |
erle |
the most difficult thing for this will be the grass note color interpolation |
| 15:49 |
erle |
node |
| 15:50 |
erle |
skunky3 what is your solution for the grass node coloring? just a reasonable 256 color palette? |
| 15:50 |
skunky3 |
oh i already have an idea for that |
| 15:50 |
erle |
i mean 16 × 16 = 256 |
| 15:51 |
erle |
so you can smoothly blend a bunch of things |
| 15:51 |
erle |
if you only want single-accent color |
| 15:51 |
erle |
skunky3, tell |
| 15:51 |
erle |
how are you going to solve it? maybe i can steal the idea hehehehehe |
| 15:51 |
MTDiscord |
<Benrob0329> kilbith: Black Mesa's menu reminds me that I need to play it, there was a ton of thought and effort behind that game |
| 15:51 |
skunky3 |
making 16 blocks, so that i can define colours via rgb |
| 15:52 |
erle |
skunky3 you mean my R5G5B5 proposal? or just 16 base colors? |
| 15:52 |
MTDiscord |
<Benrob0329> But I think that the tab hierarchy would do us some good |
| 15:52 |
erle |
with my proposal, you get 4096 different colors for blending |
| 15:52 |
skunky3 |
R5? isnt it 4 bits? |
| 15:52 |
erle |
Benrob0329 there are similar, but much better thought out ideas in the issue |
| 15:53 |
skunky3 |
or am i having a moment |
| 15:53 |
MTDiscord |
<Benrob0329> Ok, and? |
| 15:53 |
erle |
skunky3 you are right! off by one error. |
| 15:54 |
erle |
2¹⁵ = 4096 = 16 × 256 |
| 15:54 |
erle |
no |
| 15:54 |
|
fling joined #minetest |
| 15:54 |
erle |
that was what i thought |
| 15:54 |
erle |
but it is 2¹² |
| 15:54 |
erle |
so you have only 4 bits |
| 15:54 |
erle |
per color channel |
| 15:54 |
skunky3 |
i think your proposal before had more textures in it |
| 15:55 |
skunky3 |
but, with 16 i can parse hex colour codes of the form #AAA quite easily |
| 15:56 |
skunky3 |
so the interpolation is actually the least difficult thing for me |
| 15:57 |
erle |
well, it is if you want more colors |
| 15:58 |
erle |
you need some way to distribute the colors to the nodes palette |
| 15:58 |
erle |
programmatically |
| 15:58 |
erle |
each color representing a base node and a param2 |
| 15:58 |
erle |
which with R4G4B4 are 2⁴ bits + 2⁸ bits |
| 15:59 |
skunky3 |
well that seems quite simple to me, ima be honest |
| 15:59 |
erle |
it is not too hard if you don't have an issue with all your base nodes being red or so hehe |
| 15:59 |
MTDiscord |
<MisterE> Isn't there a colorized drawtype, that has a whole palette? |
| 15:59 |
erle |
MisterE, yes, but only 256 colors |
| 15:59 |
erle |
and ironically, it does not read the colormap of the image i think |
| 15:59 |
erle |
not sure |
| 16:01 |
skunky3 |
well it will be 16 nodes, with each 256 colours, where all of them use the same textures but do some other funk to make them different colours |
| 16:02 |
skunky3 |
there wont really be a way to get grass node items without param2 set, except via creative |
| 16:03 |
|
est31 joined #minetest |
| 16:04 |
erle |
skunky3 i suggest to be a bit smarter about the bit fiddling. for the 16 nodes, use 1 red bit, 2 green bits and 1 blue bit |
| 16:04 |
erle |
for the base nodes |
| 16:04 |
erle |
IIRC the human eye is more sensitive to green color changes |
| 16:04 |
erle |
in terms of perceived brightness |
| 16:05 |
erle |
so you could make it so that the node you pick up is approx. the same color as the blended node ig |
| 16:05 |
erle |
like one of 16 colors matching closest ideally |
| 16:06 |
|
est31 joined #minetest |
| 16:06 |
skunky3 |
i mean the node you pick up, keeps the param2 informantion tho |
| 16:06 |
|
fling joined #minetest |
| 16:06 |
skunky3 |
like, the colour is kept the same |
| 16:06 |
erle |
in the inventory too? |
| 16:07 |
skunky3 |
yes |
| 16:07 |
skunky3 |
as long as its not gotten through creative |
| 16:07 |
erle |
so it is rendered with that color? |
| 16:07 |
skunky3 |
yes |
| 16:07 |
erle |
nice |
| 16:07 |
erle |
then it's super easy |
| 16:07 |
skunky3 |
yeah |
| 16:07 |
erle |
i did not know |
| 16:08 |
erle |
please show when you are done |
| 16:08 |
erle |
this is interesting |
| 16:08 |
skunky3 |
i can send a link to the script and image once it works with growing, just wont be doing stuff with biomes yet |
| 16:09 |
erle |
yeah do |
| 16:11 |
erle |
now i am also motivated to try it |
| 16:11 |
skunky3 |
just keep in mind its just standalone rn, and will neither generate, nor set its colour when given through creative inventorys |
| 16:12 |
skunky3 |
and im just setting param 2 through the development test param2 tool for testing |
| 16:13 |
|
sys4 joined #minetest |
| 16:16 |
erle |
i am writing code to generate the palettes now |
| 16:20 |
skunky3 |
my implementation will use a singular palette |
| 16:21 |
erle |
for 16 nodes? |
| 16:21 |
skunky3 |
yes |
| 16:22 |
erle |
why |
| 16:22 |
erle |
you can have 16 palettes |
| 16:22 |
skunky3 |
and you can have 1 palette |
| 16:23 |
erle |
but then you only ever need 1 node |
| 16:23 |
skunky3 |
no |
| 16:23 |
erle |
why |
| 16:23 |
skunky3 |
i will show soon when it works |
| 16:39 |
erle |
skunky3, check this out https://mister-muffin.de/p/cjLg.txt |
| 16:41 |
erle |
1. use R4G4B4 colors for blending (bit shift if you are lazy, better use the sample depth rescaling algorithm like i did here) |
| 16:41 |
erle |
2. have 16 different nodes. red color channel is the node |
| 16:41 |
erle |
3. green color channel × blue color channel is the param 2 |
| 16:44 |
erle |
skunky3, does that help you? |
| 16:45 |
erle |
oh |
| 16:45 |
erle |
i did something wrong here |
| 16:47 |
erle |
i thin the correct param2 is (15 - green) × blue |
| 16:53 |
skunky3 |
unsure where i would upload the image files, or the zip file |
| 17:04 |
skunky3 |
but i deffinetly got all the colours there |
| 17:07 |
|
sys4 joined #minetest |
| 17:44 |
|
___nick___ joined #minetest |
| 17:46 |
|
___nick___ joined #minetest |
| 17:54 |
erle |
skunky3 https://mister-muffin.de/paste |
| 17:56 |
skunky3 |
does that support files besides text ones tho? erle |
| 18:02 |
|
sys4 joined #minetest |
| 18:04 |
skunky3 |
erle its a tar.gz file https://mister-muffin.de/p/xiUb.gz |
| 18:06 |
skunky3 |
ima still need to make the growing code tho |
| 18:07 |
erle |
skunky3 i don't think the multiply trick works |
| 18:08 |
erle |
you do not get a uniform color distribution that way |
| 18:08 |
erle |
admittedly, in perceptual terms, my approach also does not give you one |
| 18:08 |
erle |
but it's much better |
| 18:10 |
erle |
i see why you are only using one palette though |
| 18:10 |
erle |
it's pretty color |
| 18:10 |
skunky3 |
i dont really mind tbh, my goal is just to get all colours that can be written as #3f6 |
| 18:10 |
erle |
but i think you will get banding issues |
| 18:10 |
erle |
well, that is what my approach does too |
| 18:11 |
erle |
maybe i am missing something here |
| 18:11 |
skunky3 |
it only multiplies one of the colour channels of the palette |
| 18:12 |
skunky3 |
and thats the colour thats always at 255 |
| 18:12 |
erle |
oh hmm |
| 18:12 |
erle |
ig it's equivalent then |
| 18:13 |
erle |
palette = "neko_grass.png^[multiply:#FF"..ihex.."FF" |
| 18:13 |
erle |
nah i doubt it |
| 18:13 |
skunky3 |
yes, the green channell is the one that doesnt get changed |
| 18:14 |
erle |
but if you muiltiply the other channels with 255, you always get 255 for them? |
| 18:14 |
erle |
like i am not sure, maybe i do not understand how the texmod works |
| 18:14 |
erle |
but i assumed it would multiply channel-wise |
| 18:15 |
skunky3 |
remember that colour values in multiply work with 0-1 instead of 0-255 |
| 18:15 |
erle |
and in multiplication the neutral element is not 255. is it in the texmod? |
| 18:15 |
erle |
oh |
| 18:15 |
erle |
i see |
| 18:15 |
skunky3 |
thats how colour multiplication works in general |
| 18:15 |
erle |
then i guess our solutions are equivalent, only i use more disk space and you use more RAM and CPU |
| 18:15 |
erle |
i can assure you that i just assumed and did not look it up |
| 18:16 |
erle |
i had a teacher once who said “if you assume, you make an ass out of you and me” höhöhö |
| 18:16 |
erle |
so yes, smart trick. but where did you get the base image? |
| 18:16 |
erle |
for the colormap |
| 18:17 |
skunky3 |
well i tried it in game, and works |
| 18:17 |
erle |
did you do this in an earlier project? |
| 18:17 |
skunky3 |
i just made it using gradients in an image editor, was quite simple |
| 18:17 |
erle |
oh lol |
| 18:18 |
erle |
we have widely different ideas on reproducibility of artifacts i think |
| 18:18 |
erle |
ig it's pretty much the same then anyways |
| 18:18 |
erle |
wildly |
| 18:19 |
skunky3 |
i dont like to automate something that ill only do once anyway |
| 18:24 |
skunky3 |
dont get me wrong, i like automating things, but i usually weigh it out as to how usefull it is |
| 18:47 |
|
Verticen joined #minetest |
| 18:55 |
|
vampirefrog joined #minetest |
| 19:03 |
|
mazes_83 joined #minetest |
| 19:04 |
|
mazes_83 joined #minetest |
| 19:11 |
|
vampirefrog joined #minetest |
| 19:20 |
erle |
skunky3 oh, i built my think in a way that you can have different color depths |
| 19:20 |
erle |
or different color distributions |
| 19:20 |
erle |
in the future |
| 19:21 |
skunky3 |
ahh i see, im fine with it being at this scale, since its best fit to colour codes |
| 19:24 |
|
xigoi left #minetest |
| 19:25 |
erle |
skunky3 well, you might, for example, not use the entire rgb colorspace |
| 19:25 |
erle |
anyway, pls tell me about your mod when you are finished! |
| 19:27 |
skunky3 |
im planning to make a game rather then a mod, just making mods rn to learn the ins and out of minetest |
| 19:49 |
skunky3 |
does minetest.find_nodes_in_area make sense for an area as small as 3x3x3? |
| 19:51 |
erle |
skunky3 benchmark it |
| 20:04 |
MinetestBot |
[git] RichardTry -> minetest/minetest_game: Update Russian translation 2bdc7c8 https://github.com/minetest/minetest_game/commit/2bdc7c8c7d9eb35185c9f90c371f65f105ab112a (2022-07-29T20:03:06Z) |
| 20:20 |
MinetestBot |
[git] Wuzzy2 -> minetest/minetest: Update builtin translation files 3b37bcd https://github.com/minetest/minetest/commit/3b37bcd994ee455496818708b8efc488aaf2777b (2022-07-29T20:20:48Z) |
| 20:20 |
MinetestBot |
[git] Wuzzy2 -> minetest/minetest: Update German builtin translation 6a269d5 https://github.com/minetest/minetest/commit/6a269d58ef28be6428a5fb62c0eb966fc1ff5883 (2022-07-29T20:20:48Z) |
| 20:27 |
|
proller joined #minetest |
| 20:28 |
|
appguru joined #minetest |
| 21:06 |
|
sys4 joined #minetest |
| 21:08 |
|
Yad joined #minetest |
| 21:14 |
|
WhiskeyPotato joined #minetest |
| 21:39 |
|
dabbill joined #minetest |
| 21:42 |
|
erle joined #minetest |
| 21:44 |
skunky3 |
the grass is growing, and merging |
| 21:46 |
|
dabbill joined #minetest |
| 21:52 |
|
specing joined #minetest |
| 21:54 |
|
Talkless joined #minetest |
| 21:54 |
|
olliy joined #minetest |
| 21:57 |
arielaw[m] |
hey y'all, are there any tutorials for grokking the map generation process? there's some funky stuff I want to do with terrain generation for a game but I'm not sure how I should go about it |
| 22:02 |
|
Thermoriax joined #minetest |
| 22:04 |
arielaw[m] |
the wiki isn't quite doing it for me, I was wondering if there's something more along the lines of "hey here we have these parameters, with the valley/v7 mapgen and THIS is how the result looks" |
| 22:10 |
|
Verticen joined #minetest |
| 22:17 |
|
kaeza joined #minetest |
| 22:22 |
|
Boingo joined #minetest |
| 22:28 |
MTDiscord |
<MisterE> I don't think so. |
| 22:28 |
MTDiscord |
<MisterE> There are plenty of mapgen mods tho |
| 22:34 |
|
panwolfram joined #minetest |
| 22:38 |
arielaw[m] |
Yeah I've read a little about subterrane for example and it seems like it'll be useful for what I want to do, but you still need to more or less understand how perlin noise works to use this stuff, no? |
| 23:03 |
|
est31 joined #minetest |
| 23:26 |
|
Verticen joined #minetest |
| 23:30 |
|
Lesha_Vel joined #minetest |
| 23:36 |
|
Juri joined #minetest |
| 23:55 |
|
sys4 joined #minetest |
