DDraceNetwork - developer

ReiTW

yes but that's useless asf

not really, their main route is https://ddstats.tw/player

00:10

thats actual web page so

00:11

a better option would be something like https://api.ddstats.tw/player but doesnt matter idk

ChillerDragon

[ddnet/ddnet] Issue closed: #8142 Maintainers personal opinions should not block merges

sjrc6

[ddnet/ddnet] Pull request closed: #8969 Enable cl_sub_tick_aiming by default

sjrc6

[ddnet/ddnet] Pull request opened: #9124 Apply exponential easing to smooth camera

I don't like linear interpolation, I think it always looks bad. Comparison: (Sorry for the crunchy bitrate, github is limiting to 10mb now) https://github.com/user-attachments/assets/75245d73-684a-4183-9c47-94000dc5195d

Checklist

[x] Tested the change ingame
[x] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Written a unit test (especially base/) or added coverage to integration test
[ ]...

Jupstar ✪

does someone have an API to fetch the last player skin of a given player name? gerdoe sent me the API from furo: https://ddstats.tw/player/json?player=n9 but it's way too verbose, i'd dos the server xD

last or most used?

Matodor

last or most used?

last rn

zhn

last rn

rn no

I'll add this feature to my API in a few days

1

github-merge-queue[bot]

[ddnet:master] 2 new commits

b6bc5d1 apply exponential easing to smooth camera - sjrc6 10df82a Merge pull request #9124 from sjrc6/pr_smooth_expo_ease - def-

07:00

def-

[ddnet/ddnet] Pull request closed: #9124 Apply exponential easing to smooth camera

@Scrumplex poggers2

07:08

Prolly one of the best picks

meloƞ

@Scrumplex poggers2

what is this?

Scrumplex is nominated to be part of the nix steering committee!!!

07:16

https://github.com/NixOS/SC-election-2024

GitHub - NixOS/SC-election-2024: 2024 Election for the Steering Com...

2024 Election for the Steering Committee. Contribute to NixOS/SC-election-2024 development by creating an account on GitHub.

gl hf

https://blog.cloudflare.com/how-cloudflare-auto-mitigated-world-record-3-8-tbps-ddos-attack/

How Cloudflare auto-mitigated a world record 3.8 Tbps DDoS attack

Cloudflare’s DDoS defenses have automatically and successfully detected and mitigated a 3.8 terabit per second DDoS attack — the largest attack on record — as part of a month-long campaign of over a hundred hyper-volumetric L3/4 DDoS attacks.

4

07:18

im pro gentoo

im anti gentoo

@Jupstar ✪

09:08

https://query.rs/

Query.rs

A search engine for Rust

Ryozuki

@Jupstar ✪

Looks fair

Are we able to specify a path to the autoexec_server.cfg file ?

09:24

Or any config file in general

Serbia when Rust became popular:

Replying to @Ryozuki https://query.rs/

zhn

1. clstartinfo clchangeinfo are self replicating 2. svextraprojectile is redudant(?) 3. input is a system message? why? 4. ready/conready are self replicating 5. mapdata contains the same crc of map, why? 6. and another questions that im not able to recall

1. that's by design, I think 2. not sure how it's redundant, but it's also not used anymore, so whatever 3. because the engine deals with input timing etc 4. I don't understand what you mean by self-replicating 5. probably to make sure that we get map data for the correct map

09:26

"break the protocol" isn't a goal, it can be a means to achieve a goal

09:27

what you posted looks more like a couple of questions about the current protocol though. I'd be worried if we decided to break the protocol because someone had questions on it

Ryozuki

map format

can you write up your vision for a map format?

meloƞ

Are we able to specify a path to the autoexec_server.cfg file ?

Or let me reword that, can the .cfg file handle absolute paths ? (eg exec ~/.config/ddnetserver/myconfig.cfg)

its rewind time

Learath2

Btw did y’all see Valve managed to break through the wayland stalemate that has been holding back that vsync proposal?

no. link?

heinrich5991

no. link?

https://www.phoronix.com/news/Mesa-frog-fifo-v1-MR https://www.supergoodcode.com/My-Wayland-Your-Wayland-Our-Wayland/

Frog-FIFO-V1 Aims To Address Mesa's "Fundamentally Broken" Wayland ...

Joshua Ashton of Valve's Linux graphics team has opened a Mesa merge request to support a proposed 'frog-fifo-v1' protocol for Wayland to address the matter of 'FIFO is fundamentally broken under Mesa's Wayland WSI right now.'

My Wayland Your Wayland Our Wayland

I <3 Open Source That should be obvious by now, right? I’ve been out here blogging about Open Source stuff for over a decade, and occasionally I still have time to actually write code.

1

09:38

Mike's blogpost is about how they'd fix the governance issue. And the first link is about the temporary solution they came up with to get things moving

meloƞ

Scrumplex is nominated to be part of the nix steering committee!!!

:3

meloƞ

Scrumplex is nominated to be part of the nix steering committee!!!

make sure to vote if you are eligible ^^

Scrumplex

make sure to vote if you are eligible ^^

❤️

heinrich5991

1. that's by design, I think 2. not sure how it's redundant, but it's also not used anymore, so whatever 3. because the engine deals with input timing etc 4. I don't understand what you mean by self-replicating 5. probably to make sure that we get map data for the correct map

first of all it was a small review of 0.6 protocol and i don't try to force anyone to break the protocol, it's more about the time when protocol breaking will be a thing (i doubt current protocol messages will migrate like this though), but if i'll elaborate anyway: 1. but they have same layout with same information 2. that's why it's redundant :D, it's literally unused 3. i guess, it's fair, i didn't really dive into teeworlds engine part 4. situation like clstartinfo and clchangeinfo, they have same layout and the only difference that one's being sent by server and other by client, why they can't use the same message to indicate it 5. but in 0.7 protocol they send crc32 and sha256 only once, because it's useless to send them on every chunk, i guess ( i don't really know, maybe, im missing something)

clstartinfo and clchangeinfo are both sent by the client AFAIK (they also both have the cl prefix). one is used as part of the login sequence. I don't think this has something to do with "backward compatibility"

if you are about 4 then conready and ready is just message id, with no information at all

12:05

i don't really know why they should be divided in two identical messages

yes, I saw that the 0.7 protocol makes a different choice for the map data. they also make some other weird choices, like map data messages not being self-contained — you need to have outside knowledge to know the length of the contained data

yeah serverinfo is kinda meh

I think conready and ready signal different readiness levels of client/server

12:06

i.e. different stages of the loading process

wait, but conready is sent right after server receives ready from client

heinrich5991

yes, I saw that the 0.7 protocol makes a different choice for the map data. they also make some other weird choices, like map data messages not being self-contained — you need to have outside knowledge to know the length of the contained data

but anyway, is there any reason to keep the same crc (of whole map i mean, it's not crc chunk or anything else afaik) on every chunk packet

it makes the protocol a little bit more robust against stray map download packets

12:10

but I guess it's kinda random

meloƞ

Are we able to specify a path to the autoexec_server.cfg file ?

./ddnet_server -f file.cfg

meloƞ

Or let me reword that, can the .cfg file handle absolute paths ? (eg exec ~/.config/ddnetserver/myconfig.cfg)

yes

12:35

12:39

@heinrich5991 hi, can you review 2 of my prs catrose

, i mean finish reviewing AB one and check new one

Anime.pdf

yes

Thanks darling

Anime.pdf

./ddnet_server -f file.cfg

with "-f file.cfg" data/autoexec_server.cfg it will be executed anyway

Anime.pdf

@heinrich5991 hi, can you review 2 of my prs catrose

, i mean finish reviewing AB one and check new one

no promises

heinrich5991

no promises

that's fine, just don't forget about me

12:49

Matodor

with "-f file.cfg" data/autoexec_server.cfg it will be executed anyway

delete it

1

Anime.pdf

delete it

It is better not to execute these files if a specific config is specified via launch parameters exec data/autoexec_server.cfg if you need it

12:56

I didn't know about this, I recently launched a public server and was very surprised when players started voting for "Shutdown server"

https://github.com/rust-lang/rust/pull/129086 omg finally

Stabilize is_none_or by slanterns · Pull Request #129086 · rust-l...

Closes: #126383. @rustbot label: +T-libs-api r? libs-api

12:56

in 10 days

12:57

i can rerwite my code lol

12:57

i hated that so much

Matodor

I didn't know about this, I recently launched a public server and was very surprised when players started voting for "Shutdown server"

you don't need autoexec cfg in most cases, just start a lot of servers with script specifying per-server config with -f that will set individual server settings + execute common.cfg that will have settings common for all servers

12:57

thats how ddnet and most servers work

Jupstar ✪

https://github.com/rust-lang/rust/pull/129086 omg finally

you can implement stuff like this by yourself, just needing an use util::IsNoneOr as _; at the top of each file where you use it

12:59

no need to wait for official stabilization

i can do a lot of stuff if i dont wait for other things. but manually writing an include is defs less attractive than just having it

there's even a crate that implements it apparently: https://docs.rs/is_none_or/0.1.0/is_none_or/

is_none_or - Rust

Checking conditions with the IsNoneOr trait

Jupstar ✪

i can do a lot of stuff if i dont wait for other things. but manually writing an include is defs less attractive than just having it

can your IDE also automatically include a use when you use a trait method? I know rustrover can do that

heinrich5991

can your IDE also automatically include a use when you use a trait method? I know rustrover can do that

yes it can

so no need to manually write the use there

but i'd still need to link the crate everytime i use it

yes

that is smth my ide cant do

but it's one of the cool features of rust. you can extend standard library types without waiting for the standard library to do it

13:01

I'm missing that when I write C++

@MilkeeyCat THATS IT your language needs to automatically add dependencies xDD just type in vk:: and bam dep added

heinrich5991

but it's one of the cool features of rust. you can extend standard library types without waiting for the standard library to do it

i agree with that yeah

Jupstar ✪

@MilkeeyCat THATS IT your language needs to automatically add dependencies xDD just type in vk:: and bam dep added

i wrote ~350 loc in it and it didn't segfault even a single time

MilkeeyCat

i wrote ~350 loc in it and it didn't segfault even a single time

nice

it already sends connect packet to server, and recieves data back

13:03

now im trying to decode all dis stuff

hello my god friends

Anime.pdf

you don't need autoexec cfg in most cases, just start a lot of servers with script specifying per-server config with -f that will set individual server settings + execute common.cfg that will have settings common for all servers

yes, that's why I say the server should not automatically execute autoexecs files if a specific config was specified

camka23

hello my god friends

hello, what do you want from us gods?

Jupstar ✪

hello, what do you want from us gods?

fire, obviously

13:09

now we gotta figure who's prometheus

MilkeeyCat

now im trying to decode all dis stuff

Don't need to decode much of anything if you just want to get a tee ingame

i need a token

13:13

no?

Oh actually you do, I forgot it's 2024 and we have tokens

13:14

Look at the logs I have to go off of to fix an error at work today: 3:09PM ERR Activity error. ActivityType=ListInventoryLevels Attempt=4 Error="Unknown Error"

13:15

idk what part of the stack is throwing that very informative "Unknown Error" all the way to the top, but thank

next rust stable is a pretty huge deal: https://github.com/rust-lang/rust/pull/122792

Stabilize min_exhaustive_patterns by Nadrieril · Pull Request #12...

Stabilisation report I propose we stabilize the min_exhaustive_patterns language feature. With this feature, patterns of empty types are considered unreachable when matched by-value. This allows: e...

huh, nice

13:26

although I'd say it's only a QoL change, not a dealbreaker, since you could achieve the same with more code before

Learath2

Look at the logs I have to go off of to fix an error at work today: 3:09PM ERR Activity error. ActivityType=ListInventoryLevels Attempt=4 Error="Unknown Error"

temporal (probably) in sight

jxsl13

temporal (probably) in sight

It was goshopify

urgh :0

the map format should be more extensible,

Like with images, it should also bundle the game/front tile graphics (and editor tiles if they differ) if they arent included in the base client (external), this allows us to not need to bundle all the new mods game tiles, etc
Revise any compression algorithm if it can be improved
Revise any data format if it can be improved, i think it should be improved with the following prioritis if needs be: extendability > compression > load speed, just for the map format case, since its mostly static data that needs to be stored and loaded.
Also maybe a external optional file describing the tiles, like we do for the current game tiles iirc, but for other mods.

KebsCS

[ddnet/ddnet] Pull request opened: #9125 Fix color validation in some commands

Fixes #9122

Checklist

[x] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Written a unit test (especially base/) or added coverage to integration test
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#u...

Ryozuki

the map format should be more extensible,

Like with images, it should also bundle the game/front tile graphics (and editor tiles if they differ) if they arent included in the base client (external), this allows us to not need to bundle all the new mods game tiles, etc
Revise any compression algorithm if it can be improved
Revise any data format if it can be improved, i think it should be improved with the following prioritis if needs be: extendability > compression > load speed, just for the map format case, since its mostly static data that needs to be stored and loaded.
Also maybe a external optional file describing the tiles, like we do for the current game tiles iirc, but for other mods.

If I were doing it today, I'd probably offload everything to zip

14:12

A metadata json, an assets folder for all embedded assets, and the layers directly as uncompressed binary files. Deflate should take relatively good care of it

14:14

Also layer types as uuids, tele-v1@maplayer.ddnet.org

Ryozuki

the map format should be more extensible,

Like with images, it should also bundle the game/front tile graphics (and editor tiles if they differ) if they arent included in the base client (external), this allows us to not need to bundle all the new mods game tiles, etc
Revise any compression algorithm if it can be improved
Revise any data format if it can be improved, i think it should be improved with the following prioritis if needs be: extendability > compression > load speed, just for the map format case, since its mostly static data that needs to be stored and loaded.
Also maybe a external optional file describing the tiles, like we do for the current game tiles iirc, but for other mods.

the first is defs interesting, but ppl like to mod the entities, but as a base image sounds cool

extensibility > compression > load speed

Interestingly I'd go the exact opposite, extensibility can be done on a higher level

Also maybe a external optional file describing the tiles, like we do for the current game tiles iirc, but for other mods.

Could be part of the physics module, which the editor could load. Thought about doing that, but isn't high prio enough for me

Ryozuki

the map format should be more extensible,

Like with images, it should also bundle the game/front tile graphics (and editor tiles if they differ) if they arent included in the base client (external), this allows us to not need to bundle all the new mods game tiles, etc
Revise any compression algorithm if it can be improved
Revise any data format if it can be improved, i think it should be improved with the following prioritis if needs be: extendability > compression > load speed, just for the map format case, since its mostly static data that needs to be stored and loaded.
Also maybe a external optional file describing the tiles, like we do for the current game tiles iirc, but for other mods.

due to #7777 being closed with the reasoning stated there, moddability r.i.p.

https://github.com/ddnet/ddnet/issues/7777

Source-level moddability discussion · Issue #7777 · ddnet/ddnet

How much do we want to support moddability in the DDNet codebase? Note that this doesn't talk about the network protocol but about the actual code.

such maps are probably also "out of scope".

The problem is some ppl don't want a mod, they want a different game

14:48

So they gotta wait for dd-pg, just another 20 years, like always

def-

[ddnet/ddnet] Pull request closed: #9125 Fix color validation in some commands

14:49

github-merge-queue[bot]

[ddnet:master] 2 new commits

6d7aeab Fix color validation in some commands - KebsCS 46abb54 Merge pull request #9125 from KebsCS/pr-command-argument-validation - def-

14:49

def-

[ddnet/ddnet] Issue closed: #9122 player_color_body has incorrect arguement validation

Jupstar ✪

So they gotta wait for dd-pg, just another 20 years, like always

Jupstar ✪

The problem is some ppl don't want a mod, they want a different game

meeee :3

15:07

teeworlds looks like a good start point, it has everything and nothing at the same moment

heinrich5991

[ddnet/ddnet] Pull request closed: #8307 Add cl_show_chat_mention (Closed #8299)

It supports only 16 players which is nonsense nowadays. Plugging the 64 players support means a lot of work. Then a mod needs the new ddnet registration to be listed. Then it needs NetEx uuid-based system (and various CNetObj_GameInfoEx, CNetObj_DDNetCharacter, etc). 100+ hours to get the ground.

Replying to @zhn teeworlds looks like a good start point, it has everything and nothing a…

Kaffeine

It supports only 16 players which is nonsense nowadays. Plugging the 64 players support means a lot of work. Then a mod needs the new ddnet registration to be listed. Then it needs NetEx uuid-based system (and various CNetObj_GameInfoEx, CNetObj_DDNetCharacter, etc). 100+ hours to get the ground.

im not about well-prepared mod developers though, im about programming newbies that have literally no experience in both programming and gamedev

16:23

you have assets, sounds, prepared net and bunch of things to start with

16:23

teeworlds itself as a game engine sucks ofc

MilkeeyCat

i wrote ~350 loc in it and it didn't segfault even a single time

there was no segfault but it was writing to wrong memory location lol

1

heinrich5991

[ddnet/ddnet] Issue closed: #7299 Mod Support: Support custom sound events

Ryozuki

the map format should be more extensible,

Like with images, it should also bundle the game/front tile graphics (and editor tiles if they differ) if they arent included in the base client (external), this allows us to not need to bundle all the new mods game tiles, etc
Revise any compression algorithm if it can be improved
Revise any data format if it can be improved, i think it should be improved with the following prioritis if needs be: extendability > compression > load speed, just for the map format case, since its mostly static data that needs to be stored and loaded.
Also maybe a external optional file describing the tiles, like we do for the current game tiles iirc, but for other mods.

I think I'd actually like to debundle images etc. to be downloaded separately from the server
agree that another compression algorithm might be nice
too vague IMO. I'd probably do it the way @Learath2 proposed, dump most stuff into JSON inside a zip
I don't understand the last point about describing tiles, do you mean the game tile descriptions? I really don't think that should be part of actual maps, it'd get out of sync with the mods instantly. perhaps we could get mod packs for the editor that include game tiles plus their descriptions instead

16:44

wdym with more extensible btw, what are you missing?

heinrich5991

[ddnet/ddnet] Issue closed: #7809 fix strong and weak indicator when sv_no_weak_hook 1

heinrich5991

I think I'd actually like to debundle images etc. to be downloaded separately from the server
agree that another compression algorithm might be nice
too vague IMO. I'd probably do it the way @Learath2 proposed, dump most stuff into JSON inside a zip
I don't understand the last point about describing tiles, do you mean the game tile descriptions? I really don't think that should be part of actual maps, it'd get out of sync with the mods instantly. perhaps we could get mod packs for the editor that include game tiles plus their descriptions instead

but what if players want to share map, they should archive their map with images every time?

zhn

but what if players want to share map, they should archive their map with images every time?

that is what ddnet is kinda doin rn

Jupstar ✪

that is what ddnet is kinda doin rn

i mean ddnet bundles images, heinrich proposes to debundle them

zhn

i mean ddnet bundles images, heinrich proposes to debundle them

he wants to zip everything i think

17:41

but generally i think at least the server should debundle them

17:41

and distribute them individually

17:41

but i defs understand that this is less intuitive for a normal user

then i can't process "debundle" and "zip archives with map info" in one message

zhn

then i can't process "debundle" and "zip archives with map info" in one message

well all file formats individually

17:42

into a zip

eh

That's complicated. Unfortunately a detailed discussion on this topic was wiped (or hidden from users) when DDNet Season 2 (or what season it was) was removed from server channels list. One of the options is to implement a sparse map download. In that case the client downloads the map without bundled assets, and then ask the (game or https) server for missing parts, and then bake it into a complete .map in downloads/.

mh and then?

17:45

then u can also download the whole container at once

17:45

if they are unbundled on the filesystem it saves space

In that case users can share maps as-is, and the assets are still debundled e.g. not downloaded if the client already have them. I don't remember the detailed plan right now but the map should reference the assets by names or by UUID or by sha256, and then (depending on what we want) an asset can be updated (uuid) or be carved in stone forever (sha). Like: the client has downloaded assets/<uuid>, download missing stuff and inserts it into the map file if/when necessary.

but in any way, the map format has other problems than distribution or size problems

Are we really struggle to save players disk space?

are we struggling to save a bit network traffic? xd

Sure. The download time is a real problem for some players/mods, and it would be nice to have assets update.

Replying to @Jupstar ✪ are we struggling to save a bit network traffic? xd

is that a network limitation, or simply our way to download files

17:53

the more interesting format is actually demo anyway.

it needs to store the map
in future anything mod related
server settings

etc.

it'd be nice if a map could be updated and only the changed layer could be downloaded

heinrich5991

it'd be nice if a map could be updated and only the changed layer could be downloaded

that sounds like quite a bit of complexity added tho

17:54

to save 500KB download size?

i.e. even if the mod has 20 MiB of assets, it'd be nice if all of the images and sounds don't need to be redownloaded

Our way to download files. IIUC it is the same limitation which makes us using 'rates' instead of sending all commands list on rcon client authenticated. See e.g. https://github.com/ddnet/ddnet/pull/7650

Massively improve the speed of rcon commands sending. by furo321 · ...

Previously it took forever to receive all rcon commands. It's now sent 64x faster! Checklist Tested the change ingame Provided screenshots if it is a visual change Tested in combination...

if the map changes

i agree

17:54

hi heinrich

you probably want to download maps via HTTPS anyway — that's doable for everyone today

heinrich5991

i.e. even if the mod has 20 MiB of assets, it'd be nice if all of the images and sounds don't need to be redownloaded

but you cannot have x without changing y. how do you want to store your map related assets on disk then?

u dont have to send the whole file to reload map

17:56

just some payload indicating what changed

Ewan

just some payload indicating what changed

compared to what

Jupstar ✪

but you cannot have x without changing y. how do you want to store your map related assets on disk then?

either bundled or unbundled, but that's a separate decision. what's x and y?

Jupstar ✪

compared to what

map filename and/or hash of current map (which should have been sent when the client connected)

live updates sound more complicated though

heinrich5991

either bundled or unbundled, but that's a separate decision. what's x and y?

if it's bundled, u will have either 2 files or delete one if u have 2, you increase disk space by 20MiB (assuming your assets) if u unbundle them, then you could say it's user unfriendlier again

they both just keep a working in-memory copy already

17:57

keeping it in sync just sounds like the potential issue

yup, those are the tradeoffs of bundled vs unbundled on disk

u dont have to send any more than the original map file

17:58

assuming u know what

17:58

is changing

generally from a programmer perspective i'd go with unbundled. but i also can understand that sharing the files sucks then

Ewan

u dont have to send any more than the original map file

yes

17:59

rn on dd-pg i tried to go with unbundled for pretty much everything

17:59

and later think how much it really sucks

18:00

(and ofc also split assets from map then)

Can it be that selfbuilt will use other versions of libs, than official release?

heinrich5991

I think I'd actually like to debundle images etc. to be downloaded separately from the server
agree that another compression algorithm might be nice
too vague IMO. I'd probably do it the way @Learath2 proposed, dump most stuff into JSON inside a zip
I don't understand the last point about describing tiles, do you mean the game tile descriptions? I really don't think that should be part of actual maps, it'd get out of sync with the mods instantly. perhaps we could get mod packs for the editor that include game tiles plus their descriptions instead

I think I'd actually like to debundle images etc. to be downloaded separately from the server

E.g. this is also what i want. server side assets can then do lot of things

Anime.pdf

Can it be that selfbuilt will use other versions of libs, than official release?

y<es

18:01

for shared libs that are not shipped with the client that is the case

that is unfortunate

Makes sense but it is complicated in some cases. E.g. infclass generates client maps on demand — depending on the current server settings, and e.g. it adds map assets on the fly.

Replying to @heinrich5991 you probably want to download maps via HTTPS anyway — that's doable fo…

18:03

Still, can you please recommend an https server (software)? python -m http.server? nginx? Apache? Maybe it is possible to generate some client maps set for usual server settings at least.

Why debundle anyway? What feature do you want to support with that?

Kaffeine

Still, can you please recommend an https server (software)? python -m http.server? nginx? Apache? Maybe it is possible to generate some client maps set for usual server settings at least.

i like flask, it's very easy. axum is also very good for making a high performance web server without a lot of headache

Learath2

Why debundle anyway? What feature do you want to support with that?

Think of the demo format, and lets additionally assume u have wasm files, lua scripts what do i know. + server side assets that are REQUIRED (so cannot miss)

18:05

u end up with pretty annoying huge files

I'd suggest nginx or caddy, but any https capable server works

Thanks

caddy has automatic https built in

Ah you are thinking of stuff like that. Yeah a map bundle can not contain everything needed for a server that can send wasm/lua at runtime

can you make the infclass server output the map file into a directory? Kaffeine

18:08

a HTTPS server could then directly serve this file

heinrich5991

it'd be nice if a map could be updated and only the changed layer could be downloaded

sounds like docker images

I would keep it bundled, let the server send patches to the map for runtime stuff. Said patches would be saved in the demo as messages which would keep it reproducible

how would you handle skipping?

heinrich5991

how would you handle skipping?

(either pre or post) I would scan through the demo to get a list of ticks where patches are applied

and add file offsets somewhere

This could also be generated while saving the demo removing the cost from demo load

sounds sane

Learath2

(either pre or post) I would scan through the demo to get a list of ticks where patches are applied

how can this be known

18:11

actually what even is the goal

18:11

why demo file

18:11

i missed all that

heinrich5991, yes, the generated client maps are saved to files. It means the host 'd have to run own https server but yeah, it can work

. In other case I could upload maps to github pages, and skip https cert acquiring.

if you don't like the https cert acquiring process, I'd suggest using caddy

18:15

you just start it and it deals with acquiring the certificate automatically

that does sound very handy

caddy is pretty easy to use

cert management never sounds like a problem until ur on a distro that doesn't have the nginx letsencrypt plugin or whatever

18:19

historically i have used cloudflare to proxy my http traffic at the endpoint and then you literally don't need to think about SSL certs

18:19

but then ofc there's a whole section of your pipeline where the traffic is not encrypted

an example config: /etc/caddy/Caddyfile

zcat.ch {
    root * /var/www/html/zcatch/public
    file_server
    log {
        output file /var/log/caddy/zcat.ch.access.log {
            roll_size 2mb
            roll_keep 5
        }
    }
}

pic.zcat.ch {
    redir https://discord.TLD/38aDRSfVKt
    log {
        output file /var/log/caddy/pic.zcat.ch.access.log {
            roll_size 2mb
            roll_keep 5
        }
    }
}

(edited)

lol