DDraceNetwork - developer

DDraceNetwork

Development / developer

Development discussion. Logged to https://ddnet.tw/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 2020-06-20 00:00:00Z and 2020-06-21 00:00:00Z

[ddnet:master] 2 new commits

91c44f1 Sixup needs it's own maps - Learath2 9f53623 Merge #2285 - bors[bot]

07:29

[ddnet/ddnet] Pull request closed: #2285 Sixup needs it's own maps

Has anyone looked into the horrendous map download speeds in 0.7? It seems our old fastdownload hack was better

Up to 0.7.4 it's only twice as fast as 0.6 by default. 0.7.5 increased this to x8, but the whole thing still heavily depends on latency

@redix can you point me towards the change?

There is sv_map_download_speed. In 0.7.5 it was just increased to 8 (https://github.com/teeworlds/teeworlds/commit/1b0e9abd41aaeb4ed8372f7fb6680984aaafa358)

it set a configuration variable

@heinrich5991 I couldn't figure out why cmake won't update git_revision.cpp on the actual servers btw

It does on my computer

@Learath2 from a quick glance, the problem seems to be that someone build an in-tree version

08:57

I wonder how this hasn't fallen apart yet

08:57

I guess I'll add an hard error when cmake detects this

[ddnet/ddnet] Pull request opened: #2286 Times are no longer negative in 0.7

@heinrich5991 huh, you sure that's the problem? it'd be odd to have that happen on every server

hm, no doesn't seem to be the problem

Oh, but deen might be scping the src folder over

still odd that we have two times the generated files

09:03

WARNING: The file has been changed since reading it!!!

09:03

@Learath2

09:03

you also on ger?

ger2

*ger2

09:04

yes

I'm done with it, just reverted back to 7d428676e296cf2f50aea9feb3bf435cc6490c46 until the times are also in

[ddnet/ddnet] Pull request opened: #2287 Fix `git_revision.cpp` updating for out-of-tree builds

[ddnet:master] 2 new commits

6db72b2 Times are no longer negative in 0.7 - Learath2 14e6227 Merge #2286 - bors[bot]

09:19

[ddnet/ddnet] Pull request closed: #2286 Times are no longer negative in 0.7

@Learath2 let's put GER2 on master again and try sixup?

09:24

Done

[ddnet/ddnet] Pull request closed: #2287 Fix `git_revision.cpp` updating for out-of-tree builds

09:25

[ddnet:master] 2 new commits

6f3a7a6 Fix git_revision.cpp updating for out-of-tree builds - heinrich5991 ff705a3 Merge #2287 - bors[bot]

Nice, seems to work

09:27

Let's update the rest of the servers when you feel like it's tested enough

Okay first bug, ninja seems to make you invisible

ddnet 0.7 maps must have the 0.7 layer compression in them, no?

@Learath2 is it because armor/health==0?

10:01

had a similar problem when converting fng

0.7 cant do call votes

10:02

after trying to call vote

10:02

https://i.imgur.com/XcjYmH7.png

Embedded image

10:05

oh, call votes crash the entire server

rip

emotes dont appear on 0.7

10:19

ok, just one of the emotes work idk why

10:24

if you type in chat /emote happy 1, then the ooop! emote works but the rest doesnt. the ooop emote applies to all emotes, doesnt matter which emote you pick, it always shows the ooop! emote

okay let me revert this once more

10:27

@noby when the ninja timer runs low the character becomes visible again

10:27

so odd

10:27

The client seems to think that the character died

had a bug on .6<->.7 fng where ur tee would disappear for u when u got shot

10:28

fixed it by setting the health/ammo to nonzero if sixup and frozen

10:28

near the very end of ccharacter::snap

    if (m_FreezeTime > 0 || m_FreezeTime == -1)
    {
        if (Server()->IsSixup(m_pPlayer->GetCID()))
            pCharacter->m_Health = pCharacter->m_Armor = 1;
    }

10:29

probably shitty fix but hey it worked

10:29

maybe u can take this info and make a proper fix from it

oh

10:30

we send negative armor?

10:30

wtf

that would probably cause invisible tee too lol

Why do we send negative armor even? some hack for encoding times?

where do u send negative armor i havent seen this actualy

m_Armor = 10 - (NinjaTime / 15);

10:32

for sufficiently large ninjatime this is negative

ohh wow

10:32

turns out i have this code too o.o

10:32

but changing whats sent in snap works around it i guess

Yep that fixes it

hehe

ill be interested to see how yall properly fix it

need to find a better way to map Ninjatime to 10 armor

10:34

just a little math

10:35

although honestly I wonder which clever 0.7boi thought he was the cleverest when making negative armor == death

it's just being validated I guess?

@Learath2 you dont need the armor indicator for freeze/ninja in 0.7

10:38

0.7 has its own ammo for ninja

10:38

its a bar

10:38

. (edited)

Image attachment

@Xusep m_Armor = absolute((10 - (NinjaTime / 15)) % 10); Is what I went with to keep things the same between 0.6 and 0.7

absolute doesn't really make sense, I think

10:39

clamp would be more useufl

It's like a countdown that keeps looping until the last round

u tagged the fake noby

10:39

lol

that's how it used to work anyway

10:40

If we are changing that I'd rather map it properly to 10 shields

whats wrong with just sending a different value if 0.6 vs 0.7

hello?

10:40

just use the bar for 0.7

Like m_Armor = ((float)NinjaTime/NinjaDuration) * 10;

10:41

Hello, 0.6 clients can still spec 0.7 clients and they need the duration aswell

10:41

cleverboi

@Learath2 are you sure we kept looping?

Hm, let me double check

I think 0.6 clamped it in the client

https://github.com/ddnet/ddnet7/blob/master/src/game/server/entities/character.cpp#L1082

DDNet port to TW 0.7. Contribute to ddnet/ddnet7 development by creating an account on GitHub.

lol i am noby

@heinrich5991 yep you are right

10:42

I guess a proper mapping to 10 shields is better

10:44

Ah I see what we were doing with clamping

10:44

yeah that looks better

[ddnet:master] 2 new commits

c53528b Allow spectators to start votes - def- 4a81009 Merge #2284 - bors[bot]

10:52

[ddnet/ddnet] Pull request closed: #2284 Remove sv_spectator_votes

10:56

[ddnet/ddnet] Pull request opened: #2288 Fix snapping for ninja on 0.7

[ddnet/ddnet] Pull request opened: #2289 Fix votes called from 0.7

We can keep testing after those get merged

@Learath2 this also allows for the ninja bar in 0.7?

yes heinrich

@heinrich5991 about #2281 https://devblogs.microsoft.com/oldnewthing/20061108-05/?p=29083

How do I test that return value of ShellExecute against 32? | The O...

We discussed earlier the history behind the the return value of the ShellExecute function, and why its value in Win32 is meaningless aside from testing it against the value 32 to determine whether an error occurred. How, then, should you check for errors?

11:31

I guess we can just cast the result to intptr_t and not rely on implementation defined behaviour

"We want the general concept, judging from Discord. :)"

11:31

what does this mean

As in we want some sort of learn link

that the idea of linking the wiki from ingame is wanted

[ddnet:master] 2 new commits

8d422df Fix snapping for ninja - Learath2 c464e3e Merge #2288 - bors[bot]

11:35

[ddnet/ddnet] Pull request closed: #2288 Fix snapping for ninja on 0.7

does this work (int)(uintptr_t) or do i have to wrap it with ()

should work like that, not sure if the double cast is absolutely necessary, you can probably just cast to an (intptr_t) and compare that against 32

11:37

I wonder who even has 16bit code

the docs say you can only compare this as an int

I guess the docs are more authoritative, just cast it twice then

the security warning is not stern enough for me

11:44

passing user input to this function gives you arbitrary code execution

[ddnet/ddnet] Pull request opened: #2290 Restore old message behavior when using /load and /save

Only notify the player initiating /load about the error and don't send 'Loading initiated by ...' and 'Saving initiated by ...' messages resulting in two messages per /load and /save. Not sure if the old behavior is better or if we want to keep the new with sending two messages per /save and /load.

@heinrich5991 can you just tell me a remark that will make u satisfied

11:53

so we dont waste more time

from the comment: "this may not be called with untrusted input or it'll result in arbitrary code execution."

i rebased

[ddnet:master] 2 new commits

07f0239 Add Learn button which opens the ddnet wiki in the browser. - edg-l 4edac31 Merge #2281 - bors[bot]

12:06

[ddnet/ddnet] Pull request closed: #2281 Add Learn button which opens the ddnet wiki in the browser.

12:10

[ddnet/ddnet] Pull request opened: #2291 Make PlayerResult messages an union variant

and some small style and readability fixes in sql_score.

[ddnet:master] 2 new commits

b032edd Fix votes called from 0.7 - Learath2 07f6e71 Merge #2289 - bors[bot]

12:34

[ddnet/ddnet] Pull request closed: #2289 Fix votes called from 0.7

@Learath2 github cog is live now and cleaned up

12:36

$build_status #2281

https://github.com/12pm/ddnet-discordbot/commit/72ba37f1aae2bafe214195f25621e20c74a67a83

cleanup github cog · 12pm/ddnet-discordbot@72ba37f

nice pic

ty

12:37

#2291

https://github.com/ddnet/ddnet/pull/2291

Make PlayerResult messages an union variant by Zwelf · Pull Request...

and some small style and readability fixes in sql_score.

\o/

12:39

teeworlds#1234

12:39

teeworlds/teeworlds#1234

https://github.com/teeworlds/teeworlds/issues/1234

Message/sound on joining a server is missing · Issue #1234 · teewor...

I'm noticing that on joining a server there is no message anymore that the spectator group was joined and due to this the related sound is not played. It was working in Teeworlds 0.6.2 so t...

heh

uh

12:40

oh right it needs owner

12:40

if its not from ddnet

[ddnet:master] 2 new commits

673f59d Restore old message behavior when using /load and /save - Zwelf a3ab572 Merge #2290 - bors[bot]

12:41

[ddnet/ddnet] Pull request closed: #2290 Restore old message behavior when using /load and /save

ddnet-web#1

https://github.com/ddnet/ddnet-web/issues/1

[FEATURE] Sub-menu in ranks to filter by sub-difficulty (stars) · I...

[ddnet/ddnet] Issue opened: #2292 Learn button blocks client

on Linux. Chromium is set as my default browser, I don't have it running. So the new chromium process will stay in foreground, thus DDNet client is blocked. @edg-l

12:47

[ddnet/ddnet] Issue opened: #2293 Learn button doesn't fit

!screenshot_2020-06-20_14-45-22 I'll try to fix it

12:51

[ddnet:master] 3 new commits

67643e8 Make PlayerResult messages an union variant - Zwelf 1e4ccd7 Small style and readability fixes in sql_score - Zwelf 17e77d7 Merge #2291 - bors[bot]

12:51

[ddnet/ddnet] Pull request closed: #2291 Make PlayerResult messages an union variant

12:57

[ddnet/ddnet] Pull request opened: #2294 Make the learn button fit

!screenshot_2020-06-20_14-57-03

13:00

[ddnet/ddnet] Pull request opened: #2295 Open links in background (fixes #2292)

Only tested on Linux

[ddnet/ddnet] Pull request opened: #2296 Store saves client-side in ddnet-saves.txt

Only appends saves, just in case you forgot where you saved. Unfortunately only stores your own name, not your entire team.

@Learath2 On Sunny Land I noticed that graphics are borked, just at the start already

Image attachment

kek

@Learath2 and the finish time shown is also wrong

14:05

it's supposed to be 60:46, not 00:03

Imagine Sunny Land was released as Novice map afacepalm

afacepalm

@deen Well the map I load as is from servers7/maps, I'd guess it's broken there too

oh no

can i still glitch through kills on 0.7 monkaS

monkaS

yeah

feelsamazingman

1

which converter tool is used?

@Patiga the one in the repo iirc

14:12

afaik it only embeds images, doesn't do anything else

yes

14:12

turns out images for tiles need specific sizes in 0.7, while 0.6 doesn't care

14:13

so I'll need to pad the images that are used for tiles, hm

and I need to figure out how the times are actually supposed to be calculated

14:14

I actually don't know how we store the times tbh

emotes still broken feelsbadman

feelsbadman

so actually it's not just pad, but you have to scale the image. I hope no other maps use this, don't want to adapt my script. I'll check if I can find other maps

[ddnet/ddnet] Pull request opened: #2297 Ingame time is in milliseconds on 0.7

@deen [5:31 PM] redix: in theory it should be enough if width and height are divisible by 16

tthanks

[4:34 PM] redix: that would work but maybe something in the implementation requires 1024*1024

[ddnet/ddnet] Pull request opened: #2298 Log save/load in teehistorian

I didn't write tests yet.

how do you join team on 0.7? (edited)

15:12

cuz /team switches to teamchat

[2020-06-20 17:13:45][server]: shutdown from game server (sql open wordlist error) │························· [2020-06-20 17:13:45][sql]: SQL connection disconnected │························· [2020-06-20 17:13:45][sql]: SQL connection disconnected Happend when I try to start ddnet6 srv

15:14

(The server is crashing)

using /team

lol

2

@murpi If you use vanilla client you have to write first /team then press esc and then type your team number

15:18

and then go for it :D

15:18

or /team <number> and then esc

ah, ok. thanks

timeout protection for 0.7 doesnt work

15:28

shows unable to connect to the server after trying to rejoin the server

15:31

also trying to use /timeout (code) on regular 0.6 client causes connection problems

oh I should probably send a RemoveCommand for /team

@Trafalgar Law need a wordlist in data/ since a recent commit

15:44

we should ship it though(?)

That means?

https://github.com/ddnet/ddnet/pull/2247/files#diff-a9fdcdb86ce6db757bee21b305d723bd

15:46

you need that data/wordlist.txt fiel

holy jesus, 0.7 timeout protection is really broken

Should probably just disable it for 0.7 clients

@heinrich5991 And where do I get this file? :O

It should be there in the source

15:48

yep it's in data/wordlist.txt

Let me look :X

@Learath2 can you test timeout protection with me?

True its theire

15:50

But why the srv can't start then?

We have a rather large issue with the gametimer, we use the warmup timer since necropotame changed it but we can't really use the warmup timer for 0.7 nor can we go back to the gametimer as that'll break animations

15:51

what do?

15:51

@murpi

15:51

sure

@Learath2 can we use the race time from 0.7?

@Learath2 find me and join with 0.7.4 client

@heinrich5991 is there really a racetime? I see support for it in score but that's it

15:53

@murpi can it be 0.7.5 instead?

datasrc/network.py: NetTick("m_RaceStartTick"),

uh, yeah try

@Learath2 ^

Blind I am

Well, @heinrich5991 do you knwo why he can't open then the file if its exist?

is it in the data directory the server uses?

15:56

(watch the messages in the very beginning with $APPDIR and similar)

The issue with timeout seems to be that the 0.7 client doesn't notice that the local client id has changed

15:58

not quite sure how to convey that to it

@Learath2 only possible with fakeids

Yeah, just found out

but then then every 0.7 client always has to think itself is id 0

Timeout protection, atleast the hacky way we do it is dead

ye

It now needs to happen very early in the connection

16:37

(which is the correct place for it, but alas)

yea

Okay, I'll disable it for 0.7 clients, but it's rather easy to fix for custom clients

16:41

we can later add an extended message to let custom clients enable timeout protection

[ddnet/ddnet] Pull request opened: #2299 Disable timeout for 0.7 clients

Turns out there is no way to influence m_LocalClientID anymore. Another thing changed for the sake of change.

Why do we even change client ids?

[ddnet:master] 2 new commits

28905c4 Disable timeout for 0.7 clients - Learath2 5ca6e9c Merge #2299 - bors[bot]

17:24

[ddnet/ddnet] Pull request closed: #2299 Disable timeout for 0.7 clients

@fokkonaut I'm sure east and @deen had a reason to do it this way around when they implemented it

[ddnet:master] 2 new commits

865afa6 Ingame time is in milliseconds - Learath2 4dd72df Merge #2297 - bors[bot]

17:31

[ddnet/ddnet] Pull request closed: #2297 Ingame time is in milliseconds on 0.7

Did east go to the west or where did he go?

east and hmh are both in tahiti

really?

probably not

...

Haven't heard of either for months now

they moved on from this kid game

17:59

poggers

I'm having trouble deciding what to do

Why

$build_status

@Learath2 why is it red?

mac infrastructure problems in the github build i think

18:30

at least that's what I saw in my PRs

https://github.com/ddnet/ddnet looks green to me

DDraceNetwork, a cooperative racing mod of Teeworlds - ddnet/ddnet

on github all checks are green

18:30

Odd

are there docs for the github api?

18:38

with all possible values?

Yep

18:45

https://developer.github.com/v3/

18:45

This is what I referred to

18:46

$build_status #2297

Did you remove the header maybe? the checks feature is marked experimental

18:46

$build_status #2296

nah then it would say commit not found

18:48

oh i think i found the issue

18:48

i reversed the check for failed

18:49

$build_status #2296

now it should be right xd

$build_status

@Learath2 dont send 10 stars to the client when shooting pistol

19:21

when it collides

19:21

teh client doesnt use the angle you send to it

19:21

in 0,7

19:21

also, 0.7 clients dont need the broadcast timer, they have timer and finish netmsgs

19:23

and you shouldnt send 3 single stars for freeze, instead 10 with one object

The timer I'm going to pull request

19:24

Merging the damage indicators together is hardly simple considering I see one at a time

then pass a sixup amount of stars to the function

19:24

and handle both seperated

eh? damageinds get translated to damage messages

(that way you could also only send one single star for the pistol)

a simple one to one mapping

yeah, i mean dont "translate" them, just create one event for sixup clients and one for non-sixup

Can't, there is one array of events that's sent to every client

latest api uses graphql btw

of course you can

19:26

in eventhandler::snap

19:26

you can decide

Please enlighten me than, I don't see it happening without some massive hacks

lol

Or shoving everything in there and sorting it out later, which is wasteful of the space in the events array

[ddnet/ddnet] Issue opened: #2300 double hook sounds player attach

dont send the hooksound player attach from character.cpp to sixup clients, they handle hooksound based on coreflag_hook_attach_player clientside, so we get two sounds

I guess an especially unelegant way would be to just double the size of the event storage

could do that (edited)

In the end we're just running 2 servers in one with a translation in between

[ddnet/ddnet] Pull request opened: #2301 map_convert_07: Check image dimensions for divisibility by 16

As noticed on Sunny Land.

[5:31 PM] redix: in theory it should be enough if width and height are divisible by 16

Now running this over our maps and fixing them manually.

@deen https://discordapp.com/channels/252358080522747904/293493549758939136/723910935772659714

yeah, I'll try to replace in a few maps and see if required

20:05

about 50 maps affected btw

I don't really see a good solution for the damage indicators tbh, they work like this is there really a reason to change them up?

they look super ugly

20:22

shoot a wall

20:22

then youll notice its a random side

20:22

and the Freezer stars a basically too close to each other i believe

It'll always be a random side, no directions anymore

20:23

Well not really random, the bottom left third to begin wiht

20:24

Freeze stars are indeed too close but is it really that important?

20:26

I guess doubling MAX_EVENTS is only an 18kb per server increase in memory use

sounds ok

[ddnet/ddnet] Issue opened: #2302 ctrl-s not working in editor

[ddnet/ddnet] Pull request opened: #2303 map_extract: Only access map info if it exists

Noticed on map Paranormal

[ddnet/ddnet] Issue opened: #2304 Client side word filter list

I don't know if it may happen, but if the game grows, and maybe someone streams it, you may not want to get people banned because racial slurs being seen on the chat by a troll.

[ddnet/ddnet] Pull request opened: #2305 Version 14.0

Waiting for https://github.com/ddnet/ddnet/pull/2303 https://github.com/ddnet/ddnet/pull/2301 https://github.com/ddnet/ddnet/pull/2296 https://github.com/ddnet/ddnet/pull/2295 https://github.com/ddnet/ddnet/pull/2294 https://github.com/ddnet/ddnet/pull/2280 https://github.com/ddnet/ddnet/pull/2279 https://github.com/ddnet/ddnet/pull/2242 and probably some sixup fixes by Learath2 :)

[ddnet/ddnet] Pull request opened: #2306 Make ctrl-s/ctrl-l etc work everywhere in editor (fixes #2302)

instead of just in the layers view

Exported 328 message(s)