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 2020-08-25 00:00:00Z and 2020-08-26 00:00:00Z

[ddnet/ddnet] Issue closed: #2633 Have the server sort default to some combination of number of players and ping

05:37

[ddnet:master] 3 new commits

4abe8ca Sort servers by a combination of players and ping as default - trml d5c0300 Sort players by descending order first - trml 7d1ce2f Merge #2638 - bors[bot]

05:37

[ddnet/ddnet] Pull request closed: #2638 Sort servers by a combination of players and ping as default

05:44

[ddnet-web:master] 1 new commit

4e21bd1 Add donation - def-

[ddnet-web:master] 1 new commit

466167b Update highcharts and include in this repo - def-

Is anyone here familiar with crawling? I'm looking to crawl a website that doesn't really want it's user list public, so I need to crawl the internet for the links to user profiles

09:30

Though I'm not sure how best to do that

crawling internet uh oh

Yeah, tough job with millions of pages. So I was thinking of scraping google instead

I think google won't like if you do unto them as they do unto others

09:43

Anyway, if you have any idea how I could go about scraping large-ish parts of the internet

09:44

hmu

Is there a way to show what bind I have written for my keys in the console? Or only in config file

09:52

Smth like show bind 'x'

dump_binds x

Ty

[ddnet/ddnet] Pull request closed: #2637 Send larger distance when zooming

11:14

[ddnet:master] 2 new commits

d24dbd0 Send larger distance when zooming - def- 836863b Merge #2637 - bors[bot]

11:20

[ddnet:master] 2 new commits

d5045ce Enable "Switch weapon on pickup" by default - def- 5b8e8ba Merge #2631 - bors[bot]

11:20

[ddnet/ddnet] Pull request closed: #2631 Enable "Switch weapon on pickup" by default

[ddnet/ddnet] Pull request closed: #2639 Give a more instructive message on DNSBL ban

12:01

[ddnet:master] 2 new commits

d41a6e6 Give a more instructive message on DNSBL ban - def- 3d1bc41 Merge #2639 - bors[bot]

12:04

[ddnet-web:master] 1 new commit

0ea1922 Add static charts.js - def-

@Learath2 about the nick situation: currently, you'll get the steam name by default, it will be "shadowed" in the nick text fields

12:10

until you put someting of your own in there

12:11

I'd prefer to allow changing nicks for steam, I'm open to more intuitive UI. if you guys think we shouldn't allow to change nicks in the steam version, I'm also fine with that

Unchecked full screen option and came across this, haven't been able to replicate it though so might have been problem my end. Epilepsy warning.

I'm not quite sure what the best practice is

@Clavata is it also flashing if you don't record it with a camera?

couldnt get it to show up with obs

but you can also see the flashing without the camera?

yes

is ur gpu dying

wouldnt know

i mean have u noticed any weird behavior like that in other applications/games that arent tw

havent noticed anything else, and tw works still

ig not that then

I've had that bug as well, usually after using ScreenToGif and turning fullscreen back on

hmm, I was actually in the process of taking screenshots when i turned full screen off and got that

[ddnet/ddnet] Issue opened: #2641 Vote to move players to freeze instead of kicking?

Kasia suggests so that newcomers don't get kicked when they have trouble doing a part and are considered to be blocking. They can then still watch how the part is done when frozen.

isnt that sv_vote_pause

tee_thinking

2

Mb we should use steam name for dummy too, like [D] prefix on it?

[ddnet-web:master] 1 new commit

59ae7ca Don't need tooltip, add news - def-

13:48

[ddnet-web:master] 1 new commit

6df01a7 fix - def-

13:49

[ddnet-web:master] 1 new commit

6d1732a capital - def-

[ddnet-web:master] 1 new commit

9d4059d update news - def-

[ddnet-web:master] 1 new commit

f591efc Add 1422 - def-

[ddnet/ddnet] Issue closed: #2640 Enabling DNSBL banning should not ban IPs that are playing already, only new ones

[ddnet-web:master] 1 new commit

a0980f7 Forgot files - def-

[ddnet/ddnet] Issue closed: #2605 Use Steam API to set player name?

15:30

[ddnet:master] 5 new commits

3c6f058 Add SteamAPI + stub - heinrich5991 a1875e4 Rename libddnet-antibot.so back to libantibot.so - heinrich5991 5d83e91 Use player name from the Steam API as fallback - heinrich5991 bfe7b3c Use player name prefixed with "[D] " for the dummy name - heinrich5991 f2cb651 Merge #2615 - bors[bot]

15:30

[ddnet/ddnet] Pull request closed: #2615 Use player name from the Steam API as fallback

15:33

[ddnet/ddnet] Issue opened: #2642 Use Steam API to provide links to join to another player

We already support running DDNet ddnet:ger.ddnet.tw:8303. @heinrich5991 recently added Steam API support, so I think this can be extended more easily now: https://github.com/ddnet/ddnet/pull/2615

15:34

[ddnet/ddnet] Issue closed: #2641 Vote to move players to freeze instead of kicking?

[ddnet/ddnet] Issue closed: #2619 Mipmap generation for OpenGL <= 1.5

15:43

[ddnet:master] 8 new commits

c2ce132 Allow specifying an OpenGL version manually - Jupeyy 7450e35 Add OpenGL 4.x support & cleanup - Jupeyy a07e9ac Add OpenGL 1.x, 2.x, 3.x support (fixes #2619) (fixes #2607) - Jupeyy 5b9269f Check NPOT textures and stronger 1.x fallback (fixes #2630) - Jupeyy f04ec93 Use OpenGL 3.0 as default - Jupeyy 2c02314 Stronger 1.x fallback, if OpenGL 3.0 was invalidly created - Jupeyy 3117a93 Do not create 2D textures, if not required - Jupeyy 694db44 Merge #2598 - bors[bot]

15:43

[ddnet/ddnet] Issue closed: #2607 3D textures and 2D array textures

15:43

[ddnet/ddnet] Issue closed: #2630 Use Rescale function in the backend to convert image sizes to powers of 2

https://youtu.be/1U0BB2D7wx4

SI TE RES PIERDES#1--mi primer video__AR YV

wrong channel

yo teeworlds developers wassup

sup

15:50

ddnet*

15:50

this is not teeworlds

bruh

15:56

you use the donations to pay for yourself or what?

to pay servers

15:58

developers dont get money

see https://ddnet.tw/funding/

Time to trigger all devs

15:58

I love Python

good

What

me too

What

15:59

visible confusement

I love windows

@deen the announcement is wrong

16:00

it says its brutal

16:00

but its insane 1 map

16:00

Image attachment

WHO HAS THE NEW DD_NET ME IF

wat @AR YV

?

[ddnet/ddnet] Issue opened: #2643 Client segfault when trying to join teesmash mod

Client crashes after connecting (but you never get ingame) ``` Thread 1 "DDNet" received signal SIGSEGV, Segmentation fault. CCollision::Init (this=0x55555596a278 , pLayers=0x55555596a228 ) at ../src/game/collision.cpp:66 66 m_Width = m_pLayers->GameLayer()->m_Width; (gdb) bt #0 CCollision::Init(CLayers*) (this=0x55555596a278 , pLayers=0x55555596a228 ) at ../src/game/collision.cpp:66 #1 0x00005555557b1a10 in CGameClient::OnConnected() (this=0x555555969980 ) at ../src/game/client/g...

16:21

[ddnet:master] 1 new commit

1810393 Version 14.5 - def-

@deen did you see my comment about the steam version?

16:22

https://github.com/ddnet/ddnet/pull/2615#issuecomment-680118672

Use player name from the Steam API as fallback by heinrich5991 · Pu...

Fixes #2605.

[ddnet-scripts:master] 1 new commit

18de8f2 Try adding steam api - def-

I take this as a yes

is it possible for game layer to be null? xD (edited)

[ddnet-scripts:master] 1 new commit

bd95cc0 macosx build fix - def-

I mean do a print m_pLayers->GameLayer() to find out @Ryozuki

@heinrich5991 i guess checking that with gdb is not trivial? xD

@Ryozuki enter what I said in gdb

oh

16:28

$1 = (CMapItemLayerTilemap *) 0x0

16:28

ye

16:28

its null

you found the problem

well solving it is another story

16:28

it worked in previous ddnet versions

16:28

justatest

find out why it is 0 or bisect

damn

18:17

i cant reproduce it again with self compiling

18:17

but steam one crashes

18:18

oh

18:18

it crashes again

18:19

maybe i should have deleted build entirely

[ddnet-scripts:master] 1 new commit

0f53f73 More steam build fixes - def-

are you sure it's libsteam_api.dll not steam_api.dll?

18:23

@deen

yes

18:27

@heinrich5991 Also, on osx, should I put it in Frameworks or ddnet?

[ddnet/ddnet] Pull request closed: #2634 Use borderless by default on Mac (fixes #2632)

18:28

[ddnet/ddnet] Issue closed: #2632 Fullscreen is completely broken on macOS

18:28

[ddnet:master] 2 new commits

23a3d46 Use borderless by default on Mac (fixes #2632) - def- 955ccb9 Merge pull request #2634 from def-/pr-borderless-mac - def-

I don't know

18:29

you could run install_name_tool to find out

18:29

@Learath2 tested it on mac

@deen you want a path relative to @executable_path on mac usuallu

18:32

y*

?

18:33

@Learath2 we have the other dylibs in Frameworks, where should I put the steam api dylib?

Same place

thanks

Might need to mess around with install name tool to get it to link properly and to the correct dylib

Did you test that the non-steam mac build still starts? https://ddnet.tw/downloads/DDNet-14.5-osx.dmg

[ddnet/ddnet] Pull request closed: #2627 Fix GUI Icons for OpenGL <= 1.5 (fixes #2619)

18:36

[ddnet/ddnet] Pull request closed: #2616 Add localization support for server messages

18:37

[ddnet:14.5] 2 new commits

23a3d46 Use borderless by default on Mac (fixes #2632) - def- 955ccb9 Merge pull request #2634 from def-/pr-borderless-mac - def-

@deen the nonsteam version seems broken

@heinrich5991 wait, I can't update to 14.5 on Linux. We don't have a way to download so files

Dyld Error Message:
  Library not loaded: /home/*/libsteam_api.dylib
  Referenced from: /private/var/folders/*/DDNet.app/Contents/MacOS/DDNet
  Reason: image not found

Do we really need to link against the stub? Why not just #ifdef it out?

makes it easier to test

18:39

one less configuration

We could do a really hacky nginx hack to serve the correct .so file depending on the client name

@deen seems it linked with an absolute path for the non steam version

does cmake rpath work on macos?

18:42

perhaps it's as simple as removing the if(PLATFORM STRNEQ mac) check

we have it working for libfreetype and libsqlite

Seems the steam version is also linked with an absolute path

so if you do it the same way as those libs, I can put it in ~/Frameworks

18:42

in Frameworks

SDL also looks off to me, you almost always want to link to @executable_path/../Frameworks

18:43

For the bundled release

I didn't change anything there

i think https://github.com/ddnet/ddnet/commit/4da0592d78ed57b15738cc6cc63ce5e59723f070

Replace Kobra 4 and Goo! with Gold Mine · ddnet/ddnet@4da0592

and add maps7 version

18:45

is the first bad commit

Sorry, I meant @loader_path/../Frameworks

@Ryozuki I doubt it

hmhmhhmh

Anyway, just linking to @rpath should also work as we have @loader_path/../Frameworks as LC_RPATH

ok wait

Could you provide a fix? I'll be afk a bit

I can try, but you are crosscompiling, I'm not. You might have a different version of cmake

now it says its this https://github.com/ddnet/ddnet/commit/53c43536d5be86c9fb880ff5a7799f620acb40e6

map_convert_07: Support drag'n'drop · ddnet/ddnet@53c4353

18:50

im confused

18:50

Image attachment

18:50

my git bisect skills justatest

justatest

18:52

ok ill make a script

@heinrich5991 how do you link libsteam with rpath?

for good bisect

it seems to link to absolute path on mac

@Learath2 maybe check how the other libs are done

We don't build the other libs ourselves though

@Learath2 try allowing mac in the set_own_rpath function in cmakelists.txt

18:54

maybe that works

We don't want $ORIGIN on mac anyway

18:57

How do I keep finding myself reading cmake documentation at 2200 every time?

18:57

I absolutely abhor cmake

hm okay

@heinrich5991 how can I get the path we built some target at?

$<TARGET_FILE:name>

19:02

doesn't work everywhere though

I honestly don't know why it links with an absolute path

how is the search path determined?

19:04

on mac?

19:04

on linux it's completely independent from the link path

Actually, it links to @rpath correctly on mac, maybe a cross compilation issue?

19:05

Guess I'm not sleeping tonight

19:06

@heinrich5991 Search path while loading the libraries?

@Ryozuki I doubt it

@deen it tells me its 4da0592d78ed57b15738cc6cc63ce5e59723f070 again

where the executable searches its dependencies

first bad commit

I think your bisect is broken

19:09

that file doesn't even get compiled into the client

@heinrich5991 there is no search done on mac, it's either an absolute path or it's a relative path

i bisected 3 times

19:09

im done

ok, so without doing bisect

19:10

that commit doesn't work and the one before works?

rm -rf build
mkdir -p build
cd build
cmake ..
make -j12
exit_status=0
./DDNet "connect 88.198.184.126:8318" && exit 0
exit 1

19:10

it doesnt like segfault signal so i wrapped it

19:10

ignore that variable

19:11

lemme test the commit before

19:11

the one before works

19:12

(cf15ef6cc0bb48496a60a906284eec85b5807bbc)

and that one doesn't?

oh wait

19:12

maybe thats 2 before

19:12

ill test 53c43536d5be86c9fb880ff5a7799f620acb40e6

19:13

works

and does 4da0592d78ed57b15738cc6cc63ce5e59723f070 work if testing manually?

19:13

wait, did you test the other two manually?

4da0592d78ed57b15738cc6cc63ce5e59723f070 doesnt work manually

19:14

does linus use ddnet-libs?

19:14

linux

19:15

i may have forgot to git submodule update

19:15

the one before works manually

19:15

idk why 4da0592d78ed57b15738cc6cc63ce5e59723f070 doesnt work

19:16

[2020-08-25 21:15:58][datafile]: loading. filename='maps/dm1.map' [2020-08-25 21:15:58][datafile]: could not open 'maps/dm1.map' [2020-08-25 21:15:58][datafile]: loading. filename='downloadedmaps/dm1_f2159e6e.map' [2020-08-25 21:15:58][datafile]: could not open 'downloadedmaps/dm1_f2159e6e.map' [2020-08-25 21:15:58][client/network]: starting to download map to 'downloadedmaps/dm1_f2159e6e.map.100899.tmp'

19:16

kinda funny

19:16

idk why it tries to download dm1

19:16

the server map is not named like that

19:16

its named Smashy

vanilla token?

hm?

some version of the vanilla token system is implemented with dm1

@heinrich5991 does it use the client map too?

19:18

https://github.com/ddnet/ddnet/commit/4da0592d78ed57b15738cc6cc63ce5e59723f070

Replace Kobra 4 and Goo! with Gold Mine · ddnet/ddnet@4da0592

and add maps7 version

19:18

this commit removes dm1

19:18

i think

19:18

maybe its that

ah yea, maybe that

so git bisect never lies

19:19

poggers

19:20

we should add dm1 then?

Or embed dm1 I think

19:22

My cross compilation vm broke somewhere along the way, so I'm recreating that

19:22

I bet I won't be able to reproduce it

19:26

pacmangang

Image attachment

19:26

extra smart package manager

sudo pacman -S pacman-keyring

Waiting for llvm to compile on a 2 core vm is actually very annoying

19:42

So let's pick on python while waiting

19:43

How can a language with so much hand holding be so horrible at package management?

invented before the advent of useful package managers I guess?

19:46

they should fix it though

I'm also confused how someone else didn't take it in their hands

I think they're doing pyenv these days

Seems like there are no alternatives to pip, nor are there any import managers fixing the insanity that is python module resolution

I don't like it though

19:47

try pyenv, but it has emojis

19:47

[TRIGGER WARNING]

Facebook didn't like npm, so they made yarn. Made npm quickly get their shit together

19:47

@heinrich5991 no env variable to disable cringy emojis?

exists

brew and yarn also come with emojis but thank god they let me disable them

[ddnet/ddnet] Issue closed: #2643 Client segfault when trying to join teesmash mod

19:49

[ddnet:master] 1 new commit

5ffcab1 Add back Vanilla maps for compatibility (fixes #2643) - def-

@Learath2 anything I should try?

@deen can you check otool -L for the built libsteam?

Hey Someone helping nobos being blocked on multimap

Also what sdk were you building with?

@Learath2 the one we built?

19:54

the one you sent me

@deen works with latest commit

19:54

it fixed it

@Ryozuki sorry about that

I sent you 4 sdks I think

MacOSX10.13.sdk.tar.xz

19:57

apparently we don't bundle one at all @Learath2

Well one should be being built in your build folder

19:58

can you check that?

that got deleted, have to rebuild

Will DDNet eventually be rebased on TW 0.7? I really like the new UI in 0.7 (edited)

20:01

If yes. Is there anything a potential contributor could help with? (edited)

I think people said they like the simple start screen in 0.7, so we could adopt that

20:02

you can try make a pull request

@Scrumplex I did work on a 0.7 upgrade for a while, but I'd like to have a rewrite for that and I don't want to spaghetti it like the last time around. However, teeworlds source isn't very open to modding. There are some classes that are very heavily intertwined that makes it very hard to keep modifications to the source minimal

20:04

So a 0.7 upgrade is kinda unlikely until I get some more free time. However any UI improvements will be appreciated

I don't think we neccessarily would have to rebase everything at once. This could be done incrementally

20:05

I guess cherrypicking only what we want could work

@Learath2

$ x86_64-apple-darwin17-otool -L libsteam_api.dylib
libsteam_api.dylib:
    /home/deen/isos/ddnet/macosx/ddnet-master/libsteam_api.dylib (compatibility version 0.0.0, current version 0.0.0)
    /usr/lib/libc++.1.dylib (compatibility version 1.0.0, current version 400.9.0)
    /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1252.50.4)

@Scrumplex i guess you get then the spaghetti learath mentioned

@deen uh, that's concerning. Can you do a clean and a verbose build?

20:08

ninja -v for ninja. make VERBOSE=1 for make iirc

20:10

When compiling natively, I get

[learath2@l2mbp] ~/D/C/d/build $ otool -L libsteam_api.dylib 
libsteam_api.dylib:
        @rpath/libsteam_api.dylib (compatibility version 0.0.0, current version 0.0.0)
        /usr/lib/libc++.1.dylib (compatibility version 1.0.0, current version 902.1.0)
        /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1281.100.1)

20:10

The install name being absolute makes ld link with absolute path too, which causes the issue

Oh, I can fix that easily

With install_name_tool, but is that really how you want it fixed?

yes

Also need to fix linking for libsqlite for the server and libfreetype for the client

20:14

client is not using the libfreetype we bundle

I guess I'm not getting the release done today .(

20:15

my poor vm is compiling as fast as it can

probably faster than my Celeron

Great ./build.sh failed to extract the sdk

20:18

wtf

20:18

the easiest part

@Learath2

344.91 KB

Eh, I don't get why my cmake decides the install name should be relative and yours decides it should be absolute

20:22

what version of cmake are you using?

cmake 2.7?

[ddnet-scripts:master] 1 new commit

377258c build fixes - def-

cmake 3.18.2

Me too, for me the generated ninja files have INSTALLNAME_DIR = @rpath/ which gives the desired behaviour

20:27

@deen are you also building with ninja?

no, make

let me try make

@Learath2 can you try https://ddnet.tw/downloads/DDNet-14.5-osx.dmg ? (edited)

20:29

$ x86_64-apple-darwin17-otool -L DDNet
DDNet:
    /System/Library/Frameworks/Carbon.framework/Versions/A/Carbon (compatibility version 2.0.0, current version 158.0.0)
    /System/Library/Frameworks/Security.framework/Versions/A/Security (compatibility version 1.0.0, current version 58286.51.6)
    @executable_path/../Frameworks/libfreetype.6.dylib (compatibility version 24.0.0, current version 24.1.0)
    @rpath/SDL2.framework/Versions/A/SDL2 (compatibility version 1.0.0, current version 8.0.0)
    @executable_path/../Frameworks/libsteam_api.dylib (compatibility version 0.0.0, current version 0.0.0)
    /System/Library/Frameworks/Cocoa.framework/Versions/A/Cocoa (compatibility version 1.0.0, current version 22.0.0)
    /System/Library/Frameworks/OpenGL.framework/Versions/A/OpenGL (compatibility version 1.0.0, current version 1.0.0)
    /usr/lib/libc++.1.dylib (compatibility version 1.0.0, current version 400.9.0)
    /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1252.50.4)
    /System/Library/Frameworks/AppKit.framework/Versions/C/AppKit (compatibility version 45.0.0, current version 1561.40.112)
    /System/Library/Frameworks/CoreFoundation.framework/Versions/A/CoreFoundation (compatibility version 150.0.0, current version 1452.23.0)
    /System/Library/Frameworks/Foundation.framework/Versions/C/Foundation (compatibility version 300.0.0, current version 1452.23.0)
    /usr/lib/libobjc.A.dylib (compatibility version 1.0.0, current version 228.0.0)

make also links it properly for me

20:30

trying your build

why is libsteam_api.dylib linking to itself btw?

20:30

i didn't fix that yet

It doesn't link to itself

20:31

it's the odd way mach-o binaries work

20:31

the first entry is it's install name

20:31

Anyway, this works. You can also change the @executable_path/../Frameworks with @rpath/ as LC_RPATH is correct now

all of them?

For freetype and libsteam_api

and sqlite?

Yep that too, but don't forget to strip the rpath and insert the correct one

?

20:33

how?

I do it for the client

why there's new master server, what is this for?

You run the darwin_strip_rpath.py script. Then you insert @loader_path/../Frameworks

so I have to do that for the server too I guess

@nori The master server often gets attacks, we didn't want to host it alongside ddnet.tw

20:34

@deen yep

and for libsteam_api.dylib too and also change the path there?

Idk what your CMakeLists file looks like right now

complicated

20:36

Can you just show me the part we mess around with the install names?

    COMMAND ${PYTHON_EXECUTABLE} ${PROJECT_SOURCE_DIR}/scripts/darwin_strip_rpath.py ${CMAKE_OTOOL} ${CMAKE_INSTALL_NAME_TOOL} ${DMG_TMPDIR}/${TARGET_CLIENT}.app/Contents/MacOS/${TARGET_CLIENT}
    COMMAND ${CMAKE_INSTALL_NAME_TOOL} -add_rpath @loader_path/../Frameworks ${DMG_TMPDIR}/${TARGET_CLIENT}.app/Contents/MacOS/${TARGET_CLIENT}
    COMMAND ${CMAKE_INSTALL_NAME_TOOL} -change /usr/local/lib/libfreetype.6.dylib @rpath/libfreetype.6.dylib ${DMG_TMPDIR}/${TARGET_CLIENT}.app/Contents/MacOS/${TARGET_CLIENT}
    COMMAND ${CMAKE_INSTALL_NAME_TOOL} -change ${PROJECT_SOURCE_DIR}/libsteam_api.dylib @rpath/libsteam_api.dylib ${DMG_TMPDIR}/${TARGET_CLIENT}.app/Contents/MacOS/${TARGET_CLIENT}
    COMMAND ${CMAKE_INSTALL_NAME_TOOL} -change ${PROJECT_SOURCE_DIR}/libsteam_api.dylib @rpath/libsteam_api.dylib ${DMG_TMPDIR}/${TARGET_CLIENT}.app/Contents/Frameworks/libsteam_api.dylib

20:36

and:

    COMMAND ${CMAKE_COMMAND} -E make_directory ${DMG_TMPDIR}/${TARGET_SERVER}.app/Contents/Frameworks/
    COMMAND ${CMAKE_COMMAND} -E copy ${PROJECT_SOURCE_DIR}/ddnet-libs/sqlite3/mac/lib64/libsqlite3.0.dylib ${DMG_TMPDIR}/${TARGET_SERVER}.app/Contents/Frameworks/
    COMMAND ${PYTHON_EXECUTABLE} ${PROJECT_SOURCE_DIR}/scripts/darwin_strip_rpath.py ${CMAKE_OTOOL} ${CMAKE_INSTALL_NAME_TOOL} ${DMG_TMPDIR}/${TARGET_SERVER}.app/Contents/MacOS/${TARGET_SERVER}
    COMMAND ${CMAKE_INSTALL_NAME_TOOL} -add_rpath @loader_path/../Frameworks ${DMG_TMPDIR}/${TARGET_SERVER}.app/Contents/MacOS/${TARGET_SERVER}
    COMMAND ${CMAKE_INSTALL_NAME_TOOL} -change /usr/local/lib/libsqlite3.0.dylib @rpath/libsqlite3.0.dylib ${DMG_TMPDIR}/${TARGET_SERVER}.app/Contents/MacOS/${TARGET_SERVER}

does anyone know why its gfx_resizable 1 by default?

@Patiga because I like resizing windows

20:37

does it cause problems?

it was causing troubles on linux

yes

KDE

not just kde

Meh, I use it on Linux since forever

im on i3, but ye i remember it also happening on kde

20:37

it changes the resolution of your monito

20:38

*monitor

in fullscreen mode?

@deen LGTM, let me give this a try. But I doubt the last change you make does anything

didnt test whether that changes anything

@Patiga it's releated with x server

It's linking libsteam_api.dylib incorrectly to start with, which means DDNet is linked to the incorrect install name. Which you are fixing after the fact

my fix was usually gfx_resizable 0, but it got reset a few days ago, and today it happened to me again, although gfx_resizable shouldve been 0

20:39

i think it also changed the hz my monitor was set to to 60

@Learath2 you mean the

    COMMAND ${CMAKE_INSTALL_NAME_TOOL} -change ${PROJECT_SOURCE_DIR}/libsteam_api.dylib @rpath/libsteam_api.dylib ${DMG_TMPDIR}/${TARGET_CLIENT}.app/Contents/Frameworks/libsteam_api.dylib

can be removed?

@deen kde appearently sends resize events when minimizing etc

ouch

@deen I'm not completely sure, might aswell keep it

@deen you should be able to resize your ddnet manually while gfx_resizable is 0

20:41

so i was wondering if it actually serves a purpose in the usual use case

20:41

ah, i suppose its supposed to work so you never have to change the resolution yourself?

yeah, my window manager just resizes the windows

20:42

and it works fine

20:42

@Learath2 try https://ddnet.tw/downloads/DDNet-14.5-osx.dmg?version=1 please

@Patiga wym by issues? I even got it to work with the worlds most finnicky WM, macOS windowmanager

[ddnet/ddnet] Pull request opened: #2644 Fix MacOSX cross-compilation linking

it changes my monitor resolution, hz to some wrong values on some events

20:44

maybe connected to fullscreen -> switching windows

@Learath2 this is a thing, i experienced that resolution change too once

search for "gfx_resizable" in the channels, happened to a few players

and gfx_resizable is 0 by default btw

@deen works and is linked properly

yay

20:47

now let me build the steam version

Can I test the steam version too?

did that change recently? pretty sure it turned itself on

https://github.com/ddnet/ddnet/pull/2108

Make gfx_resizable default to 0 by def- · Pull Request #2108 · ddne...

because of common graphics issues. As suggested by Jupstar

oh wasnt changed back, then its fine xD

I'll change a bit about how gfx_resizable works to get it more correct for macOS when SDL decides to get their shit together and merge my patches

@deen the current version isnt in steam yet, is it?

not yet

20:50

have to fix some more things

ok bcs suddenly so many ppl report white textures

20:51

maybe just big number of players?

20:51

the fallback is already very aggressive xD

20:51

and old ones just get used to rare white textures and just restart?

20:52

I use 0.7 mostly, seen a white texture once in my life

20:52

@Deleted User with the client in #announcements that should happen anymore

20:52

@Deleted User yeah bcs 0.7 doesnt fix it either

20:52

saw your post

1

20:53

but it probably was consistent white on that signle map?? probs the background wasnt power of two

@Learath2 https://ddnet.tw/DDNet-14.5-osx.zip for steam

Dunno I've played the map many times but once it was loaded white. I've restarted the client and it was gone

20:53

I mean background layer

20:54

yeah. ok

20:54

I have RTX2070 if that matters

20:54

OHH

20:54

such a good card

20:54

no out of video memory

20:54

then its really more about a driver bug or smth xD

20:54

yeah it's perfect for playing teeworlds

We can also have a beta key for testing steam releases properly btw

and 8 GB helps me to lag less

@Deleted User is ddnet using 3d textures with old opengl now?

@Learath2 should I upload it there or can you test with what I sent you?

@redix up to opengl 2.0

20:55

before that its complete fallback

20:55

@Deleted User btw never seen white textures in any other games.

20:55

I've tested many.

20:55

@Deleted User opengl is always a miracle on windows xD

20:55

So it's not nvidia driver bug I think.

20:55

@Deleted User I am an Archlinux user

okay. there seem to be 2.x cards that don't support 256 layers for 3d textures

most games probs have either a very new opengl version or directx/vulkan

I can test with steam_appid.txt I think

mhh ok @Deleted User but i also have a modern card and never had it

20:56

@Deleted User it was only once

20:56

@redix yeah, but i prefer a complete fallback, bcs its less than 1% of all users

20:56

I'd forget to report it if smbd didn't reported the same this day

have to go, bye

true

hm.. if that helps, I have a notebook and nvidia is routed trough intel in hardware I believe

20:57

@redix https://feedback.wildfiregames.com/report/opengl/feature/GL_MAX_3D_TEXTURE_SIZE says around 4% and is 5 years old report https://opengl.gpuinfo.org/displaycapability.php?name=GL_MAX_3D_TEXTURE_SIZE says around 1%

@deen steam seems broken anyway

Image attachment

though it shouldn't affect anything

20:58

and I don't have intel drivers installed or bumblebee so it cannot fallback to intel card

20:58

but even modern intel shouldnt fail

20:58

ofc, that didn't bothered me. Are there players who get this constantly?

20:58

and since it only happens rarely its more likely to be either a really rare bug, or a rare driver bug

well there have been two cases for 0.7 in the teeworlds forum

or just there are more players after the steam release now and we just have big numbers? (edited)

@Learath2 that's probably because you downloaded it manually. So we need to do the beta I guess, can you do that yourself?

@redix yeah saw that, good we keep the old renderer for now xD

yep. people really are running this game on ancient hardware

@Deleted User you are the first to report white textures with such a good GPU really.. most ppl have some GPU from <2010 who report that

@deen I don't have access to the steam thing. Maybe @heinrich5991 can set it up

it was background though on a really big and complex map

21:00

no white things in menu

opengl 2.1 capable

Image attachment

21:01

if users dont respect that

21:01

¯\_(ツ)_/¯

yeah but i really think that GPU has the ability to accept any texture xD

21:01

the texture must have been extremly broken xD

21:01

Dunno how that could be possible.

21:01

It is my server and map I edited.

are u sure ur using the gpu

And I've played it a lot.

justatest

[2020-08-25 23:59:57][sdl]: SDL version 2.0.8 (compiled = 2.0.8)
[2020-08-25 23:59:58][gfx]: Created OpenGL 3.0 context.
[2020-08-25 23:59:59][gfx]: unable to create OpenGL context: Failed creating OpenGL context at version requested
[2020-08-25 23:59:59][sdl]: SDL version 2.0.8 (compiled = 2.0.8)
[2020-08-25 23:59:59][gfx]: unable to create OpenGL context: Failed creating OpenGL context at version requested
[2020-08-25 23:59:59][sdl]: SDL version 2.0.8 (compiled = 2.0.8)
[2020-08-25 23:59:59][gfx]: Created OpenGL 2.0 context.
[2020-08-26 00:00:00][OpenGL]: Version string: 2.1 ATI-3.9.15

even the intel GPU should easily do it tbh

Gotta love macOS

21:02

Atleast gl2 is still there

i have a laptop with ivybridge intel and it can render 3.3 completly correctly

21:02

@Ryozuki sure

Image attachment

any cheapcpu should do the trick

@Learath2 nice, does the maplad work correctly?

What is a maplad?

just load a map sry

@Deleted User that doesnt say anything

@Ryozuki wut?

ah

21:03

nvidia

@Deleted User i doubt glxinfo works on proprietary nvidia does it?

21:04

Why it shouldn't?

21:04

it works ofc

21:04

bcs its a mesa tool

@Deleted User maps load just fine

I've played Dying Light with nice perfomance on linux

poggers

Image attachment

downlaoded from steam

21:04

@Learath2 nice, then the chance is good that macos doesnt give too much trouble

macOS is honestly the most confusing thing I've ever encountered in my life

@Deleted User u sure you don't have the opensource Nvidia drivers enabled

It's the mix of great and horrible engineering in one single machine

last time i had NVidia i couldnt use mesa tools xD

https://www.youtube.com/watch?v=iYWzMvlj2RQ

Linus Torvalds: Nvidia, Fuck You!

@Learath2 ikr

greenthing

sure

Image attachment

21:06

thats mesa

21:06

I don't have this white texture now. It have appeared only once.

21:06

isnt it

21:06

I don't know what you're talking about. It's a proprietary driver from nvidia

21:06

well cannot tell rn too long with nvidia on linux

21:06

only had trouble with it xD

I had a very decent experience with nvidia on linux too on a 1080

Image attachment

21:07

nvidia is perfectly fine in linux, I use it for ML sometimes

21:07

anyway @Deleted User dont u have an nvidia control panel?

21:07

I have ofc

I wish nvidia would open a bit more of the driver like amd

it should show the correct opengl version

21:07

Image attachment

I don't mind a binary blob being uploaded to the card or a binary blob running but it'd be great if it'd integrate better with the kernel

@Learath2 i had a GTX 760 and had alot of crashes with it xD

@Learath2 i can'T find the beta thing

21:08

@Learath2 could you maybe just override that security thing temporarily?

@Deleted User it's ancient

21:08

@Deleted User yeah then its fine, maybe still a driver bug xD

[ddnet/ddnet] Issue opened: #2645 Mouse not trapped in DDNet window after game start

Every single time I start the game using Steam (haven't tried somewhere else) the mouse is not responsive. I first need to click so the mouse is usable. I have no extra cursor while the mouse is not trapped and alt-tab does not work for some reason during the same time. If other Windows 10 Steam users like me have this issue, this is a bad first impression for the common newbie. Some additional info that may help:

The Steam overlay notification "Access the Steam community while playin...

@deen I disabled SIP and it works

bcs i really never heard of anyone with modern gpu

21:09

Image attachment

21:09

Image attachment

@Learath2 thank you

Image attachment

21:09

yeah probs u should have use glxinfo | grep string

21:09

server glx vendor string: NVIDIA Corporation
server glx version string: 1.4
client glx vendor string: NVIDIA Corporation
client glx version string: 1.4
OpenGL vendor string: NVIDIA Corporation
OpenGL renderer string: GeForce RTX 2070 with Max-Q Design/PCIe/SSE2
OpenGL core profile version string: 4.6.0 NVIDIA 435.21
OpenGL core profile shading language version string: 4.60 NVIDIA
OpenGL version string: 4.6.0 NVIDIA 435.21
OpenGL shading language version string: 4.60 NVIDIA
OpenGL ES profile version string: OpenGL ES 3.2 NVIDIA 435.21
OpenGL ES profile shading language version string: OpenGL ES GLSL ES 3.20

@Learath2 server works too?

@Deleted User yeah looks good

21:10

Are there any problems? I am playing Dying Light with a full framerate.

21:10

on linux

21:10

@Deleted User i can just tell you again, i never seen anybody in tw hading trouble with textures on modern GPU

21:10

so the chance this is a driver bug isnt small

21:11

I am not having troubles

it was funny

21:11

but if by a chance of 0.00001% that would be my first start of a game I'd thought that it is a trouble indeed

21:11

though restart of a client helped

21:11

yeah I don't think it's a driver bug

21:11

I've reported this just to give you info about a modern card

21:12

I hope it will help

@deen uh, the server is missing the rpath entry

21:12

it works out because macOS ships with libsqlite which it finds in the flat namespace

but a white texture is commonly something in the opengl pipeline

21:13

bcs it replaces non bound textures with white textures

if I do install_name_tool -add_rpath @loader_path/../Frameworks ./DDNet-Server it loads the bundled lib just fine

21:17

God, I have an awful headache. I'll go try to sleep

1

1

[ddnet/ddnet] Pull request opened: #2646 Fix permanent pain sound with dummy hammer in freeze

Fixes #2532 for newer client versions by checking if the player released fire in between presses instead of checking for autofire.

who is on windows and good with SDL btw? this steam mouse glitch was reported a few times already

[ddnet/ddnet] Pull request opened: #2647 No sampler calls in old OpenGL API

Sadly didn't see it earlier, old API still binds its samplers to textures directly. Need this before steam release @def-

Exported 527 message(s)