Time Nick Message 06:53 Zeno` Hi all. Just here to hand in my official resignation 06:54 Zeno` Maybe I shall return one day (I had a great time working with all the devs and other people on this project). I just don't seem to have the interest at the moment. *sigh* 06:57 Zeno` Removed myself from the repo (You left minetest. You will be missed.) 06:57 * Zeno` sheds a tear 06:58 * Zeno` waves 07:05 kahrl nooo Zeno` :( 07:05 * sofar waves back 07:05 kahrl I hope it wasn't the Sisyphean nature of the irrlicht abstraction layer that burned you out 07:18 sofar ho hum. why does minetest.get_meta() not return nil if the meta field isn't present? 07:24 kahrl sofar: at the API level, node metadata is always considered present 07:25 kahrl sofar: but it may be empty 07:25 kahrl (if it's empty, nothing will be stored in the database) 07:25 sofar I'm getting a '0' back on nodes that don't have the property 07:25 sofar so I can't use that to distinguish between nodes that have an int property set and ones that don't 07:26 sofar I know the meta stuff is always present 07:26 sofar this is about a specific int value inside the meta fields 07:27 kahrl check if get_string(name) returns an empty string 07:27 sofar I'm sorry, I just had convulsions 07:27 sofar lol 07:29 sofar I'd rather use '1' as minimal value then, doing a totally odd workaround like that... shudder (if it even works, that's hella ugly) 07:32 kahrl I guess it would be nice to have a 'contains' method in NodeMetaRef 07:32 kahrl (though it could technically implemented in lua using to_table(), that's quite wasteful) 07:32 sofar I can see why you'd be worried about returning nil 07:33 sofar probably break a few mods 07:33 sofar maybe add warnings to the code if a property is asked that isn't present in the nodemetaref? 07:33 kahrl yeah, changing the behaviour of get_int at this point is out of the question 07:36 kahrl warnings would just lead to warning spam, even though in most cases there's nothing wrong 07:37 sofar if you think that returning a value for a requested field is OK, sure, then it's just spam 07:37 sofar I'd argue that it's a bug 07:37 sofar or, bad programming of API's 07:37 sofar maybe I'm too picky 07:37 kahrl sofar: the API is what is has been for several years 07:38 kahrl it can't be changed now since many mods rely on it 07:38 sofar never changing an API means you never improve 07:38 sofar but sure, I get it 07:39 kahrl breaking every mod ever is not an improvement 07:40 sofar without a warning you have no data that mods are relying on that behavior 07:41 sofar I'm not that well into the core code ... I suppose I can see if I can try it myself to add a warning 07:41 kahrl it's inevitable with mods that rely heavily on node metadata, like technic 07:41 kahrl (not saying that I checked that technic specifically relies on it) 07:46 sofar hmm, NodeMetadataList::get() nicely returns NULL 07:47 sofar it's probably in the lua bindings then 07:52 kahrl add haskell bindings to minetest and make get_int returns a Maybe Int :P 07:52 kahrl return* 07:53 sofar ahahaha 07:53 sofar someone was talking about javascript 07:53 sofar tbh lua is fine... 07:55 sofar it's too late for me to care... 09:04 realbadangel__ mods like technic are using meta a lot 09:04 realbadangel__ pipeworks and mesecons too 09:05 realbadangel__ and those are the most complex mods for mt out there 09:14 realbadangel__ breakin api will piss lotsa folks 15:42 est31 bye Zeno. 16:32 est31 :( 17:29 VanessaE_ sorry to see you go, Zeno` :( 17:49 hmmmm ahhh noo zeno` :( 17:50 hmmmm lol I guess we're not getting a graphics engine-independent minetest 17:50 hmmmm back to the drawing board... 17:52 VanessaE_ well I guess the bright side there is that by now, it would have been such a big rebase effort that someone would have to rewrite it anyway. 17:55 who_wants_some minetest development pace is too fast 17:55 who_wants_some sorry, wrong channel! 18:34 realbadangel__ est31, i have analyzed the possibilities, only solution is real client side modding 18:34 realbadangel__ and im not experienced enough on that field to introduce it 18:35 realbadangel__ jeija on the other hand made a PR with proof of concept 18:35 realbadangel__ https://github.com/Jeija/minetest/tree/clientside_mods 18:36 realbadangel__ if somebody make this in, i can move further with client side shaders 18:37 realbadangel__ atm i dont wanna invent another override file 18:38 realbadangel__ ive read the code for all overriding stuff, like noise, textures 18:39 realbadangel__ all is just wrestling, freestyle 18:39 realbadangel__ imho real client side mods are the only solution 18:41 Calinou yeah, we should have it some day 18:41 Calinou even Quake modders eventually stopped doing everything server-side 18:41 Calinou they introduced CSQC ;) 18:41 Calinou and now they cram more and more things into it 18:42 Calinou (because clients are often more powerful than servers…) 18:42 realbadangel__ ofc clients are more powerfull 18:42 realbadangel__ we do buy ROG boxes 18:43 realbadangel__ or build them on our own 18:43 realbadangel__ server can run on a hostin space 18:43 realbadangel__ client is about performance and FPS 18:44 realbadangel__ and NICE lookin 18:45 Calinou my VPS is 1 CPU (2 GHz), 1 GB RAM 18:45 Calinou my laptop is 8 threads (4C/8T), 8 GB RAM, and a dedicated card… 18:45 who_wants_some my fear is when it becomes client-side now the clients will have even more bad stutters 18:45 who_wants_some or crashes 18:45 Calinou that's the server owner's problem if they distribute poorly performing mods 18:46 Calinou crashes are preventable, but work needs to be done on that 18:46 realbadangel__ who_wants_some, when we will have client side modding you will get mobs, carts etc 18:46 realbadangel__ with no lags 18:46 realbadangel__ atm its impossible 18:46 who_wants_some I hope 18:47 who_wants_some please fix liquids over ignore and slow mesh gen that causes big stutters (ap-tly) 18:48 realbadangel__ liquids flow is not my area of interest at all 18:48 realbadangel__ i mean physics 18:48 realbadangel__ i do gfx only 18:49 who_wants_some texture tear problems *cough* 18:49 who_wants_some or 3166 18:49 realbadangel__ thats hardware 18:49 realbadangel__ and 3166? ooops ;) 18:49 realbadangel__ im stuck 19:05 who_wants_some he is stuck 19:05 who_wants_some he needs some new idea from outside 19:05 who_wants_some "suggestion" 19:06 RealBadAngel i know how it should work, but i dont know how to code it 19:07 RealBadAngel i started the work, its far enough to be considered a workin solution 19:07 RealBadAngel it just have some flaws which i cant fix atm 19:07 VanessaE_ it works, it improves things, and doesn't break anything, right? 19:07 VanessaE_ and nothing is worse off 19:08 VanessaE_ then it should be merged. The double-update can be solved later. 19:08 RealBadAngel propably i could but with spending lotsa time which i could use to develop other things 19:08 VanessaE_ better two nodes' worth of updates than 7 entire mapblocks 19:09 RealBadAngel i gave up dancing with double-update and picked shaders again 19:09 VanessaE_ hmmmm: ^^^^^ 19:09 who_wants_some that is slow 19:09 RealBadAngel so in two days i made mod definitions of shader use 19:09 RealBadAngel you can try now water and lava 19:10 RealBadAngel somebody already said that lava is "piece of art" 19:10 est31 yippee server is up 19:10 RealBadAngel and i really think it is 19:11 RealBadAngel just compile and watch it :P 19:11 est31 I try to keep out of these discussions 19:11 est31 some will love it, some will hate it 19:11 rubenwardy Calinou, putting more code client side is bad if it makes the server vulnerable to cheating 19:11 est31 well, it depends 19:11 est31 if the server side code trusts the inputs, then yes. 19:11 RealBadAngel est31, can you take a look on Jeija's client side modding proof of concept? 19:11 est31 that code is years old 19:11 rubenwardy est31, "makes the server vulnerable to cheating" 19:12 RealBadAngel est31, it works 19:12 rubenwardy it works doesn't mean it's good 19:12 RealBadAngel code doesnt have age 19:12 rubenwardy hmmmm was working on a prototype 19:12 RealBadAngel it works or not 19:12 rubenwardy but he stopped for some reason - fixing stuff elsewhere 19:12 rubenwardy celeron55 wanted to do it after farview 19:13 rubenwardy RealBadAngel, it's not black or white. 19:13 RealBadAngel but neither its grey 19:13 rubenwardy if something "works" but it's unreliable, then it might as well not work 19:13 RealBadAngel or at least it shouldnt 19:14 who_wants_some server needs some cheating protection already, there are hacked clients already :( 19:14 RealBadAngel atm development is stuck 19:14 Calinou rubenwardy, obviously it's not going to be cheat-sensitive code 19:14 Calinou but Minetest is pretty vulnerable to cheating as it is right now 19:14 rubenwardy who_wants_some, fly and fast are vulnerability 19:14 RealBadAngel see? 19:14 est31 fast is no vulnerability 19:14 est31 you can just adjust the speed setting 19:14 rubenwardy because the server doesn't do proper client-side prediction and server reconcilliation 19:14 RealBadAngel we are all needing help from each other 19:14 rubenwardy yes, to cheating 19:14 who_wants_some rubenwardy, so server needs to check player movement anyway 19:14 Calinou est31, everyone disables anticheat actually :P 19:15 Calinou it's buggy, constantly resets innocent players 19:15 rubenwardy anticheat isn't real server side reconcilliation 19:15 est31 Calinou, then dont complain that minetest doesnt have anti-cheat 19:15 est31 but that its anti-cheat is buggy 19:15 est31 those are different things 19:15 est31 but yeah fly and noclip still remains 19:16 hmmmm i think anticheat should be a mod 19:16 rubenwardy Why? 19:16 hmmmm it's ancilliary functionality to begin with 19:17 hmmmm its inclusion into the core implies that minetest is a "game" where people can "cheat" whereas really that all depends on the modpack installed 19:17 RealBadAngel i think we have started a storm in a glass of water 19:17 RealBadAngel hi Taoki :) 19:54 rubenwardy I agree that farview as it is last time I used it looked out of place. But, WIP 19:56 hmmmm [02:44 PM] c55 and hmmmm are too fucused on pure code, both treating it as kinda excercise, to show own abilities 19:56 hmmmm no, it's because I realize that talking about things is useless because the code is not getting done 19:56 hmmmm minetest is a big project for sure but it needs even more development 19:57 hmmmm talking is bad because it makes you feel like you're accomplishing something when you aren't 19:57 est31 it looked great IMO last time I looked at it, just had two bugs. 19:58 est31 first is hard to describe 19:58 est31 I guess some driver thingy or so 19:58 est31 second is that it allocates tons of memory 19:58 est31 but the second was promised to get fixed 20:00 who_wants_some_ est31, is it possible to upload to GPU not 40 meshes in very limited time (causing big stutter), but spread them over time, so stutter is say 1ms per each rendered frame? 20:01 RealBadAngel hmmmm, for the single commiter thats nearly impossible now to get something done without touchin a few other areas of experitse 20:01 RealBadAngel and thats what im talkin about 20:02 who_wants_some_ est31, i've tried very simple flatgen mod and I still got pretty big stutter when new meshes generated (if trust F5 graphs), it is basically 15ms -> 30ms -> 15ms (that is too big and not acceptable) 20:03 RealBadAngel i can know everything about shaders (and i know that you dont know a single shit bout it). and vice versa, im trying to get something done in threads and all i get are lols 20:04 RealBadAngel even when what i did to get my idea workin was copy paste of existing already solutions 20:04 est31 yea I guess the shaders I did were shit too. 20:05 est31 and your shaders are great 20:05 RealBadAngel hah, i could do better :P 20:06 RealBadAngel but seriously est31 what shaders code you made? 20:06 RealBadAngel no offence, just asking 20:07 est31 tbh I haven't coded any real shaders yet 20:08 RealBadAngel so what have you meant? 20:08 hmmmm real developers don't pidgeonhole themselves into a single technology subset 20:08 RealBadAngel we are not real 20:08 hmmmm saying that you can't do threads but you can do shaders is ridiculous 20:09 hmmmm you are not real maybe 20:09 RealBadAngel well i am, sort of, a nick name :P 20:09 est31 I have expressed wrongly, it should have been "the shaders I would do would be shit" 20:09 est31 haha 20:09 RealBadAngel est31, you have tested water and lava 20:10 RealBadAngel can u express your opinion now? 20:10 RealBadAngel and here 20:11 est31 I said its nice, but I dont know whether it should be the standard shader. 20:11 who_wants_some_ is someone has binaries for windows with new shaders, pls share with me, i want to see it on ati 20:12 RealBadAngel those are not standard shaders 20:13 RealBadAngel could be if mt would go HD and fine lookin 20:13 who_wants_some_ and slow working 20:13 RealBadAngel theyre pretty fast 20:14 RealBadAngel and moreover they eliminate some nasty visual bugs 20:14 RealBadAngel like different looking of block with same texture 20:15 RealBadAngel +s 20:15 RealBadAngel point of my latest PR is that, any node can have own shader 20:16 RealBadAngel independent from the engine at all 20:16 who_wants_some_ that is useful 20:17 RealBadAngel i wanted it first to be able to send shader code over net as a mod 20:17 RealBadAngel i dont think that could be a security problem but whatever 20:17 RealBadAngel client side modification pack is better place for it 20:18 RealBadAngel client is having good box and gpu, then hes choosing candies 20:18 RealBadAngel no matter the server 20:27 RealBadAngel so, for starters, https://github.com/minetest/minetest/pull/3391 20:27 RealBadAngel what are your votes? 20:30 RealBadAngel above pull with will need client side modding for the feature to be complete ofc 20:31 RealBadAngel but both sides will require those definitions 21:12 RealBadAngel splitty day, i will ask again tomorrow ;)