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 2020-12-29 00:00:00Z and 2020-12-30 00:00:00Z

@deen i think the cl_showdistance is almost twice as wide and high as the fov

for envelop animation on vanilla, we just fmod the time with envelop loop length to avoid large float/int conversion and division instead of changing precisions.

i just removed roundstarttime accidentially

01:59

floats stay inconsistent, even if u only use small floats

02:05

much more horrible is that i found a GL 3.3 bug while debugging this issue xd even tho its just a "renders too few things" bug, not critical

[ddnet/ddnet] Pull request opened: #3455 Readd round start tick to the calculation

fixes #3454 @bretonium would you be so kind and test this quickly, if everything looks fine again? :D

Checklist

[ ] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Written a unit test if it works standalone, system.c especially
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[ ] Tested the c...

thonk

02:43

but it did work for synced animation with high tick numbers for vanilla.

[ddnet/ddnet] Pull request opened: #3456 Fix quad shader

OGL 3.3 only

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 if it works standalone, system.c especially
[ ] 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/#using-address...

@TsFreddie working != working accuratly

thonk

sure

also synced animations are more accurate anyway

02:45

bcs they recalculate the time every time

I thought the problem is they lag due to lack of precisions

yes, im not sure if ctf2 has synced envs, but i fixed it on ctf2 #3002

https://github.com/ddnet/ddnet/pull/3002

Make animations more consistent with high float values by Jupeyy · ...

fixes what is described in #2997 Old client animation when server time is huge: vid_old.zip client animation when server time is huge: vid_new.zip @Fireball-Teeworlds maybe u can try, if that fixes...

there is a video xd

ctf2 has synced snow for some reason

appearently not on 0.6?

02:47

just opened it in editor to look

Maybe it's a 0.7 thing

anyway

02:47

if u want to check my impl against 0.7 u need to simulate adding floats every time

02:48

e.g. adding a constant time that happens per frame to the float

02:48

and if they not the same after lets say 1h

02:48

then the floating point inaccuracy kicked in

02:48

my guess is, that 0.7 is around 100ms off after 1h

Nvm it was dm3

02:50

I means if all clients are 100ms off then it is still synced

02:51

it depends on the frame times probs

02:52

adding a float 1000 times instead of 60 times should make a relativly huge difference

02:52

depending on what u call huge xd

Oh you mean client animation?

02:54

Synced animations are always calculated iirc

yes

02:54

i mean non synced

02:54

thats why linked the issue

02:54

what

02:54

wait

02:55

no

02:55

we talked about the 1h thing

02:55

02:55

im confused now too xD

yeah non sycned for 1h

Ok

so yeah ctf2 snow should be synced i guess

Well if anyone want a playtime clock on the map I guess making it accurate is good.

true, it obviously doesnt matter too much

02:57

tw used this shitty implementation for like 10 years and nobody really cared

02:57

but i love accuracy

02:58

and so or so, CPUs dont even give accurate timers anyway

02:58

accurate in a matter of ACCURATE

One sec is localtime() busted

02:59

Oh 0.7 changed non synced animation completely I guess

well u will know better

03:02

i dont look into the code regulary

Non synced animation is just s_Time = Client()->LocalTime() on vanilla for some reason.

ddnet is as accurate as 0.0000000000000001 compared to what current pysics can currently meassure

03:03

so pretty bad xd

03:03

oh and *100 if we dont choose 1 as 100% (edited)

So you can't make a clock with non synced envelope at all on vanilla. Huh

isnt localtime just increasing value?

03:05

its a function i'd prefer to remove from ddnet, bcs of the reasons of inaccuracy

I see

anyway, its just me over dramatize

Well localtime is time_get() - starttime

03:10

Just checked

saved as float?

03:11

or as int?

It's a int64 arithmetic divided by time_freq.

it only matters what it is stored in

float

but u said s_Time = Client()->LocalTime()

03:12

so its fine anyway

03:12

the 0.6 code did not do that

thonk

I wonder why they changed it

bcs its more accurate

And probably less code to worry about

s_Time += pThis->LocalTime() - s_LastLocalTime;
            s_LastLocalTime = pThis->LocalTime();

03:13

that was the old ddnet version

I would imagine they care about simplicity more

i c

03:13

tho it also makes it pretty accurate, since its calculated every time freshly

But it does make sense. Why do you need to keep track of a non synced timer anyway

so that makes my claim wrong btw

So they just made non synced envelope purely for animation and nothing else then.

[ddnet-maps:master] 1 new commit

4ee3abf A Deino - ddnet-maps

my poor internet

Image attachment

[ddnet-web:master] 1 new commit

5bcf1d4 Add DB server sponsoring by KOMПOT - def-

@fokkonaut that's ok, people can just come flying in quickly and you still want to see them immediately

[ddnet:master] 2 new commits

54d937d Fix quad shader - Jupeyy 1211fcd Merge #3456 - bors[bot]

08:06

[ddnet/ddnet] Pull request closed: #3456 Fix quad shader

can i disable timeout protection on my own server and just disconnect/destroy timed-out player

conn_timeout_protection 0

thanks :)

Where is CNetObj_PlayerInput defined? Is there a macros that adds "PlayerInput" at the end or smth.? (edited)

datasrc/network.py, look for PlayerInput

10:28

You generally don't want to change the existing objects though, you can add your own NetObjectEx

ah thx

@Deleted User How risky are your bug fixes, should I ship them in next quick fix release without RC?

[ddnet:master] 2 new commits

30d089b Readd round start tick to the calculation - Jupeyy 262439b Merge #3455 - bors[bot]

11:47

[ddnet/ddnet] Issue closed: #3454 infclass class selector broken

11:47

[ddnet/ddnet] Pull request closed: #3455 Readd round start tick to the calculation

[ddnet-discordbot:master] 1 new commit

d688e84 update waiting maps timestamp properly and bump waiting duration to 60 days - 12pm

@deen yes add them

[ddnet:master] 2 new commits

59f1c2c Add New Year season theme by mind and use for 2 days - def- 5056563 Merge #3451 - bors[bot]

13:21

[ddnet/ddnet] Pull request closed: #3451 Add New Year season theme by mind and use for 2 days

@deen @Learath2 the performance sucks bcs fully transparent quads are drawn too

13:21

We could just skip them

sounds like too much work

13:27

at least for this release

[ddnet:15.2-devel] 4 new commits

93c1f37 Fix quad shader - Jupeyy e4def98 Readd round start tick to the calculation - Jupeyy d035825 Add New Year season theme by mind and use for 2 days - def- cbc45f7 Version 15.2.5 - def-

13:30

[ddnet:master] 1 new commit

78870a5 Version 15.2.5 - def-

@deen its 2 lines for non gl 3.3 and 4 lines for gl 3.3

[ddnet/ddnet] Pull request opened: #3457 Don't render fully transparent quads

@def- i mean u could test the fps diff, its quite huge and if u really only have 50fps, others will only have 30fps xd

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 if it works standalone, system.c especially
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[...

[ddnet:15.2-devel] 1 new commit

464d310 Don't render fully transparent quads - Jupeyy

13:40

[ddnet:15.2.5] 1 new commit

464d310 Don't render fully transparent quads - Jupeyy

@deen

13:42

i force pushed, can u get teh current commit instead?

13:42

for GL 3.3 its not that easy, bcs of the shader difference

I think I got the correct one

atleast in these github things it doesnt look like it

which things?

these popups

13:44

when clicking them

Ah, github wasn't updating your change for me

alright

Wait, so with OpenGL3 there will be no improvement? :/

yes, i dont want to risk too much coding rn

13:46

but with default settings its optimized

13:46

and most ppl probs play on default anyway

[ddnet:15.2-devel] 1 new commit

11c8300 Don't render fully transparent quads - Jupeyy

i'll patch it for gl 3.3 later, but it requires the correct quad offset

What was wrong with the change you made previously?

i mean i can patch it, if u want

13:48

takes me maybe 10 minutes, and maybe some testing quickly

then let's wait with the release I guess

cancel bors then?=

13:51

or new pr?

273k lines of code POGGIES

POGGIES

Image attachment

13:54

Industry average experience is about 1-25 errors per 1000 lines of code for delivered software.

13:54

according to this ddnet has from 273 to 6835 bugs

13:54

monkalaugh

@Ryozuki you should exclude the external dependencies probably

Image attachment

13:59

from 142 to 3568

That number of bugs estimation is pretty useless

14:13

do i have to try to keep the CID consistent to a character to stop them from jumping around?

[ddnet:15.2-devel] 1 new commit

3695469 Don't render fully transparent quads - Jupeyy

btw i'm manually snapping characters and players based on some fake dummy data. if it wasn't clear

@TsFreddie yes changing cids create these issues. The 64p patch has an algorithm to keep the cids stable-ish when mapping to vanilla

i'm not using that since i'm allowing people join without seeing each other, so i just hack their cid to be always 0, and 1 through 63 will be the dummies

14:22

i'll try something else then

[ddnet:master] 2 new commits

288a18f Don't render fully transparent quads - Jupeyy 1c750a7 Merge #3457 - bors[bot]

14:23

[ddnet/ddnet] Pull request closed: #3457 Don't render fully transparent quads

This is the first time in years that i'm making a server mod, where i can just mess up the code so bad and still don't need to worry about it.

14:25

feels good kek

kek

[ddnet-web:master] 1 new commit

7a5f0b0 Version 15.2.5 - def-

monkalaugh

messing up a mod is fun until you start to like the mod and put too much effort into it to rewrite xd

14:59

totally not talking out of expirience here

[ddnet/ddnet] Pull request opened: #3458 Respect possible particle size increase, of worst case rotation

Particle's achsis aligned bounding box can increase in size if the particle is rotating. Even if the particle images are all "pretty" round and so it wouldn't matter for default particles, its not correct and with a custom image it could break. following #3441 Basically this: !image

Checklist

[ ] Tested the change ingame
[ ] Provided screenshots if it is a visual chan...

[ddnet:master] 2 new commits

bce1f45 Respect possible particle size increase, of worst case rotation - Jupeyy b153164 Merge #3458 - bors[bot]

15:11

[ddnet/ddnet] Pull request closed: #3458 Respect possible particle size increase, of worst case rotation

[ddnet:master] 3 new commits

f1358b5 Same default map as in autoexec_server.cfg - def- ff2d2f6 Fix x of y servers - def- 871fd94 Merge #3448 #3449 - bors[bot]

15:21

[ddnet/ddnet] Pull request closed: #3449 Fix x of y servers

15:21

[ddnet/ddnet] Pull request closed: #3448 Same default map as in autoexec_server.cfg

[ddnet-web:master] 1 new commit

968ea64 Update stats/server: Remove RUS2, AUS - def-

[ddnet/ddnet] Issue opened: #3459 Freeview mouse should move faster when zoomed out

suggestion: add mod's name after map's name in .demo ctf5_xxxxxxxxxxxx_zCatch_xxxxxxxxx.demo (edited)

Multimap_xxx_CTF_xxx.demo

19:31

actually neat idea but im not sure if it just makes demo names longer without any additional information @gerdoe do u have a usecase?

19:32

in the demo browser u can see the gametype in the metadata anyways

19:32

might help identifiying vanilla vs zCatch demos faster

i don't see any metadata with gametype in ddnet client rn

Image attachment

oh wait im stoopid no nvm its not there xd

stoopedragon

:c

[ddnet/ddnet] Pull request opened: #3460 Update russian translation

[ddnet:master] 2 new commits

16e6ed8 Unify + typos - gerdoe-jr 9942507 Merge #3460 - bors[bot]

20:04

[ddnet/ddnet] Pull request closed: #3460 Update russian translation

look this gem i found in hackernews

20:13

https://www.bbc.com/news/technology-55475433

Brexit deal mentions Netscape browser and Mozilla Mail

Experts suggest officials saved time by copying and pasting text from old legislation.

20:13

References to decades-old computer software are included in the new Brexit agreement, including a description of Netscape Communicator and Mozilla Mail as being "modern" services.

20:13

Experts believe officials must have copied and pasted chunks of text from old legislation into the document.

>

The references are on page 921 of the trade deal, in a section on encryption technology.

20:13

hilarious

20:15

Image attachment

20:15

xd

20:16

1024 bit keys

20:16

in 2020

20:16

monkalaugh

20:18

they will sign this, thats the worst

20:18

The UK government has not yet responded to the BBC's request for comment.

apparently, tender was won by a self-taught crypto-boy

oh, that's he, actually

Image attachment

Your IP is: 127.0.0.1 . Hacked haha

hell yeah it is

"no, it's 192.168.1.1"

@Ryozuki: https://gitlab.com/ddnet-wiki/ddnet-wiki.gitlab.io/-/issues/3#note_475552788

Evaluate software choice and workflow (#3) · Issues · DDNet Wiki / ...

@Ryozuki Currently we make merge requests and let someone else merge it, but I am open to suggestion to...

Exported 204 message(s)