DDraceNetwork - developer

DDraceNetwork

Development / developer

Development discussion. Logged to https://ddnet.org/irclogs/ Connected with DDNet's IRC channel, Matrix room and GitHub repositories — IRC: #ddnet on Quakenet | Matrix: #ddnet-developer:matrix.org GitHub: https://github.com/ddnet

Between 2024-06-03 00:00:00Z and 2024-06-04 00:00:00Z

xxxxD

Replying to egyt why do you have a separate CTF capture sound for Polish

00:15

Idk what en and pl stands for but I don’t think it’s languages. Both sounds are used to differentiate which flag was captured

not which but which time

00:18

sounds differ if you grab flag from stand and from died player

The sounds he sent is capture not grab

00:19

Oh wait no grab

00:19

Xd

00:20

But the same exists for capture too

00:20

I can’t even guess what en and pl stand for

heinrich5991

@Learath2 does this look similar to what we need for bans? https://debezium.io/

I didn't have that great experience trying to use Debezium to mirror our ranks from MySQL (but could have been a mistake from my side), the setup seemed complex and finicky too

heinrich5991

anticheat is a losing game

true, but it can be used as a compromise for a smoother client experience

louis

true, but it can be used as a compromise for a smoother client experience

doesn't sound worth the limited developer and reviewer bandwidth the project has

anticheat is not always a losing game, it depends on the game type

02:03

if you never trust the client then the cheats are limited by their ability to play the game legitimately better than a human, which is sometimes not possible depending on your game.

[ddnet/ddnet] Pull request opened: #8441 Allow client to re-send past inputs to the server

Requires #8432 Does 2 main things. 1. If the client has enough prediction margin that it can send send duplicated inputs before the original input reaches the "simulation tick" on the server then it will try to send all the inputs that it sent in the past which can get there fast enough. For a better explanation please see this video at the time stamp, it has very nice animated visuals. https://youtu.be/W3aieHjyNvw?si=WQhYO44zpI4OowcC&t=1891 In our case the physics code doe...

^ the end of lag :)

03:39

(in solo, until 8372) (edited)

Tater

^ the end of lag :)

:0

Tater

(in solo, until 8372) (edited)

<ChillerDragon> https://github.com/ddnet/ddnet/blob/64f4ea1512b5abcc1ef27b90ede05cb9934df69b/src/engine/shared/huffman.cpp#L57 this overflows on init

ddnet/src/engine/shared/huffman.cpp at 64f4ea1512b5abcc1ef27b90ede0...

DDraceNetwork, a free cooperative platformer game. Contribute to ddnet/ddnet development by creating an account on GitHub.

06:33

<ChillerDragon> not sure if that is intended i really don't understand the huffman code but even if it is intended it is a bit weird

06:35

<ChillerDragon> https://github.com/ddnet/ddnet/blob/64f4ea1512b5abcc1ef27b90ede05cb9934df69b/src/engine/shared/huffman.h#L30 symbol is a unsigned char and i reaches 256

ddnet/src/engine/shared/huffman.h at 64f4ea1512b5abcc1ef27b90ede05c...

DDraceNetwork, a free cooperative platformer game. Contribute to ddnet/ddnet development by creating an account on GitHub.

06:35

<ChillerDragon> so it flips to zero in the last iteration

06:35

<ChillerDragon> thats the eofSymbol so maybe its intended but still :D

Where does the init table come from?

06:45

I thought huffman was supposed calculate the optimal frequency table for any given data then store that frequency table with the data so it can be decompressed with the optimal table, but it seems like we just use the same one for all data as far as I can tell.

06:45

so someone must have created it from some TW packet data?

06:45

or something

06:48

omg of course it's 15 years old .-. (edited)

06:48

https://github.com/ddnet/ddnet/commit/878ede3080ab2cfb627aca505c397d9765052996#diff-c39dd1d454f6249c201c0662b73ccf1aaee23f00bcc5ae34d034f96d325d8ec0

major update with stuff · ddnet/ddnet@878ede3

someone was writing ddnet huffman file in rust? https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=8505fd268fea3cba6468da5447571347

Rust Playground

A browser interface to the Rust compiler to experiment with the language

07:02

unless this frequency table is some sort of standard made by someone else it's possible we are leaving free bandwidth savings on the table by either regenerating it or sending the huffman tree along with the packet, but I think our packets are too small for that

Tater

unless this frequency table is some sort of standard made by someone else it's possible we are leaving free bandwidth savings on the table by either regenerating it or sending the huffman tree along with the packet, but I think our packets are too small for that

well aparently not free if u need to sync such a table with the client

07:06

btw @Tater ur pr sends the inputs as individual packets right?

it only does 1 flush which means they get compressed together

07:06

but they are processed individually yes

<ChillerDragon> @Tater the table is pre calculated and hardcodet into the client and server

i see, maybe a new network packet would be nice where it takes the binary diff of the inputs.. since most bytes are mostly similar

ok but it's 15 years old and highly likely does not match the current distributation of data

<ChillerDragon> someone probably ran the numbers of which format the tw trafic has

also there's no comments in the original commit so we have no idea the quality of it's generation even 15 years ago (edited)

<ChillerDragon> i dont think that much changed in the last 15 years

07:07

<ChillerDragon> idk about 15 but the last 10 years for sure it barley changed

simply changing which type of packet gets sent more often would significantly change the frequency table

<ChillerDragon> its mostly snap thats relevant i would say

true

<ChillerDragon> always has been i guess

but still we have no idea how much data was sampled to create it in the first place

07:08

or how

<ChillerDragon> yea

Jupstar ✪

i see, maybe a new network packet would be nice where it takes the binary diff of the inputs.. since most bytes are mostly similar

could you explain more?

<ChillerDragon> omg i dont wanna debug the fakin huffman code -.- im procrastinating so hard

I think compressing the inputs more would be good for sure

Tater

could you explain more?

the oldest input packet is the base and all new inputs are binary patches of that one

07:10

so all other input packets are basically only the difference between the inputs

the client does that already sorta

07:10

or you mean something more special?

but the client uses a input packet that is ack'd by the server or not?

07:10

with 300 ping that is pretty old

I'm not sure storing binary patches for a 120byte struct is practical

Tater

I'm not sure storing binary patches for a 120byte struct is practical

why not

you need pointer to location of each patch?

07:11

I guess it doesn't need to be 4 byte pointer

07:12

actually I have no idea how binary diff algorithm works tbh

07:12

maybe it's more clever

let say u have two vec<u8> then it simply puts a 0 where the binary diff is equal else it puts a diff

07:13

very simplified (edited)

07:13

and then compressions can normally very well compress these 0's

the client and server will bundle I think 1400bytes of data to pass to the huffman which is about ~10 inputs. I'm not sure to what extend the huffman algorithm will naturally perform something similar to your idea by nature of how it works, it would probably be quite good without the fixed frequency table but now I'm not sure it's even doing anything anymore

Jupstar ✪

and then compressions can normally very well compress these 0's

can it

troll

zhn

can it

troll

Tater

the client and server will bundle I think 1400bytes of data to pass to the huffman which is about ~10 inputs. I'm not sure to what extend the huffman algorithm will naturally perform something similar to your idea by nature of how it works, it would probably be quite good without the fixed frequency table but now I'm not sure it's even doing anything anymore

mh well i can only tell you compression algorithms as good as they are, are also not perfect.. for teeworlds snapshots taking binary diffs with zstd outperformed any compression i tested easily. The only disadvantage of binary patching is that the base "file" can get old (e.g. by a higher ping)

well it should always arrive in the same "cluster" of inputs so you start the diff from the start with every packet (edited)

Jupstar ✪

mh well i can only tell you compression algorithms as good as they are, are also not perfect.. for teeworlds snapshots taking binary diffs with zstd outperformed any compression i tested easily. The only disadvantage of binary patching is that the base "file" can get old (e.g. by a higher ping)

yeah probably because our huffman table is trash lol

but how do u want to create a general purpose huffman table for snapshots, inputs etc.

07:22

or do u want to use different tables per purpose?

I don't think there's a correct answer

07:23

but remember that having a fixed table means it changes the network protocol anytime we update it (edited)

07:24

I mean a seperate table for every type of message would be very good tbh but I think the network system smushes all the messages together regardless of type

07:24

which might be better anyway since many messages are much less than the 1400byte size

i think heinrich did tests in that direction

also why do we use 1400bytes

07:25

seems small

07:25

UDP can be 64kb I think

cuz its data mtu

a UDP frame usually isnt much bigger than 1400 bytes if u think cross plattform afaik

zhn

cuz its data mtu

here we have an expert

07:27

epyc

no u

07:27

i can say anything about net until Devinci comes and tells what exactly is wrong kekw

07:28

bruh this uni sucks why do i need to wait like 2 years just to start mess up with networks and its internals

So, this is the list of strings I couldn't translate, some because I don't know where they occure, and some because I don't know technical terms behind them

(paused)
All combined
Folder Link
Follow
Game paused
Grabs
Moved ingame
Replace video
Searching

[Graphics error]
An error during command recording occurred. Try to update your GPU drivers.

(edited)

last ones are pretty verbose

I never heard "command rendering", that's why I asked about them too (edited)

drawing operations and memory transfers are called render commands at least in vulkan

07:38

framebuffer is buffer of frames, practically transliteration is used

07:39

of frame*

"did not understand anything, but very interesting"

Tater

if you never trust the client then the cheats are limited by their ability to play the game legitimately better than a human, which is sometimes not possible depending on your game.

"anticheat" as was used in the discussion was explicitly about trusting the client and then trying apply heuristics/detect cheats on top of it

yeah that's fair

I'll be editing this message when I find an appropriate translation for a certain string

Replying to egyt So, this is the list of strings I couldn't translate, some because I don…

08:37

And, btw, the progress is 95% (581 out of 607 strings translated), so the PR is not really that far away!

Jupstar ✪

mh well i can only tell you compression algorithms as good as they are, are also not perfect.. for teeworlds snapshots taking binary diffs with zstd outperformed any compression i tested easily. The only disadvantage of binary patching is that the base "file" can get old (e.g. by a higher ping)

actually double checking the inputs are only like 40bytes which is a lot less than I thought

Tater

actually double checking the inputs are only like 40bytes which is a lot less than I thought

and how many packets will be deployed when having 300ms prediction margin?

plus 3 ints

hi

Jupstar ✪

and how many packets will be deployed when having 300ms prediction margin?

300/20ms = 15 inputs, so ((40+4+4+4)*15)/1400 = 0.55 packets (at most, non-unique inputs are skipped) (edited)

08:44

probably a bit more with the overhead from message struct but idk how big that is

Image attachment

08:52

Hm, for some reason "Time limit" is not translated, while "Score limit" is

interesting

08:54

maybe it’s worth doing a quick regex search to find which string literals are not translated, and then determine if they should be translated

08:54

not a catch-all but probably still worthwhile

Jupstar ✪

i see, maybe a new network packet would be nice where it takes the binary diff of the inputs.. since most bytes are mostly similar

not sure if someone already answered, but this is exactly what is done already in snap deltas ^^

egyt

Hm, for some reason "Time limit" is not translated, while "Score limit" is

There probably is no reason, just an oversight

Haha, by remaking the translation I found so many bugs...

yes, thanks for actually caring about a good translation!

egyt

Click to see attachment 🖼️

Looks like there are two time limit strings, did you also translate the other one or does it not work for some reason? For example in german.txt:

Time limit
== Zeitlimit

Time limit: %d min
== Zeitlimit: %d min

I translated both

Time limit
== Обмеження по часу

Time limit: %d min
==Обмеження по часу: %dхв

(edited)

10:57

This is the list of strings I haven't translated yet, btw

Need some help

(edited)

Replying to egyt So, this is the list of strings I couldn't translate, some because I don…

egyt: can you mark the strings that you don't know where they appear?

11:00

I can figure out where they appear

Patiga

not sure if someone already answered, but this is exactly what is done already in snap deltas ^^

of multiple snaps?

I think it rarely ever sends an actual full snap, and typically just deltas, which only include items that change, and for those items the binary diff

11:07

yea the server diffs multiple snaps to create efficient updates

mh i see, but i think u dont even need to do the items part zstd + custom dict performs just as good if not even better

All of those mentioned in my list Would be nice if you could send screenshots of GUI too (edited)

Replying to @heinrich5991 egyt: can you mark the strings that you don't know where they appear?

Patiga

yea the server diffs multiple snaps to create efficient updates

that confuses me xD

11:08

it diffs multiple snaps for a single snap or it sends previous snaps to create a snapshot chain?

well the item part helps because most items don't need to be updated coz they stay the same

Jupstar ✪

it diffs multiple snaps for a single snap or it sends previous snaps to create a snapshot chain?

Server sends only the changes

well thats why it should help

as snap delta

in reality ofc you might find that this optimization doesn't help

fokkonaut

as snap delta

yes but i don't see how that makes the binary diff better

a

well maybe it does, but i could not observe it

(edited)

11:10

@Patiga what i wonder tho is if the snapshot packet is used to unpack old snapshots

Jupstar ✪

it diffs multiple snaps for a single snap or it sends previous snaps to create a snapshot chain?

it relies on snapshots you received earlier, not ones in the future

yes

11:10

but in the input case of tater it's about actually sending old inputs

11:10

with high prediction margin they still come in before the tick that handles these inputs is used

11:11

so you basically have a higher success rate when packet drop

I'm confused about your usage of "unpack"

u basically have a chain of inputs until the current one

11:12

well it sounded a bit like snapshots are used similary like the input concept here

11:12

but i guess it just sends the current snapshot

iirc, input isn't in the snapshot

Patiga

not sure if someone already answered, but this is exactly what is done already in snap deltas ^^

well i think we just talk past each other xD

yea seems likely

like what i want is: binary patch the input chain:

very old input
old input
input old input can be binary diff of the very old one

current input of old input* (edited)

11:14

they are directly send in the same network packet anyway

11:14

if the very old input is a binary diff smth like it's done with snapshots is ofc possible yeah

11:14

basically my optimization is in first place unrelated to any binary diff of a previous packet

11:15

just so u can send a lot of inputs in a single packet

Jupstar ✪

well i think we just talk past each other xD

ah right, I jumped into the discussion at the wrong time, with a wrong thing in mind xd

11:16

sounds fun though :)

While I can guess where they appear, I can't be sure, as it happened with yesterday stuff

Replying to egyt All of those mentioned in my list

I had an interesting idea to improve performance in rendering on at least some maps, which is already tested and proven to be useful, is anyone interested? maybe @Jupstar ✪

"NOT CONNECTED"

displayed in rcon while not connected to a server

12:04

"Trying to determine UDP connectivity…"

displayed in the connecting screen when the server does not respond, but testing the internet connection hasn't finished yet

Okay, 24 left

"Loading commands…"

displayed in rcon while not all command descriptions have been downloaded from the server yet

12:11

"Personal best:"

best race time of the current map of the current player

"Server best:"

best race time of the current map on the server

12:13

"The audio device couldn't be initialised."

this one looks a bit misleading. it just means "couldn't initialize the sound system, the client will not be able to play sounds"

12:14

"The server is running a non-standard tuning on a pure game type."

displayed when the server admin changed the server source code for a vanilla gametype (DM, CTF, TDM) and changed tunings (physic parameters, like gravity) there

Yet another bug? ❤️

Replying to @heinrich5991

"The audio device couldn't be initialised." […]

"Sound error"

this is just the title of the message box of the aforementioned audio initialization error

egyt

Yet another bug? ❤️

sure

AssassinTee

I had an interesting idea to improve performance in rendering on at least some maps, which is already tested and proven to be useful, is anyone interested? maybe @Jupstar ✪

you can share ur idea

"Add"

opposite of the "Remove" button you already translated, used for the adding vote options in the vote option menu

12:19

"Try again"

shown in the "incorrect password" dialog that contains a field for the password, clicking "try again" will use the password in the field for connecting to the server again

Jupstar ✪

you can share ur idea

So, In a map I am working on, I have a lot of quads, about 8000 distributed in 50 groups. For context, I would be currently on r4 on the list with the most quads per map on ddnet, if my map would be there, anyway. I noticed, that a snow Strom, which was over 500 fields away from the spawn region, greatly dropped fps beside no quad of this layer being visible or in render distance. I improved the FPS by adding clip regions on every quadlayer, meaning I had to put every quad into it's own group and add a clipping region to it, because layers don't support clip regions. This causes the renderer to skip the group, if it's region is out of bounds, instead of checking EVERY QUAD if it needs rendering. This improved, in my special case, the rendering up to a factor of 10 (from 500 fps to 5000) (machine one), (from 200 fps to 990 (fps cap!))(machine 2) My idea: Add, when loading a map, a virtual clip region to a quad layer. This improves FPS everywhere outside of the clip and if you have multiple quad layers without overlap, it'll improve overall fps. How to do this? Check every corner of a quad in a layer, min-max x/y for to generate a clip region, you only need to do this once, when loading the map You'll need to add possible envelope offsets, which is easy, the maximum/minimum value has to be a point in the envelope (except when rotating) TODO para X, Y, can be considered and recalculated, definately possible, para 0/0 can be skipped Also, you could pre-calculate this region starting at a minimum quad count, which still needs to be found (edited)

AssassinTee

So, In a map I am working on, I have a lot of quads, about 8000 distributed in 50 groups. For context, I would be currently on r4 on the list with the most quads per map on ddnet, if my map would be there, anyway. I noticed, that a snow Strom, which was over 500 fields away from the spawn region, greatly dropped fps beside no quad of this layer being visible or in render distance. I improved the FPS by adding clip regions on every quadlayer, meaning I had to put every quad into it's own group and add a clipping region to it, because layers don't support clip regions. This causes the renderer to skip the group, if it's region is out of bounds, instead of checking EVERY QUAD if it needs rendering. This improved, in my special case, the rendering up to a factor of 10 (from 500 fps to 5000) (machine one), (from 200 fps to 990 (fps cap!))(machine 2) My idea: Add, when loading a map, a virtual clip region to a quad layer. This improves FPS everywhere outside of the clip and if you have multiple quad layers without overlap, it'll improve overall fps. How to do this? Check every corner of a quad in a layer, min-max x/y for to generate a clip region, you only need to do this once, when loading the map You'll need to add possible envelope offsets, which is easy, the maximum/minimum value has to be a point in the envelope (except when rotating) TODO para X, Y, can be considered and recalculated, definately possible, para 0/0 can be skipped Also, you could pre-calculate this region starting at a minimum quad count, which still needs to be found (edited)

mh i see, sounds pretty complicated to do it layer wise with all these side effects of animations. But that gave me a different idea. Couldn't you use the group's clip properties to completely skip that group? would be manual work which sucks, but would be safer to not imeplement it wrong

Image attachment

Here an extra cherry: If this is implemented, you could split a quadlayer into multiple with their own regions, further improving the fps

Jupstar ✪

mh i see, sounds pretty complicated to do it layer wise with all these side effects of animations. But that gave me a different idea. Couldn't you use the group's clip properties to completely skip that group? would be manual work which sucks, but would be safer to not imeplement it wrong

This is what I wrote above

AssassinTee

This is what I wrote above

and that's not enough or what?

@Jupstar ✪ Quads don't render outside the camera's view right? So regardless of clipping being off or on, this shouldn't affect performance if outside the camera's view, right?

u wrote 5000 fps

how can I create a fake bun on the server using the code?

This is enough, yes, the idea is to generalize this in order to improve map rendering in general?

Cellegen

@Jupstar ✪ Quads don't render outside the camera's view right? So regardless of clipping being off or on, this shouldn't affect performance if outside the camera's view, right?

it does, since it still has to check if the quad is inside the cameras view

yes

1

justatest

oh, then it's a miscalculation in performance

you can get a performance gain, by skipping a layer redering, because of the clip region, instead of checking every item in the layer

Jupstar ✪

it does, since it still has to check if the quad is inside the cameras view

doesn't it cache all of the quads that are inside the view and check those only? Or better yet, check those only which are the closest within the boundary of the camera's view? (edited)

no, because of animations

AssassinTee

This is enough, yes, the idea is to generalize this in order to improve map rendering in general?

yeah it's not trivial.. u also need to respect the quads rotation etc. would be the question how well the estimation works in worst case, e.g. if you actually are where most quads are rendered

12:44

@AssassinTee u do it for rain animations?

AssassinTee

you can get a performance gain, by skipping a layer redering, because of the clip region, instead of checking every item in the layer

It needs correction on what quad can and cannot be rendered

rain, snow, waterfalls, bells, no-animations-at-all

i'd rather question why u need many quads for those

12:45

do u need highly accurate rain drops xDD

I rather question, why do we need to check every single quad if it's outside the camera or not xd

brings me back to my initial: on ddnet are already maps, with more quads

Cellegen

I rather question, why do we need to check every single quad if it's outside the camera or not xd

This is where I agree

AssassinTee

brings me back to my initial: on ddnet are already maps, with more quads

can u name one

12:45

i want to look

There are many more efficient ways to look at this problem tho

Jupstar ✪

can u name one

mud, Victory 2, I can name more @Patiga had a list

Cellegen

I rather question, why do we need to check every single quad if it's outside the camera or not xd

the question is not about that there isnt smth better, more about complexity

Position based detection: based off of all quads' coordinates and the camera's front end (top left position) and back end (bottom right position), which quads are inside that position

12:47

if those quads' positions are inside that position, those can be rendered immediately.

Patiga

@Jupstar ✪ "Mud" appears to be a good map if you want an extreme amount of quads ^^

this looks way worse then what I did

Cellegen

if those quads' positions are inside that position, those can be rendered immediately.

With this method, only the quads' positions need to be cached and no need for direct comparision with an object colliding with the camera

Jupstar ✪

the question is not about that there isnt smth better, more about complexity

I agree, that having all things covered is complex, but we can still improve the easy cases: no animation, para 100/100, -> virtual clip

Count in the para multipliers, and it's done

12:48

Is my method any wrong by chance?

Cellegen

Is my method any wrong by chance?

yes, you need to check the corners

12:49

and points in between

the easiest approximation for quads is a circle around it, even if it rotates the circle stays the same

would it be more performance heavy, than checking the entire quad object directly?

additionally distance checks of circles + points are the cheapest operation

Just caching the quads' positions will make a huge performance difference, as long as there is no other way for a quad to move.

Cellegen

would it be more performance heavy, than checking the entire quad object directly?

you'd still need to check every quad, instead of just having one check for ALL quads

I see, so in that case, if the quads move by animation, then those need to update the cache

12:50

justatest

AssassinTee

you'd still need to check every quad, instead of just having one check for ALL quads

Thats not how caching should work but ok

12:51

if the position is stable, then it should be done like that

and then the player moves, how do you find which quads to render?

But now I doubt my method due to animations changing positions of Quads

12:52

Can you do signalling for objects?

12:53

stupid question, but worth to ask

12:53

Damn Im so obsessed with Godot, that was the first thing in my mind lol

Jupstar ✪

additionally distance checks of circles + points are the cheapest operation

this could solve the rotation animation case for the clip region

AssassinTee

I agree, that having all things covered is complex, but we can still improve the easy cases: no animation, para 100/100, -> virtual clip

so what do you say @Jupstar ✪ , is the idea valid? This would be an optimization in some cases, and wouldn't change a thing in others (no quad layers)

please tell me how to create a player token for my DDNEt server (edited)

AssassinTee

so what do you say @Jupstar ✪ , is the idea valid? This would be an optimization in some cases, and wouldn't change a thing in others (no quad layers)

sure but dont u need animation?

Yes, you can further improve on that idea a lot (since it's an optimization problem). I think I'll start an implementation of this when I am done with the map and need to measure frame times

maybe use that kind of method on tunes as well

13:00

tunes are still actively sucking off performance inside the editor

Cellegen

tunes are still actively sucking off performance inside the editor

that's an editor problem

FeelsBadMan

@Cellegen this is about in-game improvements (which is much more worth)

true

anyway, i'd still suggest to use less quads

13:01

it uses more VRAM too

only if you use quads with texture, or does it?

and rendering performance too

Idk about 6000 quads being on a map, but my idea is around 10k

13:02

jupstar, you have no idea about the potential of the editor

AssassinTee

only if you use quads with texture, or does it?

independent of if texture or not

13:02

the vertices of the quads need to be uploaded

The fix with the clip regions is working very good, it's not one layer with 8000 quads, I have a lot more in this map, the biggest layer has 2000, the rest has ~500, I don't see how to improve the number of quads

13:03

Issue is me abusing the pixelart tool

but u probably have a very good GPU

13:04

try with software renderer

how/like?

remove your driver, test it out? troll

troll

AssassinTee

how/like?

mh do you use linux?

I can if I have to

for teeworlds usually not

on windows dunno xd

13:05

maybe vulkan sdk ships one

@Jupstar ✪ do you have a tool to measure frametimes on different maps?

AssassinTee

@Jupstar ✪ do you have a tool to measure frametimes on different maps?

ddnet has a tool built in, but u'd need to manually drag it into excel to see the chart

13:06

ah actually someone wrote a python script i think

if sb wrote a script, it has to be patiga xD

AssassinTee

this looks way worse then what I did

So if I understand correctly, you want to put layers into their own group to achieve more precise clipping? Such a system would not directly help on a map such as mud though, right? (as long as we don't split up the quad layer automatically)

AssassinTee

if sb wrote a script, it has to be patiga xD

it was chairn

Patiga

So if I understand correctly, you want to put layers into their own group to achieve more precise clipping? Such a system would not directly help on a map such as mud though, right? (as long as we don't split up the quad layer automatically)

I would implement the clip-feature virtually for quads (no editor), precalculate them. This has nothing to do with the group clips (well maybe similar code) Yes, this wouldn't directly improve mud, but this would allow to, in a follow up, split a layer into two only for rendering (edited)

So, 16 left (not counting this one, still thinking of a good translation)... Nice! (edited)

Replying to @heinrich5991

"The server is running a non-standard tuning on a pure game type." […]

Cellegen

doesn't it cache all of the quads that are inside the view and check those only? Or better yet, check those only which are the closest within the boundary of the camera's view? (edited)

the graphics card is very good at throwing away quads that are not on the screen by itself. I doubt that such checks would increase the performance in most cases

https://discord.com/channels/252358080522747904/293493549758939136/933046871943962704

13:09

i think this was the script

1

Patiga

the graphics card is very good at throwing away quads that are not on the screen by itself. I doubt that such checks would increase the performance in most cases

then maybe its a gpu bottleneck?

the biggest bottleneck in ddnet is that it does these calculations on the CPU

13:10

CPU are only good if the number is not too high, then they can even outperform the GPU

well, positions need to be accurate, so makes sense

the performance gain comes only from skipping checks if something needs rendering, not the rendering itself (edited)

well it depends on which method is faster, discarding quads on GPU, or conditionally skipping quads on the CPU

AssassinTee

So, In a map I am working on, I have a lot of quads, about 8000 distributed in 50 groups. For context, I would be currently on r4 on the list with the most quads per map on ddnet, if my map would be there, anyway. I noticed, that a snow Strom, which was over 500 fields away from the spawn region, greatly dropped fps beside no quad of this layer being visible or in render distance. I improved the FPS by adding clip regions on every quadlayer, meaning I had to put every quad into it's own group and add a clipping region to it, because layers don't support clip regions. This causes the renderer to skip the group, if it's region is out of bounds, instead of checking EVERY QUAD if it needs rendering. This improved, in my special case, the rendering up to a factor of 10 (from 500 fps to 5000) (machine one), (from 200 fps to 990 (fps cap!))(machine 2) My idea: Add, when loading a map, a virtual clip region to a quad layer. This improves FPS everywhere outside of the clip and if you have multiple quad layers without overlap, it'll improve overall fps. How to do this? Check every corner of a quad in a layer, min-max x/y for to generate a clip region, you only need to do this once, when loading the map You'll need to add possible envelope offsets, which is easy, the maximum/minimum value has to be a point in the envelope (except when rotating) TODO para X, Y, can be considered and recalculated, definately possible, para 0/0 can be skipped Also, you could pre-calculate this region starting at a minimum quad count, which still needs to be found (edited)

and to get me up to speed, you achieved that optimization manually by reordering the quads layers into different groups, yea?

Jupstar ✪

i think this was the script

so matplotlib, I need to collect the data manually right?

AssassinTee

so matplotlib, I need to collect the data manually right?

ddnet has some command

13:13

let me find

13:13

bench_quit or smth

Patiga

and to get me up to speed, you achieved that optimization manually by reordering the quads layers into different groups, yea?

yes, I needed to split a group into 3 or more, where the middle one is a single group for a quad layer with clipping (edited)

benchmark_quit 60

13:14

to record 60 seconds of gameplay

Thing is this needs to be collected over a range of maps

Patiga

and to get me up to speed, you achieved that optimization manually by reordering the quads layers into different groups, yea?

also, I have already done this in the map version you already have. If you are interested measure fps, disable all clip regions, measure again (edited)

nice

13:24

I'm uncertain if this optimization does much on most maps. I would expect only very few maps to profit from this. (quads afaik are typically only used only used in small amounts) For the maps on which this does matter, it would be a great improvement and I like the idea! A relatively simple approach would be to simply edit the maps themselves, and simply use the group clips. that is exactly what they are for and many groups are no issue for the renderer :)

games like minecraft might use UV coordinate animations for rain something ddnet maps sadly dont support

13:28

but u could defs do smth similar with fewer quads than 6000 xd

Embedded image

Use both

this is a map where i created the rain for

frames_by_oreo_64a72a71.map

1.82 MB

13:30

it uses fewer quads

BTW I know I could just use para 0/0, add a clip where I want rain and call it a day

u can defs improve it, bcs i am laziest mapper but looks also cleaner in the editor

13:31

Image attachment

I wonder what you'd say about my envelops if you see them, you'd probably throw a stone at me (edited)

Jupstar ✪

Click to see attachment 🖼️

if the midground quads don't have para change at all, then you can resize the boundary to be smaller (edited)

13:34

If as you said it affects performance, then this should also increase it's performance a bit

Jupstar ✪

u can defs improve it, bcs i am laziest mapper but looks also cleaner in the editor

This is probably the prefered solution in 99% of the cases, but not mine, because my snow flakes are 4x4 virtual pixels following a grid. The animation itself only uses step type 'N' and the pixels jump to their positions, this can't be done without para x/y being 100

AssassinTee

This is probably the prefered solution in 99% of the cases, but not mine, because my snow flakes are 4x4 virtual pixels following a grid. The animation itself only uses step type 'N' and the pixels jump to their positions, this can't be done without para x/y being 100

u can also create an issue on github and try to explain it in more detail, i would not prevent it if u find maintainers that like your idea

I think I'll do a test implementation, measure, and add a PR if the stats provide it. Which maintainer would say no to 1% on average, and 1000% worst case improvements?

wow jupstar pro mapper

AssassinTee

I think I'll do a test implementation, measure, and add a PR if the stats provide it. Which maintainer would say no to 1% on average, and 1000% worst case improvements?

Aware

13:52

"just use less quads"

AssassinTee

I think I'll do a test implementation, measure, and add a PR if the stats provide it. Which maintainer would say no to 1% on average, and 1000% worst case improvements?

btw i think u can generally improve the quads rendering performance. i quickly looked into the code, rn it calculates the envelop for every quad (even tho most quads probably have the same animation) additionally env eval uses a linear search, which is probably no problem for small number of env points, but defs doesnt scale

13:57

that alone would maybe already double the fps xd

13:57

ah but animation time offsets could potentially be problematic

13:57

dunno if u use them too

yes, and I randomized them

13:57

13:58

a randomize pos/col offset button would be a nice feature as well, but can't have everything

Jupstar ✪

ah but animation time offsets could potentially be problematic

yea in twmap I calculate the envelope values for each envelope_index, envelope_offset pair

Patiga

yea in twmap I calculate the envelope values for each envelope_index, envelope_offset pair

ah nice, do you know a map that profits from it?

13:59

let say min. 128 quads

Jupstar ✪

btw i think u can generally improve the quads rendering performance. i quickly looked into the code, rn it calculates the envelop for every quad (even tho most quads probably have the same animation) additionally env eval uses a linear search, which is probably no problem for small number of env points, but defs doesnt scale

I am currently lookin into the OnMapLoading function, too. This virtual clip would be easy to implement, you are already iterating through all quads in a layer https://github.com/ddnet/ddnet/blob/64f4ea1512b5abcc1ef27b90ede05cb9934df69b/src/game/client/components/maplayers.cpp#L811

I don't. I did it because yk I did everything differently and upload the envelope values to the gpu ^^

14:01

so I just did that as a why-not optimization :d

Patiga

I don't. I did it because yk I did everything differently and upload the envelope values to the gpu ^^

ah u still calc them on CPU? didnt even know, but probs makes sense considering it's a loop

yea, I don't think you can do that well on the gpu

14:05

I put some thought into it and it just didn't sound all that good

Cellegen

well it depends on which method is faster, discarding quads on GPU, or conditionally skipping quads on the CPU

only if you know that 1 gpu thread will process contiguous-by-predicate data

14:11

branch prediction powers it but not just random discarding

I suspect about an algorithm being ran on the GPU to select quads

does anyone know if quads are sorted inside a map? Or do I expect them randomly/the order the mapper added them?

14:12

probably the latter,

latter, ye

14:13

wdym by sorted btw

ofc, because the rendering order matters

ah this

sorted by x/y, in order to improve rendering (like I am thinking about)

hm you can rearrange them on client side

yes, but this would change the rendering order, so this isn't an option

14:15

this may work in most cases, but it breaks when quads in the same layer overlay each other

AssassinTee

yes, but this would change the rendering order, so this isn't an option

but you can index them, no?

oh god, @Patiga our np-problem is getting worse, if we allow overlays with different colors (edited)

happy

1

ah wait

hmm, you can get the most common color and overlay it with the rest of the pixelart tho

zhn

but you can index them, no?

yes, indexing would solve this, however this could run into other issues, e.g. caching

How do I move to the bottom line when writing motd?

This is a\nMulti line\nMessage of the\nDay!

1

Sooo, 97% (590/607)... Just a little and the PR is here! Melon, can you help here? Or Heinrich? Anyone? (edited)

Replying to egyt So, this is the list of strings I couldn't translate, some because I don…

louis

wow jupstar pro mapper

pro mapper requirements: 1. knows how to copy paste bush quads 2. knows how to rotate them

ravie speedrun any%

16:22

rip

rabies

egyt:

"Vote command:" "Vote description:"

both used when adding votes to the server via the UI

"Super"

used to indicate that a player has used the super command in the scoreboard, used instead of a team number (because internally, everyone using super joins the super team) (super makes you invulnerable to freeze tiles and gives you a similar thing to endless jumps)

"Save"

like file→save. currently only used for ghosts

"No results"

same as "Match %d of %d" that you alrready translated (edited)

I'm not Egypt, but ok...

You are egypt now, it's decided

No.

sorry, that's my fingers autocorrecting the words I type (edited)

OH, MATCH IS A VERB

Replying to @heinrich5991 egypt: […]

no, match is a noun

16:40

"[this is] match 5 out of 12"

well, noun that comes from verb

"[there were] no matches"

16:40

ah

Match as a result, yeah?

yes

Okaaay, time to fix

Match "as a result"? huh

16:42

Match as in "contest"~ is the meaning there

Won't understand

Replying to @Learath2 Match "as a result"? huh

16:42

???

Learath2

Match as in "contest"~ is the meaning there

are you sure?

heinrich5991

are you sure?

Match %d of %d?

yes

16:43

it's in src/game/client/components/console.cpp

Oh, I see. Yeah match as in "match result"

it's about rcon commands that match a search string, I think

16:43

English makes me sad

16:44

~~next time use Esperanto as the source language~~ (edited)

All natural languages have deficiencies because well they are mostly emergent, not of design

so, which "match" it is??? (edited)

result, something that matches

okaaay

english is particularly bad about pronunciation. I don't know any other language that is as bad

16:45

okay, maybe except japanese

Japanese isn't that bad

学校

Tough, though and plough are criminal

nothing about that tells you something abotu the pronunciation

16:47

you need to learn each symbol, and they might even have different pronunciations depending on context

Oh, that's what you mean, well yeah. But as soon as you know how it's written in hiragana it's mostly unique

yes, if you type it out in syllables, it's fine in japanese

heinrich5991

yes, if you type it out in syllables, it's fine in japanese

(There are a few sound changes here and there, e.g. は -> わ)

"Super" used to indicate that a player has used the super command in the scoreboard, used instead of a team number (because internally, everyone using super joins the super team) (super makes you invulnerable to freeze tiles and gives you a similar thing to endless jumps)

Now I wonder where "Alpha" is used

egyt

"Super" used to indicate that a player has used the super command in the scoreboard, used instead of a team number (because internally, everyone using super joins the super team) (super makes you invulnerable to freeze tiles and gives you a similar thing to endless jumps)

Now I wonder where "Alpha" is used

Alpha is always used as "Opacity"

16:49

Higher values being more opaque

Learath2

(There are a few sound changes here and there, e.g. は -> わ)

oh true. and へ → え

えい -> ええ is also very common, and the す losing it's "う" in です

Adding votes to the server via UI

Could you tell me more about this, please? (edited)

Replying to @heinrich5991 egypt: […]

egyt: log into a server via rcon and open the vote menu

How do I log into a server?

start a server locally and check what password it tells you in the console. enter that in f2

Good thing that I have two laptops, one with Windows and one with Linux...

17:42

By the way, I replaced "resources" (as "assets" translation) with "textures", is that fine?

egyt

Good thing that I have two laptops, one with Windows and one with Linux...

you don't need windows or linux for this, or two computers

you just need to press 1 button

egyt

By the way, I replaced "resources" (as "assets" translation) with "textures", is that fine?

stick to the common terminology, assets are assets, resources are resources, textures are textures

17:49

when someone plays with your translation, and we only really offer help in english, telling them to remove 4K assets, will tell them nothing if its translated to something else

but what are "assets" in ukrainian?

17:50

if the word that means "textures" is the best fit there, it'd make sense to translate like that

What's the difference between "resource" and "asset"? They are also synonymous in English

in german at least, I'd think "resources" would be a good translation for "assets"

17:54

seems like resource is also a word in ukrainian for that: https://uk.wikipedia.org/wiki/%D0%A0%D0%B5%D1%81%D1%83%D1%80%D1%81%D0%B8_(Windows)

Ресурси (Windows)

Ресурси Microsoft Windows — дані, вбудовані в EXE, DLL, CPL і (починаючи з Windows Vista) MUI -файли. Доступ до цих даних можна отримати через функції Windows API.

yeah it seems ассет or асет (asset) is translated to Ресурси (resource) in some contexts: that right egyt? (edited)

"ассет" or "асет"

checking wiktionary, these don't seem to be russian or ukrainian words

YEP

i'll drop out of this since i dont know ukrainian and i seem to google stuff wrong

Image attachment

"asset (noun)"

Yes ("Ресурси" is plural and "Ресурс" is singular) (edited)

Replying to @meloƞ yeah it seems `ассет` (asset) is translated to `ресурс` (reso…

18:06

That's russian

Replying to @meloƞ image.png

egyt

That's russian

a

And, while "Asset" also translates as "актив" (that one), I haven't seen it being translated like that in video game context even once (edited)

Robyt3

What's the difference between "resource" and "asset"? They are also synonymous in English

this might just be a personal thing, but to me assets are supposed to be edited, and resources are not (edited)

18:09

an asset would be a png, but a resource would be a GPU texture format or something

really, use Esperanto as a source language to avoid confusion like this one :^)

does esperanto not have these problems of homonyms?

but I'm not sure it matters in this context

https://esperanto.stackexchange.com/q/3446/5667

Are there any true homonyms in Esperanto?

Okay, I’m cheating a bit, since I already know about one true homonym in Esperanto: radio (meaning radio or ray). Do any others exist? I’m looking for single word roots with divergent meanings, so...

18:12

apparently esperanto also has homonyms

egyt: Time limit translation works for me, maybe you are running into the weird bug with the line reader. Or you accidentally changed the original translation somehow, like adding a space at the end

Image attachment

Image attachment

18:13

Also, why does it say "Finish time: <...> -0.05" when it is "-0.06"? (edited)

Replying to egyt зображення.png

you can report that as a bug

I checked it now and, well, I accidentaly removed the space at the start :^) (edited)

Replying to @Robyt3 egyt: Time limit translation works for me, maybe you are running into th…

18:17

The problem is, it sometimes happens, and sometimes it doesn't

Replying to @heinrich5991 you can report that as a bug

18:18

Also it happens when you arrived later than your best time

18:18

Image attachment

just report it with the screenshot, should be enough

[ddnet/ddnet] Issue opened: #8442 Differences in delta

For some reason, delta (time of the race-best time) in top right corner and under Finish time: in the middle differs from one in the chat (which shows the correct value) For example, here it shows -00.05, while the actual delta is -00.06 !зображення And there it shows +00.03, while the actual delta is +00.04 ![зображення](https://github.com/ddnet/ddnet/assets/113429882/752d3e89-0406-49e7-a92f...

Image attachment

Replying to egyt The problem is, it sometimes happens, and sometimes it doesn't

18:30

Here it shows correct delta both in chat and top right corner

Current translation, Belarusian translation and Wiki uses "Textures", so that's why I chose this one

Replying to egyt By the way, I replaced "resources" (as "assets" translation) with "textu…

care to link the wiki?

here (edited)

18:49

Also, it mentions "Тії" ("Tiyi") as the name for Tee tab

18:49

I don't really like it because it is not the original pronounciation (edited)

Replying to egyt Also, it mentions "Тії" ("Tiyi") as the name for Tee tab

18:52

but at the same time it differs from the word "ті" ("those"), so now I don't really know (edited)

18:54

I also saw "персонаж" ("character") being used, but it is a regular noun, and won't really fit here (see picture below) (edited)

18:55

Image attachment

18:57

so, either I stop with "Тії", or we have to change practically all of these pages

18:57

(please, make "Tee" tab translatable already! https://github.com/ddnet/ddnet/issues/8436) (edited)

19:02

Or should I open a separate issue?

Replying to egyt (please, make "Tee" tab translatable already!)

19:03

https://wiki.ddnet.org/wiki/Common_Terminology/uk did I struck gold?

no, that issue is fine

19:04

if you want to expedite the resolution of the issue, I suggest you create a pull request to make "Tee" translatable (edited)

Why translate tee

Replying to @heinrich5991 if you want to expedite the resolution of the issue, I suggest you creat…

MilkeeyCat

Why translate tee

e.g. because latin characters are usually not used in the language

because we actually have a translation used on Wiki

Replying to @MilkeeyCat Why translate tee

I find the wiki reason meh

Thanks god i never open those pages

we can also adjust the wiki instead

my lang got casts poggers

poggers

Image attachment

what's the result of casting -1 to u8?

19:07

is that a good result? ^^

It can parse casts but it cant print out stuff

19:08

Xd

^^

time to git clone again (edited)

Replying to egyt

The time has come either to add bool or 16 bit ints

19:09

Probably bool will be a better choice

egyt

time to git clone again (edited)

no need to delete the clone btw

And i redid all From impl into TryFrom justatest

justatest

heinrich5991

is that a good result? ^^

it kinda works ™ (edited)

Image attachment

MilkeeyCat

And i redid all From impl into TryFrom justatest

justatest

nice

but my compiler is so high tech that it cant represent -128 as 1 byte pepeW

pepeW

(edited)

Compiling time!!

Replying to egyt time to modify the source code again

[ddnet/ddnet] Pull request opened: #8443 Minor refactoring of CControls

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
[X] Considered possible null pointers and out of bounds array indexing
[X] Changed no physics that affect existing maps
[ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addres...

MilkeeyCat

but my compiler is so high tech that it cant represent -128 as 1 byte pepeW

pepeW

(edited)

off by one error somewhere?

Learath2

off by one error somewhere?

it's because of the way i handle integer literals

20:18

i have int_repr struct which stores unsigned value

20:19

and then it can be put in unary expr to make it negative

20:19

so first it's +128 which takes 2 bytes

20:19

and then it becomes negative

Well, it works, now it is localisable

Replying to egyt Compiling time!!

20:49

Image attachment

[ddnet/ddnet] Pull request opened: #8444 Make "Tee" tab name localisable

See #8436 for details

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
[ ] 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/ddne...

Here we are, having a nice number

20:57

Also, 10 strings left to translate!

Replying to egyt So, this is the list of strings I couldn't translate, some because I don…

21:00

So, do I need to add

Tee
==

into my local translation? (edited)

21:02

Or the next release will be without my translation for that tab?

yes

Oh, and also https://github.com/ddnet/ddnet/pull/8404

"Spectate" to "Spectating" in Spectator UI by EGYT5453 · Pull Reque...

Changing "Spectate" to "Spectating" in the bottom right corner when in Spectator mode so it makes more grammatical sence Before: After: Checklist Tested the change ingame P...

21:03

Informative

Replying to @heinrich5991 yes

21:08

Image attachment

21:08

hehehehe

Huh, we actually have a word "форсувати" for that (edited)

github-merge-queue[bot]

[ddnet:master] 2 new commits

950bbf3 Make "Tee" tab name localisable - EGYT5453 82c2ecd Merge pull request #8444 from EGYT5453/tee-tab-l10n - heinrich5991

21:28

[ddnet/ddnet] Pull request closed: #8444 Make "Tee" tab name localisable

So, again, should I go into translation files of the repo to add

Tee
== Тії

so my translation for Tee tab name will be in the next release? (edited)

Tii?

https://wiki.ddnet.org/index.php?search=тії

Replying to @Learath2 Tii?

Huh, I did not know we even translated that one

Because you never did

Replying to @Learath2 Huh, I did not know we even translated that one

21:46

I made a PR to make the "Tee" (tab name) translatable literally less than an hour ago (edited)

21:50

I guess I'll have to ask "tomorrow"

Replying to egyt So, again, should I go into translation files of the repo to add […]

I forgot to restart my monero node after they replaced my hdd, now I fell so far behind that it'll take me days to resync

egyt

I guess I'll have to ask "tomorrow"

why not PR your entire thing at once?

It might be time to get a server upgrade too ngl, spinning rust is getting on my nerves lately

Learath2

It might be time to get a server upgrade too ngl, spinning rust is getting on my nerves lately

https://www.youtube.com/shorts/nbVSITerFkE

10 Years At Netflix

what do you mean by "entire thing"?

Replying to @meloƞ why not PR your entire thing at once?

the entire translation .txt, doesnt seem profitable to make a PR for a single translation :P

egyt

So, again, should I go into translation files of the repo to add

Tee
== Тії

so my translation for Tee tab name will be in the next release? (edited)

why are you translating Tee again

22:05

the players know what tees are

22:05

just like they know what DDNet is

Because I found the term that is used on Wiki

Replying to @Ewan why are you translating Tee again

i don't necessarily think this is the correct way to do that

Having "тії" is better than having "тее", change my mind

Tee is a identity, not something you translate

but i'd consider Tee to be a proper noun, wouldnt really translate that (edited)

"Combine" is also a proper noun, but it didn't stop translators

that’s because it’s critical to the HL2 plot and wouldn’t make sense otherwise

Slavic languages love transliterating proper nouns. Макдоналдс

^

i don’t think that makes it correct

Старбакс

harry potter is also transliterated in russian

pepe_holy

2

also, um, it is not finished yet. Again, 10 strings, and I don't know where they occur, meaning I can't really make an accurate translation

Replying to @meloƞ the entire translation .txt, doesnt seem profitable to make a PR for a s…

heinrich5991

harry potter is also transliterated in russian

And they use a г so it’s garry potter

I wouldn't say it's garry potter

egyt

So, this is the list of strings I couldn't translate, some because I don't know where they occure, and some because I don't know technical terms behind them

(paused)
All combined
Folder Link
Follow
Game paused
Grabs
Moved ingame
Replace video
Searching

[Graphics error]
An error during command recording occurred. Try to update your GPU drivers.

(edited)

are these the missing ones?

just pronounced as such

yes, those are the ones I haven't translated yet

Replying to @meloƞ are these the missing ones?

heinrich5991

just pronounced as such

Well if a russian person only speaks russian, for all they know it might be garry everywhere

I guess

22:13

but it's also gitler, so it's a consistent transcription, at least

I do wonder why leading h becomes g in russian though

22:14

They do have the sounds for it х sounds much closer to my ear atleast

Learath2

I do wonder why leading h becomes g in russian though

probaly cuz russian language developed without the "h" sound (edited)

literally 10 strings (or 13, if including PRs, who knows), and the translation is done

Replying to egyt yes, those are the ones I haven't translated yet

22:21

btw, should I try again adding context to "Name" in Server browser?

22:22

oh wait it hasn't been fixed yet

hasn't been fixed yet

can someone try and reproduce crashing the client by rendering a demo, setting the playback time to 0.1 - forwarding a bit and then quitting out of it without finishing rendering using the X on the bottom right? - my client became unresponsive for a solid minute, outputted me this:

2024-06-04 00:18:38 I demo_player: Stopped playback
2024-06-04 00:18:38 I demo_player: Loading demo 'demos/replays/asdasd.demo'
2024-06-04 00:18:38 I datafile: could not open 'maps/Adrenaline 3.map'
2024-06-04 00:18:38 I videorecorder: Recording to 'videos/asdasd.mp4.mp4'
2024-06-04 00:18:59 I client: disconnecting. reason='unknown'
2024-06-04 00:18:59 I video_recorder: ------------
2024-06-04 00:18:59 I video_recorder: ------------
2024-06-04 00:18:59 I demo_player: Stopped playback

and then closed (using DDNet 18.2) this is the output on latest upstream:

2024-06-04 00:22:47 W videorecorder/libav: Too many bits 32768.000000 > 12288 per frame requested, clamping to max

2024-06-04 00:22:47 I videorecorder: Recording to 'videos/aseasdasdasd.mp4'

2024-06-04 00:22:51 I videorecorder/libav: frame I:114   Avg QP:14.91  size:177571

2024-06-04 00:22:51 I videorecorder/libav: frame P:412   Avg QP:24.00  size: 10011

2024-06-04 00:22:51 I videorecorder/libav: frame B:831   Avg QP:25.97  size:  3511

2024-06-04 00:22:51 I videorecorder/libav: consecutive B-frames: 12.7%  9.6% 22.3% 55.4%

2024-06-04 00:22:51 I videorecorder/libav: mb I  I16..4: 32.8% 25.2% 41.9%

2024-06-04 00:22:51 I videorecorder/libav: mb P  I16..4:  0.7%  0.4%  0.5%  P16..4:  7.7%  2.8%  2.0%  0.0%  0.0%    skip:85.9%

2024-06-04 00:22:51 I videorecorder/libav: mb B  I16..4:  0.3%  0.0%  0.0%  B16..8:  7.4%  1.6%  0.5%  direct: 1.3%  skip:88.7%  L0:51.0% L1:37.3% BI:11.7%

2024-06-04 00:22:51 I videorecorder/libav: 8x8 transform intra:24.9% inter:51.0%

2024-06-04 00:22:51 I videorecorder/libav: coded y,uvDC,uvAC intra: 30.4% 34.5% 30.3% inter: 2.5% 1.3% 0.7%

2024-06-04 00:22:51 I videorecorder/libav: i16 v,h,dc,p: 77% 21%  2%  0%

2024-06-04 00:22:51 I videorecorder/libav: i8 v,h,dc,ddl,ddr,vr,hd,vl,hu: 72%  4% 24%  0%  0%  0%  0%  0%  0%

2024-06-04 00:22:51 I videorecorder/libav: i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 41% 30%  7%  3%  3%  4%  3%  4%  4%

2024-06-04 00:22:51 I videorecorder/libav: i8c dc,h,v,p: 68% 18% 12%  2%

2024-06-04 00:22:51 I videorecorder/libav: Weighted P-Frames: Y:0.0% UV:0.0%

2024-06-04 00:22:51 I videorecorder/libav: ref P L0: 60.3%  2.7%  9.1% 27.8%

2024-06-04 00:22:51 I videorecorder/libav: ref B L0: 73.7% 21.5%  4.8%

2024-06-04 00:22:51 I videorecorder/libav: ref B L1: 92.7%  7.3%

2024-06-04 00:22:51 I videorecorder/libav: kb/s:9651.27

2024-06-04 00:22:51 I videorecorder/libav: Qavg: 65418.816

2024-06-04 00:22:51 I demo_player: Stopped playback

22:24

dont know if its my fault or actual issue - so i didnt immediatly make one

waiting then

Replying to @heinrich5991 hasn't been fixed yet

22:26

as well as for #8404

changing Spectate to Spectating means you have to change the string in every translation.txt file, so the PR is incomplete it seems?

22:28

nvm deen already said something about it

no, that cna be handled differently

https://blog.rust-lang.org/2024/05/17/enabling-rust-lld-on-linux.html

Faster linking times on nightly on Linux using rust-lld | Rust Blog

Empowering everyone to build reliable and efficient software.

22:58

This feels very overdue

Exported 539 message(s)