Minetest logo

IRC log for #minetest-hub, 2018-12-19

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

All times shown according to UTC.

Time Nick Message
00:46 aerozoic joined #minetest-hub
01:39 ANAND joined #minetest-hub
02:25 CBugDCoder joined #minetest-hub
03:16 NathanS21 joined #minetest-hub
03:53 garywhite joined #minetest-hub
04:00 FrostRanger left #minetest-hub
07:00 jluc_ joined #minetest-hub
08:30 twoelk joined #minetest-hub
08:46 jas_ joined #minetest-hub
09:32 FrostRanger joined #minetest-hub
09:45 Calinou joined #minetest-hub
11:32 Fixer joined #minetest-hub
11:49 jluc joined #minetest-hub
11:53 MinetestSam joined #minetest-hub
12:25 Unarelith joined #minetest-hub
12:36 kilbith joined #minetest-hub
12:37 kilbith https://i.imgur.com/Bdc2Qd2.jpg
12:37 kilbith The first 2.5D FPS game in a formspec, using raycasting
12:37 kilbith ~10s to generate a frame
12:43 nerzhul ouch 10sec is huge
12:43 nerzhul for a such trivial model
12:44 rubenwardy well, it's a collection of boxes
12:45 rubenwardy although, that is surprisingly long
12:45 nerzhul do you want to add model viewer on formspec ?
12:45 kilbith there are half a million of boxes
12:45 rubenwardy are you using table.concat
12:45 kilbith no
12:45 nerzhul we see ~50 polygons
12:45 kilbith I will release the code soon
12:45 nerzhul maybe they are stacked
12:47 kilbith Each box is a pixel, there are no genuine "polygons"
12:47 kilbith These are fake
12:47 kilbith Well, this is how OpenGL is working actually
12:48 rubenwardy switching .. with table.concat will speed it up
12:48 kilbith Let's see
12:50 rubenwardy well, *Should
12:50 rubenwardy .. involves copying, table.concat involves slightly less
12:50 rubenwardy string.format is the best, but doesn't allow variable args
12:51 rubenwardy by table.concat, I mean appending to a table with t[#t + 1] and calling table.concat right at the end
12:53 rubenwardy there's also the copying from Lua to the engine, and across the wire, and then to the client
12:55 kilbith Yep, definitely faster
12:55 kilbith THanks
12:56 kilbith About a second to make a frame now
12:58 nore <@rubenwardy> by table.concat, I mean appending to a table with t[#t + 1] and calling table.concat right at the end <-- I'd advise not appending to the table like that
12:58 nore but instead using an index that holds the position to insert to and incrementing it
12:58 nore because computing #t is linear in #t, so appending is quadratic as well
12:59 kilbith This is even faster, yes
13:02 kilbith Any good video recording software on Windows 10?
13:03 rubenwardy oh yeah
13:04 rubenwardy OBS
13:04 rubenwardy quite heaby though
13:04 nerzhul obs, like on linux
13:04 nerzhul it's used by many youtube streamers
13:05 kilbith I don't mind if it's not OSS though
13:05 rubenwardy I use simplescreenrecorder on Linux, but no Windows support
13:05 rubenwardy there aren't any good non-OSS screenrecorders
13:05 rubenwardy well
13:05 rubenwardy all the ones I found add watermarks
13:05 nerzhul OBS is the most used as i said
13:05 rubenwardy OBS doesn't and is OSS
13:05 nerzhul and it has youtub estreaming support embedded :)
13:06 kilbith VLC?
13:06 nerzhul VLC doesn't record screen
13:07 nerzhul OBS permits to record the screen, a part of the screen, a window, do transition between record sources
13:07 nerzhul it's very nice
13:13 tumeninodes joined #minetest-hub
13:14 * tumeninodes is impressed https://i.imgur.com/Bdc2Qd2.jpg
13:15 kilbith Demo: https://www.youtube.com/watch?v=xtQXNJ8n3p0
13:16 kilbith Ok this is dog slow
13:17 tumeninodes screen recording sucks in 2018
13:18 tumeninodes I've been looking at external screen recorders, they seem the better way to go
13:21 tumeninodes just speed up the playback on youtube :P
13:27 skyfight joined #minetest-hub
13:52 scr267 joined #minetest-hub
14:04 ChimneySwift you can set a screen as a VLC capture device, not recommended
14:05 tumeninodes so, this will be fun to add into mods such as stairs, walls, etc. : /  https://pastebin.com/5erRfFfu
14:06 tumeninodes if there is an ultra easy method to do this, someone please let me know :P
14:07 tumeninodes as this will only impact nodes which have multiple versions/texture varients
14:12 scr267 joined #minetest-hub
14:35 IhrFussel joined #minetest-hub
14:36 IhrFussel Wait wait wait... you always say "my server, my rules...obey or leave" but now you consider adding a way to allow the CLIENT to NOT obey to SSCSM sent mods?? That makes absolutely NO sense
14:36 IhrFussel If a client wants to play on MY server they have to ACCEPT all terms of it INCLUDING sent mods I want them to run!
14:37 IhrFussel Don't give clients the freedom to CHOOSE which server sent CSMs run on them
14:37 IhrFussel nerzhul ^
14:38 IhrFussel Don't take "client freedom" THAT far, if a client wants full freedom they need to play SP
14:38 IhrFussel Plain and simple
14:39 nerzhul IhrFussel why i should accept all your rule, it's like disable JS on a website or block adds
14:39 nerzhul these rules are not only for server owners but also for clients if they want. It's interesting to see servers owners only think about their servers but doesn't always think about their gamers/clients
14:40 nerzhul i just returned the CSM drama point of view to the gamer point of view :)
14:40 IhrFussel You HAVE TO accept all rules on a server or else you can get punished...it was ALWAYS like that ... don't compare it to JS, cause most websites don't need it for essential features... when a server wants to run a CSM it will most likely have a very good reason for it
14:40 sfan5 IhrFussel: just like you can disable JS in a browser, clients will have the option of accepting no CSMs from the server
14:40 sfan5 if you do not like that, it's on your server to reject clients that do this
14:41 sfan5 (similarily to how modern websites or even Cloudflare's DDOS protection do not work with javascript disabled)
14:41 nerzhul yes; if server owner can have choice, the client should have choice too
14:41 IhrFussel sfan5, nerzhul wants a way to disable CERTAIN CSMs only and not a flag to enable/disable all sent mods
14:41 IhrFussel That will complicate things  A LOT
14:41 nerzhul and we can enhance protocol at a point to make serverlist attribute "SSCSM mandatory" permitting clients to filter servers like this
14:42 sfan5 certain CSMs? not sure if that is useful from a client perspective
14:42 nerzhul if particle spawners sent from server make me laggy, why not ?
14:42 IhrFussel He expects the client to then tell "if you disable this CSM then feature x won't be available"
14:43 nerzhul currently i just expose the idea that if we have server CSM restriction we need to also think about client SSCSM restrictions to make their game correctly works as intended. It's an idea, nothing is developed
14:43 nerzhul and we can flag the server with SSCSM mandatory in server list, like we did with creative for example
14:44 nerzhul and when the client connects if he disabled SSCSM or refuse SSCSM server just disconnect it
14:45 nerzhul the client should have the choice too. It's not server owners dictatorship :) freedom is what we always expected in mt
14:45 IhrFussel But if a client then refuses to run SSCSM x then the code needs to be run server side? I mean you surely cannot just remove essential server features the need to be run somewhere
14:45 IhrFussel they*
14:45 nerzhul IhrFussel, i suggest you to discuss about this when we will start to think about that, and it's not in the 5.0.0 dev cycle
14:46 nerzhul first release 5.0.0, after prepare 5.1.0 with SSCSM and how to implement the correct behaviour for everybody
14:46 nerzhul i agree with you that offloading to client is nice
14:47 IhrFussel The client shouldn't be able to "hog" the server just because it refuses to run SSCSM...that is my point
14:47 nerzhul but i think some heavy calculation requires to be restricted if needed by client. offloading particles is very very nice, but a restriction (or just a setting for a such use case) permits the client to disable particles and increase its FPS
14:47 kilbith Code's released: https://github.com/kilbith/raycast
14:48 nerzhul IhrFussel, some contributors refused to contribute on GH because of JS , we accept them. I think we should accept client who refused server mods too
14:48 nerzhul if the client doesn't want the SSCSM you have two choice: he connects on your server, has a bad experience and disconnects forever or he is informed that there is SSCSM and can just not connect
14:48 nerzhul it's its choice, not yours
14:48 nerzhul his*
14:49 IhrFussel So server then need to run a SS mod AND send a SSCSM mod to clients that allow it ... that kinda defeats the purpose of SSCSM
14:49 nerzhul it's interesting to see server owners are equal in terms of behaviour
14:50 nerzhul (some server owners)
14:50 nerzhul always extremists
14:50 nerzhul it's a compromise, and a contract with players. By default the setting should not be enabled but client who doesn't want to run code from servers can be able to refuse to go to those servers
14:50 IhrFussel No, but until now you always told people "servers decide what you do on there, if you need full control you can play SP"
14:51 nerzhul IhrFussel, extremist.
14:51 nerzhul We can have well balanced servers.
14:51 nerzhul and there is
14:51 shivajiva nerzhul: stop talking crap, how dare you presume all server owners are the same, that's like me saying all devs are rude...
14:52 IhrFussel I said CORE DEVS said the above thing and agreed on it ... I saw it multiple times on GH issues
14:52 nerzhul if i want to play multiplayer without running non controlled code it's my choice, not yours (note: i personnally don't care about this on MT)
14:53 rubenwardy I agree with allowing users to opt out, however the API should be designed in such a way that it doesn't matter to the server in a lot of cases
14:53 IhrFussel Yes it can be your choice, but then the server also has the choice to just refuse you... and that would make things very complex if the client needs to have a flag for EACH possible SSCSM mod/callback and needs to compare it with the server on connect
14:54 nerzhul IhrFussel it's not server role to refuse clients who doesn't want SSCSM, but in the implementation i think it's the only implementation possible, and it's core side :)
14:54 rubenwardy Client-side mods sent by the server should also contain a signature created by ContentDB, with certified information such as mod name author and hash
14:55 nerzhul IhrFussel, as a first step have a opt out for all is the easier to handle, but we can think in 5.2 for example to have fine grained things. Remember the particle example, i think it's a good example without being too disruptive on the gameplay
14:55 nerzhul rubenwardy i agree with you, but we need a CA i think to sign those mods
14:55 IhrFussel Don't force servers to run code they want to off-load server side JUST because clients don't want to run SSCSMs is what I say
14:56 IhrFussel Don't give clients control over which code runs server side and which doesn't
14:56 rubenwardy Anything you offload shouldn't matter to the server
14:56 nerzhul IhrFussel i don't think we can restrict CSM hud offloaded from servers for example, because it's a gameplay thing. Same with formspecs
14:56 rubenwardy If it matters to the server, then that's a vulnerability - don't trust the client
14:57 nerzhul but restricting 3D performance used by SSCSM to permit users to play correctly on low devices is a good choice
14:57 DI3HARD139 joined #minetest-hub
14:57 nerzhul particles, light...
14:57 nerzhul i will open the debate on that part when the code will be mature enough
14:57 nerzhul there is no point in 5.0.0 as there is no SSCSM
14:57 rubenwardy That's more of a graphics setting though, than a csm thing
14:58 shivajiva if we have to have this bastardisation of csm you presented then both server and client can turn it off, the server can reject the client if it wishes and the client can choose to play elsewhere. Anything else is unacceptable. I find celeron55 to be polite and you to be rude but I'm not daft enough to claim all devs are the same...
14:58 nerzhul rubenwardy, yeah maybe a restrction based on graphic settings
14:58 IhrFussel So basically...what you want is just a client info that says "This server uses SSCSMs and if you disagree to load it then you will experience inferior gameplay" and then the particles the server uses (inside SSCSMs) are just gone? That is not a good idea but okay
14:58 rubenwardy Also, I hate all these acronyms
14:59 Unarelith do you have a better one IhrFussel?
14:59 Unarelith +1 rubenwardy
14:59 nerzhul SPCSM is better
14:59 nerzhul because there is CPCSM
14:59 nerzhul it's more logical :D
14:59 nerzhul we can also remove S
14:59 nerzhul CPCM and SPCM
14:59 nerzhul client mods
14:59 nerzhul "side" is useless :p
14:59 Unarelith :D
15:00 rubenwardy I think client scripts may be a more accurate term, maybe
15:00 rubenwardy But doesn't matter
15:00 nerzhul rubenwardy: if content store sign the mods sent from servers, why don't distribute it too ?
15:00 rubenwardy Well, they could contain more information than .lua
15:00 rubenwardy nerzhul: celeron55 is against that
15:00 nerzhul okay
15:00 IhrFussel Unarelith, YES don't implement a flag for each callback and refuse the client server side as soon as it disallows CSMs sent by the server
15:01 nerzhul rubenwardy, you don't distribute SM ?
15:01 rubenwardy He's against relying on any other services to connect to a server
15:01 nerzhul but if you verify signature you do it :p
15:02 IhrFussel The client needs to either allow all or none ... everything else will make it hell for server owners
15:02 IhrFussel But you don't seem to care much about us
15:02 rubenwardy The signatures would be generated before the server installs the csm
15:02 rubenwardy Then verified without a connection
15:02 nerzhul i care about everybody who is constructive :)
15:02 rubenwardy Because the client will have cdb's public key
15:02 rubenwardy This was sofar's suggestion btw
15:02 nerzhul i don't see how to implement that :p
15:02 nerzhul but okay
15:02 nerzhul at least we can start without signing i hope
15:03 Unarelith IhrFussel, so explain why it would be hell for you?
15:03 rubenwardy well, at some point we will have a secure hashing function to validate builtin
15:03 rubenwardy do the same with client/ folders (and ignore and signature files)
15:04 rubenwardy then compare that with the hash signed by CDB's private key in the signature file
15:04 rubenwardy I mean, heh
15:04 IhrFussel If you have a server with 200 mods (like mine) and want to offload certain features to the client BUT that client disallows features x and y and you STILL want to make it work for them then you have to code ANOTHER server side mod JUST to be able to make it work for them
15:04 rubenwardy basically, I'd like there to be some method of securing saying the author of a mod at the bare minimum
15:05 rubenwardy you can't show a list of SPCMs names and authors without it being signed, because the server could lie
15:05 IhrFussel We are NOT talking about a stupid website here that doesn't need all the fancy stuff with JS to work... we talk about a GAME and every single feature a client misses over another can be dramatic
15:05 Unarelith IhrFussel, well it depends on what features can be disabled. if it's only graphical features then it's a good idea. if it forces modders/server owners/game makers to duplicate code it's not.
15:06 rubenwardy A better approach would probably have all authors have an RSA key, but this is more complicated
15:06 Unarelith rubenwardy, why would it be?
15:06 rubenwardy the benefit of CDB integration is all versions of CMs will be archived
15:07 rubenwardy well, it would reduce dependency on CDB
15:07 rubenwardy this does need more thought really
15:08 nerzhul rubenwardy: at least we have openssl dep through curl :p
15:08 rubenwardy yeah
15:08 nerzhul but we will make curl and openssl mandatory in MT build then
15:08 nerzhul (it should, MT without curl is a little bit limited)
15:08 rubenwardy I don't think that's unreasonable
15:08 rubenwardy because yeah, cURL is a little central
15:09 rubenwardy I mean, you could make this non-cryptographic just by asking ContentDB what it knows about a particular hash
15:09 rubenwardy but this adds a potentially hard dep on CDB
15:09 nerzhul rubenwardy the RSA key for MT server owners can be complicated. It's done like this on nextcloud apps, you should look at their store implementation
15:09 nerzhul i have a private key and a public key signing releases signed by their CA
15:09 nerzhul each release is signed with openssl
15:09 Unarelith rubenwardy, what if you decide to separate the launcher from the game itself?
15:10 nerzhul openssl dgst -sha512 -sign ~/Téléchargements/ocsms.key ~/Téléchargements/ocsms-2.1.0.tar.gz | openssl base64
15:10 nerzhul i do this :)
15:10 Unarelith the launcher can actually have a hard dep on CDB
15:10 rubenwardy s/launcher/menu/g
15:10 rubenwardy :)
15:10 nerzhul please don't create a launcher...
15:10 Unarelith well, it's more like a game launcher to me
15:10 Unarelith and with the integration of CDB it really looks like one
15:10 nerzhul using the same binary is nice
15:11 Unarelith why?
15:11 nerzhul less code dup, only one window to manage
15:11 nerzhul and also it's a game
15:11 Unarelith code dup? you just compile with the same code.
15:11 nerzhul we can share code like server/client yes
15:11 nerzhul but it's soo 1990 to use launcher on games :D
15:12 Unarelith ._.
15:12 Unarelith all modern games has launchers nerzhul
15:13 rubenwardy not really
15:13 nerzhul the witcher 3: no, world of warcraft : no (i can use Wow directly), assassin's creed: no, all RTS games i play (and there is plenty): no
15:13 Unarelith nerzhul, Steam is a launcher
15:13 nerzhul but yes there is some launchers, in skyrm
15:13 rubenwardy Steam is not a launcher
15:13 nerzhul i can play without Steam
15:13 rubenwardy that's stupid
15:13 nerzhul without Battle.net
15:13 rubenwardy that's like saying MyFileManager is a launcher
15:13 rubenwardy sure, it starts the program
15:13 nerzhul and download updates
15:14 nerzhul but it's the only feature
15:14 rubenwardy so does apt, though
15:14 Unarelith rubenwardy, you can download texture packs, mods, update the program, backup your saves
15:14 Unarelith sounds like a launcher to me
15:14 rubenwardy that's just stupid
15:14 nerzhul Unarelith you want to only connect to server with MT main window ?
15:14 nerzhul when disconnecting returning back to your desktop ?
15:15 Unarelith what do you mean?
15:16 Unarelith rubenwardy, Steam is just a "superlauncher" to me, and imo Minetest should have a separate launcher, I'll explain why:
15:16 jas_ joined #minetest-hub
15:16 Unarelith 1) If someone ever decides to create a new launcher from scratch using any other libraries, he could do it
15:17 Unarelith 2) Since the launcher is separated, a hard dep to CDB isn't a big deal
15:17 nerzhul except game should validate mod just before loading :)
15:17 Unarelith and?
15:18 Unarelith don't tell me what should happen, but explain why it won't.
15:27 benrob0329 joined #minetest-hub
15:29 aerozoic joined #minetest-hub
16:07 Shara nerzhul: your attitude toward server owners is disgusting
16:07 MinetestBot Shara: Dec-12 07:39 UTC <paramat> see http://irc.minetest.net/minetest-hub/2018-12-12#i_5456929 onwards
16:08 jas_ :D
16:08 Shara Your comment on github too: "i think SSCSM should have restrictions from client side too. If a client doesn't want server to use some API it should have the freedom to do it. And yeah server owners will not like this suggestion"
16:08 Shara Don't assume all server owners are tyrants who want to lord it over players.
16:08 Shara I want clients to have control as well
16:08 Shara You are just revealing your bias over and over
16:08 nerzhul If i'm so rude just re-read what you said about players and CSM :)
16:09 Shara What exactly have I said?
16:09 nerzhul Shara i like your opinion about client and freedom
16:09 Shara (or do you direct that to someone else?)
16:09 nerzhul please note in the gh original discussion opened 3 or 4 days ago nobody answered to my technical question whereas i said it 3 times :)
16:09 Shara Well, you also ignored a bunch of thigns server owners said many times
16:10 jas_ so what
16:10 Shara things*
16:10 nerzhul then if i'm acid with some people sorry, but just answer my questions when i ask some informations ;:)
16:10 jas_ stahp
16:10 Shara The point is everyone is tired
16:10 nerzhul Shara: i think it's the point yes
16:10 nerzhul my point is enhance the CSM mod loading restriction to restrict builtin and rename it in master
16:11 Shara Server owner speaks. You imply they are stupid or don't read, while you ignore what they say.
16:11 Shara Round and round in stupid silly circles.
16:11 nerzhul trigger that feature freeze, fix the real bugs (CSM is not a bug, i think we have bugs in other parts than CSM in core no ?) and release the 5.0.0
16:11 Shara nerzhul: I have tried to recommend compromise
16:11 Shara Tried to say: let's talk about defaults or similar
16:11 Shara No.. ignored every time
16:11 nerzhul my compromise is extend the CSM mod loading restriction to the whole lua stack
16:12 Shara CSM may not be a 'bug' but it should never have been merged. That alone means you should speak with a bit more respect
16:12 jas_ wow
16:13 Shara It's not okay for you to assume all server owners are bad bad people and speak like that
16:13 nerzhul Shara, sorry but i can't do compromise for a such sentence.
16:13 Shara Sorry, I don't understand what that means
16:13 jas_ what's the big deal anyway?  sorry i'm out of the loop.
16:13 nerzhul when you will answer my GH questions, not answer to questions i never ask, we can go ahead :)
16:13 nerzhul jas_ give me money dude and i give you drugs :D
16:13 Shara Oddly enough, I have
16:14 jas_ i got your money dude
16:14 jas_ check is in the mail
16:14 jas_ ""
16:14 Shara You ignored mine though, and it's been left to paramat again to try and answer
16:14 nerzhul jas_: heh
16:14 jas_ but what's the matter?  what's even the question?
16:14 nerzhul Shara: paramat has took time to understand the issue and do the compromise i proposed
16:14 nerzhul and it's the best thing we can do
16:14 nerzhul now go ahead, stop the battle and focus on real bugs
16:15 jas_ i wanna throw this out there real quick if i may:  when there are walls and walls of text, between bluebird, shara, and paramat, i honestly cannot read all that
16:15 nerzhul CSM is not used by majority of the users we should really focus on other code bugs
16:15 jas_ it's a tactic, i presume.  maybe unintentional, but effective nonetheless
16:15 Shara jas_: honestly, better not to
16:15 jas_ anyway
16:15 jas_ so what's the big conflict?
16:15 nerzhul jas_ trump wall with mexicans
16:15 jas_ check
16:15 jas_ !next
16:15 MinetestBot Another satisfied customer. Next!
16:15 Shara right now, the conflict is simply that I'm sick of nerzhul insulting server owners
16:15 jas_ ok.  nerzhul stop insulting server owners!
16:16 Shara show some respect, get some respect
16:16 nerzhul i never said: "servers owners are cowards"
16:16 jas_ (and optionally, apologize.)
16:16 jas_ good luck have fun o/
16:16 nerzhul i respect people who take time to read the doc and answer my question
16:16 Shara nerzhul: I never mentioned "cowards"
16:17 nerzhul i don't respect people complaining just for complaining without verifying the implementation, and who are based on a 0.4.16 code and think we never listen to them whereas many efforts has been done for them, but are not released yet
16:17 Shara Sure, which would be fine
16:17 Shara Until you say things like:
16:18 Shara nerzhul: it's interesting to see server owners are equal in terms of behaviour
16:18 nerzhul i fixed the sentence to add "some server owners"
16:18 nerzhul just after :)
16:18 Shara And on github you just assume server owners won't support a fair system for players
16:18 nerzhul the 5.0.0 restriction system is fair
16:18 nerzhul but nobody tested it
16:18 Shara Talking about: "i think SSCSM should have restrictions from client side too. If a client doesn't want server to use some API it should have the freedom to do it. And yeah server owners will not like this suggestion"
16:19 nerzhul complaining servers owners are just complaining based on asumption we never coded the restrictions
16:19 Shara I'm a server owner. I never assumed that
16:19 Shara I thanked you for working on it in fact
16:19 nerzhul It's real, some server owners will not like that, and just reread IhrFussel reaction ti's exactly what i said :)
16:19 Shara We're not all Fussel
16:19 nerzhul i know
16:20 nerzhul i need to fix my sentences to add some everywhere i think :p
16:20 Shara Then please remember it. That's what I am asking here.
16:20 Shara Thank you
16:20 nerzhul but also i said "server owners", neither "the server owners" or "all server owners"
16:20 nerzhul it's undefined
16:20 nerzhul in french we have a correct distinguish between "des" and "les"
16:21 nerzhul "des" is undefined , "les" point on the majority
16:21 nerzhul in english they drop the "des"
16:21 Shara Well I'm not the only one who takes offence at comments like this. Language barrier don't help, I know.
16:21 nerzhul i think so, because each language and translation done to native language can be with errors. What is your native ?
16:22 tumeninodes how the hell does everything here correlate to "Trump" and Trump wall with Mexicans... tired of seeing it. No one here I guess would like if I started spewing my political views
16:22 Shara English, or maybe bad english, or sometimes backwards english :P
16:22 tumeninodes keep IRL politics out of it all thank you, unless you want to hear the other side
16:24 Shara But I am so tired of the discussions on CSM not being constructive. I'm probably just going to out-shout anyone who fails to talk constructively at this point, at least in -hub.
16:24 Shara And that goes for both sides of the debate
16:24 tumeninodes we really should have a  no politics and no religion rule to this project, cuz it's starting to piss me off
16:24 Shara Well, they both fall under off-topic :P
16:29 Shara nerzhul: regarding the client being allowed to block things sent by the server: I really think this is fine. Servers can then decide how vital they want certain server-provided mods to be. Servers might choose to make certain things optional or required. If server-provided mods are required, the players get to make an informed choice about whether they want to accept or play somewhere else.
16:32 nerzhul Shara: the idea to have feature levels to make client choose the feature level is nice, if configured_feature_level >= feature_level => execute
16:32 nerzhul example: if CONFIG_FL_REQUIRE and feature_level is OPTIONAL => not execute
16:33 nerzhul a such thing must be implemented with SPCM then
16:33 Shara However it's done, clients need a choice
16:33 nerzhul i open an issue about that linked with 5.1.0 roadmap
16:38 nerzhul see #8002
16:38 ShadowBot https://github.com/minetest/minetest/issues/8002 -- Server Provided Client Mods: Required features
16:38 shivajiva I think nerzhul is simply avoiding the hard question, why is his version of csm a clone of the server api that lets clients add their own mods and why its acceptable to develop it to deliver mods when it didn't deliver the intent, that's what I really want to understand :)
16:40 nerzhul response: it's not a clone of server API. Some features has been cloned from server API to be offloaded on the client by client mods yes
16:41 nerzhul reponse2: why the server doesn't deliver mods: time to define the need, the serialization format, what embed, time to live outside MT with 3 childs
16:46 shivajiva so letting a client add it's own mods was the only method available to you and merging it forced that reality essentially though I suspect it was always part of your vision?
16:47 nerzhul do you know i'm not the only coredev ? :p
16:47 sofar kilbith: I'll post that to minetest-mods, thanks
16:48 nerzhul in fact client adding its own mod was the first method because we need to test APIs by ourselves while implementing them
16:48 nerzhul and after nothing more
16:48 nerzhul you can imagine we tested the API no ? :p
16:48 nerzhul it's why the code is as is
16:49 shivajiva It seems to me that you had a free hand in it and wasn't closely checked on the implications and conformity to the concept by the other devs tbh
16:49 nerzhul now go ahead, trigger the feature freeze after merging the restriction extent :)
16:49 nerzhul no i don't have a free and
16:49 nerzhul hand*
16:50 nerzhul remember one thing: mt is done by its contributors. celeron55 has given some devel ideas, but there is no real roadmap in the project becaus ethere is no project owner (in project management meaning, not a line on GH repository)
16:50 shivajiva I'm just making clear what my issue is so you don't think I haven't spent time reading and looking at this
16:51 nerzhul shivajiva: then answer my question : what CSM API found in the CSM API doc is a problem to release minetest because there is no restriction you can do server side ?
16:52 nerzhul i understand we don't have CPCM, it's a fact. But it's a feature and don't block minetest regular usage. Security issues are a blocker
16:52 nerzhul CPCM is not a security issue, i don't see how pushing mods to client can enhance their security
16:52 nerzhul SPCM sorry § in the two previous lines :)
16:52 nerzhul i'm tired ^^
16:53 shivajiva the api looks fine from that perspective, I have no issue with it. That's why I am making clear what my issue is
16:53 nerzhul you want a feature and it's not developd due to dev team time
16:53 shivajiva no
16:53 nerzhul i think it's the idea
16:54 nerzhul i will leave work, i will read your answer later today
16:56 shivajiva I wanted what was conceptualised by celeron55 in the discussion and subsequent poc he provided instead of this mess tbh
16:57 sofar it's not a mess imho
16:57 shivajiva the situation is :)
16:58 bobr joined #minetest-hub
16:58 bobr joined #minetest-hub
16:58 sofar has anyone actually looked at that poc?
17:01 sofar anyway, I don't think it's relevant as I believe that the current CSM API is more than complete enough to be used for server-sent, signature validated CSM and exclusively, meaning, flavors/restrictions can be removed entirely, and we can remove the cpcsm vector entirely
17:03 sofar I believe that's what celeron55 intended and nrzh's work would be almost entirely wholly used - the important parts are on the API side anyway
17:06 shivajiva okay, this is the time for some wisdom from celeron55 I guess :)
17:11 Andrey01 joined #minetest-hub
17:31 Andrey01 joined #minetest-hub
17:33 Andrey01 joined #minetest-hub
17:40 twoelk left #minetest-hub
17:41 Krock joined #minetest-hub
17:41 Krock hi tenplus1 (imaginary)
17:43 IhrFussel I got mad earlier because of the idea to "let clients decide which functions are allowed to run from the server side" WITHOUT thinking about how server owner would have to realize ESSENTIAL features they want to offload to clients
17:45 IhrFussel And the ONLY option I can think about it "run the exact same code server side in case the client doesn't want to run the server sent mod' which defats part of SSCM
17:45 Andrey01 joined #minetest-hub
17:46 IhrFussel defeats*
17:46 IhrFussel Cause there is nothing to offload to those "muh privacy" clients
17:47 Andrey01 joined #minetest-hub
17:50 Krock path calculating for objects could be offloaded if the server sent the LuaEntity's name
17:55 Andrey001 joined #minetest-hub
17:57 T4im prediction might be; but you probably continue to want server side physics
17:57 IhrFussel But what's the point of SSCM if the server needs to basically ask for each client function to execute? And then some clients will say 'sorry cannot run this callback'
17:58 T4im better prediction might be one point
17:58 Peppy mmm, much to read today
17:58 IhrFussel You cannot even create a mod/game that RELIABLY executes fine on every client
17:58 Peppy Android app is still broken, I guess most players will be gone before this endless debates about CSM/SSCSM ends
17:58 kilbith joined #minetest-hub
17:59 Krock yes sure. predict what happens in lags using SSCSM, and correct with the server-sent data
17:59 Krock Peppy: did you try the 5.0.0-dev build yet?
18:00 Peppy Krock, yes, but I don't use Android
18:00 Krock question was whether you tried a 5.0.0-dev build yet
18:00 Peppy yes I did
18:00 Krock okay
18:00 sofar IhrFussel: prediction and mods that coordinate server-client are only part of the story. Most mods will likely never send data to the server and just act based on instructions from the server (e.g. sounds/decorations/particles/messages/hud stuff)
18:02 Krock well, if some improvements aren't applied for iOS systems, you'd have to fix the stuff by proposing a PR - iOS is not officially supported
18:02 IhrFussel sofar, but nerzhul wants to implement a PER CALLBACK flag for SSCM which means one client allows only feature 1 and 3 and another only allows 2 and 4 while a third one only allows 1 2 and 3
18:02 Peppy Krock, let's face it many young players do use Android
18:03 sofar IhrFussel: is that documented/drafted somewhere?
18:03 Krock for me it feels like all kids use an outdated fork of the Android Minetest client
18:03 Krock except octactian. he uses the PC version :D
18:03 Peppy sure, but they have no alternative :(
18:03 IhrFussel sofar, https://github.com/minetest/minetest/issues/8002
18:04 Krock they do. there's a beta program for the official Minetest client
18:04 Krock they can try the newest clients there
18:05 Peppy they can't play their beloved 0.4.17 servers with that
18:05 sofar IhrFussel: seems a lot of work for little benefit, the other 2 points are much more critical
18:06 Krock it's called 5.0.0-dev for a reason. months after it's released the situation will surely become normal again
18:06 Krock a protocol break is quite special
18:06 Peppy krock : I understand the break was necessary
18:07 Krock It Sucks But We Cannot Help It ™
18:07 Peppy xD
18:09 Krock I spotted kilbith again! https://forum.minetest.net/viewtopic.php?p=338432#p338432
18:09 IhrFussel sofar, it would cause more work for server owners who want to implement something important as SSCM ... either the client who disallows that feature needs to disconnect or the server owners puts that much effort into it and copies the code (if possible) into a server side mod
18:10 IhrFussel But that is not the definition of "offload"
18:10 sofar offloading isn't the goal of csm
18:11 sofar it's a benefit, the goal is always better game experience
18:11 IhrFussel Huh? It was one of the major goals
18:11 Krock sofar: I think offloading is a very handy thing for CSM
18:11 sofar I'm not saying it isn't handy
18:11 Krock or at least "smoothing" the game
18:12 sofar I'm just saying the goal is better game experience, and offloading is a tool/path to some of it
18:12 garywhite joined #minetest-hub
18:12 Krock indeed
18:13 Peppy so, 0.5 will be out in months with SSCM, then server owners will struggle to have their mods running, then "hey, still anybody there ?"
18:13 sofar hehe, peppy is in a doom mood
18:13 sofar (palindrome)
18:13 IhrFussel Even if we don't talk about offloading...imagine a server relies HEAVILY on SSCM particles and now a client refuses to use that feature ... the server would need to execute the particles server side for that one client which is a rather absurd idea IMO
18:14 sofar it's not going to happen
18:14 Krock Peppy: it's the chicken and egg problem
18:14 Peppy i'd rather like being more positive :/
18:14 sofar no server owner would agree to sending thousands of individual rain drops to each client
18:14 Krock "who's going to switch first"
18:15 sofar if the client refuses, oh well, he doesn't get the experience and his game play may be significantly hindered
18:15 Krock I don't think Android allows a dual-installation or Minetest, right?
18:15 sofar end of story
18:15 Krock s/of/of
18:15 Krock s/or/of/
18:16 garywhite Krock: Nope, Android would consider the second installation as an update to the first
18:16 IhrFussel I'm sure there are things that cannot be simply dropped for "inferior experience" ... I said earlier MT is not a website where you can just disable optional JS and over 90% of them still work fine
18:16 sofar just disconnect the player
18:16 IhrFussel SSCM will be part of a game logic which requires every tid bit to run correctly
18:17 sofar "client disconnected: no SSCM ping received in 30 seconds"
18:17 sofar one CM mod, 5 lines of code
18:17 sofar I'll write it for you, too
18:18 Peppy being unable to install both Android versions of minetest sounds like a big concern to me
18:18 IhrFussel Yes but what I really hope for is a flag that tells the server if all/some/none flags are enabled from the client...so that server owners can drop those right away
18:18 Peppy some servers won't ever upgrade
18:19 garywhite Peppy: I think the only way would be to change the package name, and even then idk if it'd work
18:19 IhrFussel Or another idea -> local flags = player:get_client_flags() if flags.feature1 and flags.feature3 then --run end
18:19 Peppy "minetest legacy"
18:19 sofar IhrFussel: that's why I think it's bogus to begin with to add those 3 extra flags
18:19 sofar imho we shouldn't ever do that.
18:20 sofar the server owners will just change the code to mark everything required
18:20 sofar or kick players who don't enable optional stuff
18:20 sofar end result is that it's meaningless to even set those flags
18:21 sofar it's like saying, we won't arrest people for X anymore. But we just wrote a new law banning X
18:21 sofar don't do that :)
18:22 sofar there's something to be said for warning players before they connect, of course, unless every server uses sscm, at that point it no longer matters
18:23 IhrFussel I guess nerzhul doesn't plan to add a way to tell server owners that only "some" flags are enabled and to the server it just looks like "SSCM enabled"
18:23 Peppy Survey : who will first be killed by upgrade , server owners or players ? :D
18:24 sofar i think I've made my point, there's too many ways around it anyway
18:24 sofar you could potentially sandbox all sscm on the client and let it not affect anything in your game
18:25 sofar everything a stub
18:25 sofar except for server-client comms
18:25 IhrFussel Latest when the server expects the client to change some values or send something and gets no answer it will know that the feature is not enabled
18:25 IhrFussel Unless you also add "dummy" code that lies to the servers
18:28 sofar end result: flagging is mostly meaningless, don't even bother
18:44 scr267 joined #minetest-hub
19:23 Fixer Krock: how?
19:23 Fixer Krock: how you know it was kilbith? Also, he is banned?
19:23 Fixer Krock: or what was the problem?
19:23 Krock Fixer: see GitHub link. They also left a while ago
19:23 Krock returned now :D
19:23 Fixer Krock: they?
19:24 Krock he.
19:24 Krock using "they" when I'm not sure
19:33 pauloue joined #minetest-hub
19:36 bobr joined #minetest-hub
19:36 bobr joined #minetest-hub
19:45 Gael-de-Sailly joined #minetest-hub
20:02 nerzhul IhrFussel i think the best idea is suggested by Ezhh: restriction levels client side, as a contract between client & server, like the DNT in firefox, i think it's a good idea, client set a restriction level, and mods sent from server can enable some parts only if restrictions are sufficiently opened
20:02 nerzhul required, optional and maybe a 3rd level
20:02 nerzhul and as i said i want to work on mod transfer on a 5.1 because it can be a long time story
20:03 nerzhul because there is not only mod transport, but also verification, restrictions, verifications etcd...
20:12 VanessaE nerzhul: only if you overengineer it....
20:50 DI3HARD139-m96 joined #minetest-hub
20:55 rubenwardy what specific features would clients want to disable?
20:55 rubenwardy particles seems to be more of a graphics thing to me
21:00 DI3HARD139-m joined #minetest-hub
21:01 DI3HARD139-m joined #minetest-hub
21:03 DI3HARD139-m joined #minetest-hub
21:05 DI3HARD139-m joined #minetest-hub
21:06 nerzhul block chain ? :D
21:09 Fixer blockchain is dead
21:09 DI3HARD139-m joined #minetest-hub
21:13 DI3HARD139-m joined #minetest-hub
21:21 sofar nerzhul: there's no reason to not use the client media protocol to relay signed csms. Just before loading it in lua, you would validate the embedded signature. All that's needed is to modify e.g. https://github.com/minetest/minetest/blob/b298b0339c79db7f5b3873e73ff9ea0130f05a8a/src/server/mods.cpp#L106 to include "csm" and allow ".lua" as extensions
21:22 sofar my bet is that the code really just needs a few tweaks and you could already start transmitting and caching csms sent in no time
21:23 sofar retrieving from the cache and loading in lua is probably slightly more complex
23:30 pauloue joined #minetest-hub

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