00:59 < bridge> nim + zig = nig, call it nig :justatest: 01:37 < bridge> when did all the demo buttons become icons with no text and no tool tips? 01:37 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1260379355453063198/image.png?ex=668f1b3f&is=668dc9bf&hm=42ba9324912b42e631897ed6299be8f201876b78ec0d3c5743c87379fc8aee94& 01:37 < bridge> this is much worse 01:38 < bridge> git blame is your friend 01:38 < bridge> I suggest using `tig blame` 01:38 < bridge> it has nicer UI 01:41 < bridge> https://github.com/ddnet/ddnet/pull/7311 01:44 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1260381115647266816/273316435-87a44d15-d242-4cc4-988c-0cdb024cb3a5.png?ex=668f1ce2&is=668dcb62&hm=4c83e3f55fce6d637a54d256f28875f845ee83b69d43428c1a03650b9333ea9b& 01:44 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1260381116066566184/273316422-4b41328f-7d51-48d8-9a51-c79cc12beb0a.png?ex=668f1ce2&is=668dcb62&hm=bed13a08e58b67013b7cb4cd7aa3bcc3e2ee6520870bba2ed7d5b7c45914c79e& 01:45 < bridge> Is it only me who prefers the full text much more than the labels? 01:46 < bridge> There's plenty of space I don't see any reason to obfuscate what the buttons do 01:47 < bridge> I usually just use the github web ui blame or vscode extension 01:48 < bridge> What is DDNET\_IP supposed to be? The ip of the main ddnet server that can bypass all servers firewalls? Maybe it can use a comment so nobody has to guess 01:50 < bridge> yes @ main ddnet server 01:50 < bridge> I find the web UI to be too slow 01:51 < bridge> Idk what features tig blame has but browsing files from the console is not very comfortable for me 01:52 < bridge> if I am doing a lot of digging I just open vscode 01:54 < bridge> interesting 01:57 < bridge> Ah nice thanks 01:58 < bridge> I basically never used git blame anymore these days. 01:59 < bridge> My editor has blame hints so I see it when I look at the code. But if I am actively searching for a commit I always use `git log -S` 01:59 < bridge> chiller 02:00 < bridge> what was that one site u had 02:00 < bridge> that was a bootleg discord 02:00 < bridge> git log -SRender 02:00 < bridge> https://chat.zillyhuhn.com 02:01 < ws-client> Uwu 02:57 < bridge> hm ill take a look 02:57 < bridge> thank u 03:20 < bridge> Can anyone help me?, I'm trying to make a websocket connection with python at this link "wss://chat.modd.io/socket.io/?EIO=4&transport=websocket" but it gives me an error, can anyone help me? 03:55 < bridge> https://dontasktoask.com/ 03:56 < bridge> give error & code 04:16 < bridge> I didn't understand -_- 04:18 < bridge> @toku.kunavara: give error & code 04:19 < bridge> a 04:19 < bridge> ok 04:20 < bridge> ```py 04:20 < bridge> import websocket 04:20 < bridge> import ssl 04:20 < bridge> 04:20 < bridge> def on_message(ws, message): 04:20 < bridge> print(f"Mensagem recebida: {message}") 04:20 < bridge> 04:20 < bridge> def on_error(ws, error): 04:20 < bridge> print(f"Erro: {error}") 04:20 < bridge> 04:20 < bridge> def on_close(ws): 04:20 < bridge> print("Conexão WebSocket fechada") 04:20 < bridge> 04:20 < bridge> def on_open(ws): 04:20 < bridge> print("Conexão WebSocket aberta") 04:20 < bridge> 04:20 < bridge> if __name__ == "__main__": 04:20 < bridge> websocket.enableTrace(True) 04:20 < bridge> ws = websocket.WebSocketApp("wss://chat.modd.io/socket.io/?EIO=4&transport=websocket", 04:20 < bridge> on_message=on_message, 04:20 < bridge> on_error=on_error, 04:20 < bridge> on_close=on_close) 04:20 < bridge> ws.on_open = on_open 04:20 < bridge> ws.run_forever(sslopt={"cert_reqs": ssl.CERT_NONE}) 04:20 < bridge> ``` 04:20 < bridge> @toku.kunavara: don‘t announce you have an error and wait for a response. Give enough details so someone can solve your error. 04:20 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1260420318170976287/aaaa.txt?ex=668f4165&is=668defe5&hm=e8240a9d8bfc1c5fe90bcf55a9203172d69e42b3ab8c27306f3e363249cde152& 04:21 < bridge> Don’t ask „can anyone help me?“ Just ask your question. People might help if they want to. 04:22 < bridge> Where is that code coming from? Did you follow some tutorial or make it up? 04:24 < bridge> What is chat.modd.io 04:29 < bridge> I made the code, I was looking for ways to get chat messages from a game, it's a browser game, its name is braains.io. If I use an extension to connect to this link it works, but if I try to do it with websocket it doesn't work 04:31 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1260423113989161041/image.png?ex=668f43ff&is=668df27f&hm=dd5136e89a667b316ce079cd51a9e219bd4c630ddee145f57986a3d409e8201c& 04:31 < bridge> the name of the extension is simple web socket client 04:32 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1260423343962587146/image.png?ex=668f4436&is=668df2b6&hm=ccff6b28948d17141b4a5e41f508fc4d0e89d41db0652d51eb022036882412b2& 04:32 < bridge> but if I try to connect using python it won't work 04:32 < bridge> using websocket 04:34 < bridge> I'm sorry if you don't understand, I'm Brazilian and I'm using Google Translate to talk :,(, I've been trying to solve this problem for 3 days and I can't 05:47 < bridge> what data structure is ddnet network updates sent as (like whatever is sent each tick)? How big is it in bytes 05:48 < bridge> for a project i'm working on i'm currently sending data over network as dictionaries containing info for each field of an entity i send. I feel it'll be pretty bandwidth heavy so i'm wondering what better practices there are 06:44 < bridge> it's sorta complicated. They are sent as snapshot diffs where only the things that changes from the previous tick are sent, plus every once in a while a full snapshot is sent. (Atleast from what I understand) 06:44 < bridge> it's sorta complicated. They are sent as snapshot diffs where only the things that changed from the previous tick are sent, plus every once in a while a full snapshot is sent. (Atleast from what I understand) 06:45 < bridge> there is no fixed size but there is a limit 06:45 < bridge> there is no fixed size but there is a size limit 06:47 < bridge> basically the same idea as video encoding where you have I-Frames and P-Frames 06:47 < bridge> but not B-Frames 06:51 < bridge> ah okay 06:51 < bridge> @totar do you know how often a full snapshot is sent? 06:52 < bridge> not off the top of my head 06:58 < bridge> almost never apart from the first 06:58 < bridge> iirc the client has a CRC check, if it thinks it can not reconstruct a full snapshot it just request one from the server 06:59 < bridge> unless there are some heavy packet drops or interference, this would ideally never happen 06:59 < bridge> ah okay 06:59 < bridge> not even if a single packet is dropped? 06:59 < bridge> yes actually he's right 06:59 < bridge> are multiple previous deltas sent every tick 06:59 < bridge> it uses the last acknlowledges snapshot to create the diff 06:59 < bridge> no, client holds several at a time 06:59 < bridge> it uses the last acknlowledged snapshot to create the diff 07:00 < bridge> my game is in 10 tps so significantly less than tw, but it will definitely send more than 5x the data than a ddnet server would send 07:00 < bridge> each packet has a parent packet id, the client will delta decode based on the correct one 07:00 < bridge> but if one tick is dropped from server, doesn't the client need to ask for the full snapshot again 07:00 < bridge> the client tells the server which snapshots it has 07:00 < bridge> there is a acknoledgement system 07:01 < bridge> client tells server that it has a snapshot in full (either received as-is or decoded) 07:01 < bridge> server will only delta encode packets based on the client's acknoledged ones 07:01 < bridge> so server also holds multiple snapshots just in case clients acknoledges a older one 07:02 < bridge> ah okay 07:02 < bridge> do you know how many snapshots the client stores? 07:02 < bridge> a tad complicated 😮‍💨 07:02 < bridge> not on top of my head (for teeworlds) 07:03 < bridge> just as an example what are some things that would be in a full snapshot that are not in a delta that saves a lot of data? 07:03 < bridge> i know how it works, i don't think i know any detail. only because i coded a similar one for our game 07:04 < bridge> in teeworlds at least, delta only works because everything is integer 07:04 < bridge> there is a 7-bit variable length integer packing 07:04 < bridge> a value takes 1 to 5 bytes depend on how large the number is 07:05 < bridge> so if a player position is 1500 last tick and 1501 this tick, the delta packet encodes a 1 which is 1 byte 07:05 < bridge> so basically, for teeworlds at least, everything is sent in the delta 07:05 < bridge> literally, everything 07:06 < bridge> ah okay i see 07:06 < bridge> afaik the things it can delta are the creation of entities and their positition/properties updating. So in theory the server only sends info when an entity is created and when it moves 07:06 < bridge> cuz they are all integers, delta is just the difference between the value, the save in size if inhereted from encoding 07:06 < bridge> do they tho? 07:06 < bridge> pretty sure? 07:06 < bridge> let me see 07:06 < bridge> ig that means a lot of server calculation between the client's last snapshot and the current state on the server then 07:07 < bridge> i thought teeworlds just send everything 07:07 < bridge> wdym 07:07 < bridge> cuz skipping snap just means it disappeared 07:07 < bridge> if it sent everything why bother with delta 07:07 < bridge> it sent everything in delta encoding, but it doesn't skip entities 07:07 < bridge> if it doesn't change it sends a bunch of zero 07:07 < bridge> no I think it actually doesn't send zeros 07:08 < bridge> but in order to not send zeros, you need extra informatio for appearing and disappearing, i don't remember teeworlds has a system for that 07:08 < bridge> but in order to not send zeros, you need extra information for appearing and disappearing, i don't remember teeworlds has a system for that 07:08 < bridge> I think it has that 07:08 < bridge> unless I'm stupid 07:09 < bridge> maybe i remembered it wrong 07:09 < bridge> heinrich definitely know 07:09 < bridge> either way a bunch of zero if not that bad 07:09 < bridge> cuz teeworlds also has huffman 07:09 < bridge> maybe i will just send new position every tick since i'm calculating in floats xd 07:09 < bridge> a zero is 1 bit 07:09 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1260462988503154709/image.png?ex=668f6922&is=668e17a2&hm=b2b56be677b4bc256be8023e4c7a7e4eb427e016a8b1edc765ad94020e471d0f& 07:10 < bridge> I believe CSnapshotBuilder::NewItem handles adding items 07:10 < bridge> and deletion just happens? 07:10 < bridge> new item is for creating a snapshot, this part seems to be the part where it prepare the base snapshot 07:11 < bridge> ah wait 07:11 < bridge> I copied the wrong thing 07:11 < bridge> it check whether a snap exists, if it doesn't it is deleted. so for a snap to stay there have to be one 07:11 < bridge> even if it didn't change 07:11 < bridge> no actually 07:12 < bridge> I think I was right 07:12 < bridge> new item is for entities 07:12 < bridge> i don't think snapitem works like entities 07:12 < bridge> how are different actions labeled in the bits sent over network? (differentiating between movement data and aim data for example) 07:13 < bridge> just pure order. 07:13 < bridge> ah ok 07:13 < bridge> I mean entity like the general meaning 07:13 < bridge> I guess it's called a snapitem 07:14 < bridge> anyway, you need to send extra data if you want to skip a snap id just to indicate it didn't change 07:14 < bridge> or at least a special case for handling not changing 07:14 < bridge> either way is not more worth it than just packing the diff 07:15 < bridge> you just don't include the snapitem id in your snapshot data to indicate it didn't change? 07:15 < bridge> no 07:15 < bridge> not including a snapid means it disappeared 07:15 < bridge> snapshot is a collection of individual deltas 07:15 < bridge> I thought there was delete message 07:15 < bridge> i don't remember there being one 07:16 < bridge> but i am too lazy to check 07:17 < bridge> morning 07:17 < bridge> norming 07:18 < bridge> found this variable that does nothing 07:18 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1260465144421416971/image.png?ex=668f6b24&is=668e19a4&hm=311688c3a11d8563e816505cf033784d60bb477db21e1466502a2cd92865a32a& 07:18 < bridge> :kek: 07:19 < bridge> probably a too scared to touch situation 07:19 < bridge> maybe it's required to keep struct size 07:19 < bridge> better not touch it 07:19 < bridge> can't believe i'm reading teeworlds' code again 07:20 < bridge> well you're in the teeworlds code channel 07:21 < bridge> ok maybe you're right 07:21 < bridge> i was about to say i think you are right cuz i did see they packed deleted keys 07:22 < bridge> oh lol 07:22 < bridge> but i don't see the skipping part yes 07:22 < bridge> but i don't see the skipping part yet 07:22 < bridge> yeah 07:22 < bridge> where do deleted keys get packed 07:23 < bridge> this is.. mmm 07:23 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1260466378633248830/image.png?ex=668f6c4a&is=668e1aca&hm=a92e586f42481fca3bd5d2a2d48f55c015c11943a53283960c41d141242984ae& 07:23 < bridge> https://github.com/teeworlds/teeworlds/blob/a1911c8f7d8458fb4076ef8e7651e8ef5e91ab3e/src/engine/shared/snapshot.cpp#L215 07:24 < bridge> ah true 07:24 < bridge> honestly 07:24 < bridge> if teeworlds really didn't skip but still has a deletion system 07:24 < bridge> ok yeah I think it does send deletion messages then? 07:24 < bridge> we might just... 07:25 < bridge> cut the bandwidth more without breaking anything 07:25 < bridge> assuming it really didn't skip anything 07:25 < bridge> I think it does skip 07:25 < bridge> because the deleted items will just not be in the "pTo" snapshot 07:25 < bridge> so they don't get iterated over 07:26 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1260467120588718090/image.png?ex=668f6cfb&is=668e1b7b&hm=527b2fcd8c5c8410d3804d8445280f4379d01a8c5501f488369c43f710227d06& 07:26 < bridge> anything that was deleted won't be in the "pTo" 07:26 < bridge> + it already sent the client all the stuff which was deleted so the client is expected to delete those things first 07:26 < bridge> i mean well obvious, but what about non-deleted stuff that also skipped due to "not changed" 07:27 < bridge> i mean well obviously, but what about non-deleted stuff that also skipped due to "not changed" 07:27 < bridge> wdym 07:27 < bridge> ah 07:27 < bridge> like didn't move? 07:27 < bridge> yeah idk 07:27 < bridge> yet, aren't we talking about skipping snaps for "not changed" 07:27 < bridge> my entire argument is that teeworlds doesn't skip anything so it didn't need a deletion manager 07:27 < bridge> I think here this ``if`` will be false if they are "not changed" 07:27 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1260467491977695332/image.png?ex=668f6d54&is=668e1bd4&hm=236553025c9816e33d776d4f222adeba5d902c9ea36b1aebbfc7660c37da4925& 07:27 < bridge> ah 07:28 < bridge> ye, problem solved 07:28 < bridge> you are right 07:29 < bridge> and.... louis is gone 07:30 < bridge> lol 07:30 < bridge> tbh it's not a super complicated system, but the teeworlds code is certainly not the most elegant way to write it 07:31 < bridge> i can garrentee you our networking packer is jankier than teeworlds' 07:31 < bridge> I can't figure out why this is nessecary 07:31 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1260468450929672252/image.png?ex=668f6e39&is=668e1cb9&hm=b04d932a664a5d29080762166e30dbb325a845e4a1375da588aa36f10b039719& 07:31 < bridge> "it helps the cache" 07:32 < bridge> I mean the O(n^n) part 07:32 < bridge> hm 07:32 < bridge> 😹 07:32 < bridge> the gamedev is back! 07:33 < bridge> why is there an item hasing system in the first place 07:33 < bridge> why is there an item hashing system in the first place 07:33 < bridge> ye, idk 07:33 < bridge> bro. firefox's find is broken 07:33 < bridge> so how much money do you think this optimization saves in terms of server costs 07:33 < bridge> press . key 07:33 < bridge> it opens vscode in browser 07:34 < bridge> true 07:34 < bridge> firefox literally modified the page when i tried to select and search 07:34 < bridge> it's basically required 07:34 < bridge> it fixed itself again, didn't manage to grab a screenshot 07:34 < bridge> lol 07:35 < bridge> search on github is annoying 07:35 < bridge> I have that same issue with firefox 07:35 < bridge> I know what you mean 07:35 < bridge> ye, what tater said. it is actually quite essential just to make sure it is playable everywhere, not just server costs 07:35 < bridge> sometimes it opens firefox search, other times it opens github search 07:35 < bridge> there are other ways than snapshot / delta encoding tho 07:36 < bridge> you can use the lazy method 07:36 < bridge> if you are sending json. a basic gzip is all you need. 07:36 < bridge> for 10tps 07:36 < bridge> it's not too bad 07:36 < bridge> yeah im sending a dictionary that gets serialized into json 07:36 < bridge> are you doing snapshot prediction/interpolation and stuff which really really requires a snapshot? 07:37 < bridge> every tick is only position information for other players (usually) and for enemies it's position info every time the enemy starts a new move, and the move name (a short string, i could use a local int-to-move-string map though) 07:38 < bridge> are you doing it in browser 07:38 < bridge> nah 07:38 < bridge> do you have a reliable/unreliable channel thing? 07:38 < bridge> like udp/reliable udp 07:38 < bridge> also chat messages ig. not sure how much of an effect it has 07:38 < bridge> or just udp and tcp 07:38 < bridge> yeah i'm using godot, each tick sends both reliable (if there is data to send) and unreliable data 07:38 < bridge> godot oh 07:39 < bridge> wait, why are you sending json with godot lol 07:39 < bridge> godot has binary packers iirc? 07:40 < bridge> for location really, just push it via a unreliable and forget them. 07:40 < bridge> does it? xd 07:40 < bridge> let me check 07:40 < bridge> also you can bitpack values if you really just want locations 07:40 < bridge> maybe i'm on the newbie side of the internet but lots of the stuff i read just recommended using rpc and serializing to json 07:41 < bridge> godot tutorial on networking are all pretty basic 07:41 < bridge> yes, in terms of players i think the data can be pretty easily optimized if need be 07:42 < bridge> the only place i'm worried abt bandwidth is in main lobbies and whatnot, like for example there are hundreds of players on one server 07:42 < bridge> the only place i'm worried abt bandwidth is in main lobbies and whatnot, like for example there are hundreds of players on one world 07:42 < bridge> do you see everyone on screen at once 07:43 < bridge> https://docs.godotengine.org/en/stable/classes/class_packedbytearray.html 07:43 < bridge> in the worst case scenario probably 07:43 < bridge> but then since gameplay precision doesn't matter as much in a lobby updates could just be sent at a lower rate 07:43 < bridge> you can just `encode_s16` then `decode_s16` on the other size 07:43 < bridge> you can just `encode_s16` then `decode_s16` on the other side 07:43 < bridge> choose you own byte length 07:43 < bridge> it is pretty handy 07:43 < bridge> ah okay 07:43 < bridge> yes then that's pretty nice 07:44 < bridge> godot will send PackedByteArray no problem, it handles that for you 07:44 < bridge> how is it possible that we can have an O(n^n) algorithm in the code 07:44 < bridge> for now i use json'd dictionaries to find out what i need to send and what i dont but after that i guess packing is best idea 07:44 < bridge> how is it possible that we can have an O(n^n) algorithm in the code and it doesn't destroy the game 07:45 < bridge> it must be worst case scenario? 07:45 < bridge> given how the comment says O(n)...O(n^n) 07:45 < bridge> O(n^n) is really really bad tho 07:45 < bridge> if you see 100 players in a densed area. we can assume several things: 07:45 < bridge> 1. camera would be pretty zoomed out and does not require a lot of precision. 07:45 < bridge> 2. players will have close proximity to each other 07:45 < bridge> I think it meant to say O(n*n) 07:45 < bridge> then it is a perfect oppertunity to do delta encoding between players actually 07:46 < bridge> for example, you can divide player locations by 10, int them, then encode(p1_pos), encode(p2_pos - p1_pos) and so on 07:46 < bridge> you can save a lot of bandwidth doing that 07:47 < bridge> oh, i just assumed it meant that really. i didn't realize what n^n actually means. so i was quite confused about what you were talking about oof 07:48 < bridge> :justatest: 07:48 < bridge> O(n^n) just doesn't exist in my brain i guess 07:49 < bridge> oh you do need a 7bit packing algorithm to make that work tho 07:49 < bridge> n=16 would be 18,446,744,000,000,000,000 iterations lol 07:49 < bridge> so definitely not that 07:50 < bridge> what does this even mean // O(1) .. O(n) 07:50 < bridge> best/worst case? 07:50 < bridge> ok I'm finding who wrote these 07:51 < bridge> 07:51 < bridge> good luck on the quest lol 07:51 < bridge> he's finding them irl 07:52 < bridge> don't forget to share photos kek 07:53 < bridge> ok they were there forever 07:53 < bridge> 14 years old 07:54 < bridge> thanks magnus 07:54 < bridge> do not doubt magnus 07:54 < bridge> you're right 07:56 < bridge> there's so many comments like "// TODO: this is bad, change it later" and now it can't be changed without breaking protocol or physics 07:58 < bridge> https://github.com/Tjoosten/Funny-code-comments/blob/master/Temporary-my-ass.md 07:59 < bridge> lol 08:01 < bridge> This one is literally me against this comment https://github.com/ddnet/ddnet/blob/fa9f7b46cdf5a45be5985217f1a67c5e1553c52c/src/game/collision.cpp#L329 08:01 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1260475969865650237/image.png?ex=668f7539&is=668e23b9&hm=cad5890a622c950c5d92cd40dcf9b152f16b2eeb7d616d750b931f5bb991cf18& 08:01 < bridge> This one is literally me against this comment 08:01 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1260475969865650237/image.png?ex=668f7539&is=668e23b9&hm=cad5890a622c950c5d92cd40dcf9b152f16b2eeb7d616d750b931f5bb991cf18& 08:01 < bridge> I spent like 4 days trying to optimize it 08:02 < bridge> but preserve the original functionality 08:03 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1260476535219949619/image.png?ex=668f75c0&is=668e2440&hm=bf58b7c9cb0a37bea917d9fce64ec7c50ce72e918a664516a94e5c8461d53b0f& 08:03 < bridge> https://cdn.discordapp.com/attachments/293493549758939136/1260476535442374677/image.png?ex=668f75c0&is=668e2440&hm=2eba2fe53cd00c0303659b20b0fa59237bab3e35a09b623ecf75bf572ffaf401& 09:33 < bridge> unchanged items in the snap are not sent 09:34 < bridge> @totar is correct. you delete items by including their ID in the deleted items section of a elta 09:34 < bridge> :owo: i knew you would know 09:35 < bridge> imma name one of my character elta, sounds cool 09:36 < bridge> Im yoinking this name 09:36 < bridge> yoinkaway 09:47 < bridge> https://news.ycombinator.com/item?id=40918052 09:50 < bridge> N guaranteed low? 09:50 < bridge> It's mislabeled 09:51 < bridge> That explains it 10:01 < bridge> same as mozilla firefox btw 10:03 < bridge> I guess that usage might be less defensible. but e.g. https://addons.mozilla.org/ has special permissions, so that even ad blockers don't work there 10:55 < bridge> same 12:11 < bridge> <0xdeen> @learath2 looking better now 12:11 < bridge> <0xdeen> https://cdn.discordapp.com/attachments/293493549758939136/1260538987395678269/message.txt?ex=668fafea&is=668e5e6a&hm=4f215008dcc8edc84ec46333665f38cec9999d81646738d46841e62eceb878ea& 12:12 < bridge> <0xdeen> CAntibot::Update is still at 22%, but at least not 50 12:14 < bridge> <0xdeen> One thing that stands out is that every str_copy calls str_utf8_fix_truncation, which is kind of useless for this kind of string 12:14 < bridge> <0xdeen> I'll replace it with a mem_copy 13:13 < bridge> I want to reorganize some internals for the unique game servers and came to a general problem I wanted to ask for help here. How do I publish a directory with subdirectories and files on the internet such that others can sync it? I basically want rsync but accessible to the public. My first instinct would have been to use HTTP but I don't know any tools that can sync a folder hierarchy from HTTP and I don't know if it's even possible because HTTP d 13:28 < bridge> <0xdeen> Put it in a git repo and put it on github: https://github.com/ddnet/ddnet-maps 13:28 < bridge> <0xdeen> You can also create your own git server 13:29 < bridge> <0xdeen> You can also run rsync as a daemon to allow anyone to sync from it using rsync 13:29 < bridge> <0xdeen> https://wiki.archlinux.org/title/rsync#As_a_daemon 13:30 < bridge> <0xdeen> But with github you'll probably reach 100x as many people as with an rsync:// server 13:31 < bridge> <0xdeen> Alternatively SFTP, but you'll have to be careful that the user you generate doesn't have permissions to do anything bad: https://wiki.archlinux.org/title/SCP_and_SFTP 13:32 < bridge> <0xdeen> FTP is really out of date, I would not have used that for anything 15 years ago 😄 13:33 < bridge> <0xdeen> For HTTP there is also WebDAV, but I don't have experience with using that 13:37 < bridge> rclone seems to be able to sync from http directory listings: https://rclone.org/http/ 13:42 < bridge> I need to programmatically update files and sync files that's why I think it would be a misuse of git 13:42 < bridge> Rclone seems perfect 13:42 < bridge> For a much smaller project almost a decade back, I used unison to sync files. It can work over plain tcp too 13:43 < bridge> But I guess it's more appropriate when you need to be able to modify stuff on both sides 13:43 < bridge> Yeah for me it's only one-way 13:43 < bridge> What is the issue with git? 13:44 < bridge> You can make commits programatically just fine 13:44 < bridge> I think git pull man page specifically mentions you should not call it from a script 13:45 < bridge> I dohbt that 13:45 < bridge> I'm sure you would get it to work though it's just not what git was made for 13:45 < bridge> I don't see such a notice 13:45 < bridge> I don't see such a notice in the man page 13:45 < bridge> I doubt that 13:45 < bridge> I've been automatically pulling changes for 10 years or so from git repositories 13:45 < bridge> it seems to work fine 13:46 < bridge> <0xdeen> ddnet-maps has been on git/github for 10 years too 13:47 < bridge> (a downside of git is that you get all the history, too) 13:48 < bridge> Which you can get around by cloning shallow, no? 13:50 < bridge> yes, but that in addition to pulls is bad for performance, according to github 13:50 < bridge> they asked the cargo team to not do shallow clones as that was hard on the github infra 13:51 < bridge> yea it's not in the man page, that was wrong 13:51 < bridge> Hm, that's interesting. So the shallowing isn't something trivial 13:52 < bridge> I would assume pulls in a shallow repo would just pull in deltas like normal, apply them, then prune 13:52 < bridge> i think git is the right move tbh 13:52 < bridge> not different from automating e.g. cmake calls 13:53 < bridge> Or if you really don't need the versioning capabilities of git at all. I guess ftps or rsync are what I would go for 13:53 < bridge> i'd still use git bcs nowadays the likelihood of a system having git is prob higher than rsync 😁 + just standard 13:54 < bridge> Rclone parsing a listing meant for endusers just sounds more jank than those to me 😄 13:54 < bridge> If I use git is there anything I should be aware of? My first attempt would be just to do 13:54 < bridge> ``` 13:54 < bridge> # on main server 13:54 < bridge> git commit -m "Added map $blabliblub" 13:54 < bridge> # on locale in a systemd timer 13:54 < bridge> git pull 13:54 < bridge> ``` 13:54 < bridge> If I use git is there anything I should be aware of? My first attempt would be just to do 13:54 < bridge> ``` 13:54 < bridge> # on main server 13:54 < bridge> git commit -m "Added map $mapname" 13:54 < bridge> # on locale in a systemd timer 13:55 < bridge> git pull 13:55 < bridge> ``` 13:55 < bridge> push but ya 13:55 < bridge> how do you want to pull? via ssh? that should work out of the box 13:56 < bridge> I think http because automated pulls won't have an ssh key 13:57 < bridge> I might add a `git reset --hard` there. We still get a couple pull issues on ddnet because of that 13:57 < bridge> why? 13:57 < bridge> then you should also start an https server (if you haven't already) 13:57 < bridge> maybe just fetch and then reset --hard to avoid conflicts and just overwrite stuff 13:58 < bridge> Idk, never diagnosed it further. Somehow sometimes local servers will end up with changes 13:58 < bridge> but there should be no conflicts because nobody writes there 13:58 < bridge> then it should just work™ 13:59 < bridge> spooky 13:59 < bridge> probably because of admin actions ^^ 14:03 < bridge> Which would make sense but it happened on `bans-global.cfg`, it should never be manually edited 😄 14:03 < bridge> hmmmmm 14:06 < bridge> i think im gonna try out hyprland 14:06 < bridge> friends are saying it works with fractional scaling 14:07 < bridge> I've been using 1.5x on normal xorg on my laptop for a bit now, and I haven't had many issues 14:08 < bridge> too many variables 14:08 < bridge> with how wayland works that behavior now depends entirely on the compositor 14:08 < bridge> no more xrandr to fall back on 14:09 < bridge>