00:37 <+bridge> [ddnet] nah im fairly sure that was it 00:37 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/696851047427211324/Screen_Shot_2020-04-06_at_5.36.17_PM.png 00:37 <+bridge> [ddnet] look i tried changing gfx_highdpi 1 00:37 <+bridge> [ddnet] and the font changed 00:37 <+bridge> [ddnet] and my fans spin on 10.6.2 now 00:38 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/696851230143545414/Screen_Shot_2020-04-06_at_5.37.53_PM.png 00:38 <+bridge> [ddnet] now turned off highdpi 00:38 <+bridge> [ddnet] and font's back to normal, fans stopped spinning 00:38 <+bridge> [ddnet] @Learath2 @heinrich5991 00:39 <+bridge> [ddnet] im prety sure the versions that caused fan spinning 00:39 <+bridge> [ddnet] didnt even have this option to turn off 00:41 <+bridge> [ddnet] finding this + the result of the bisects pretty strongly points to the high dpi being the cause of all of these problems 01:44 <+bridge> [ddnet] <> https://cdn.discordapp.com/attachments/293493549758939136/696868086300672120/screenshot_2020-04-07_01-44-35.png 01:45 <+bridge> [ddnet] <> https://cdn.discordapp.com/attachments/293493549758939136/696868091795210541/screenshot_2020-04-07_01-44-36.png 01:45 <+bridge> [ddnet] <> what is this 01:45 <+bridge> [ddnet] someone is spoofing servers 01:45 <+bridge> [ddnet] bumm with egirls 01:45 <+bridge> [ddnet] <> when i join second server i join first lol 01:45 <+bridge> [ddnet] do not join it 01:45 <+bridge> [ddnet] <> why 01:45 <+bridge> [ddnet] yep dont join 01:46 <+bridge> [ddnet] theyre probably trying to get playerips 01:46 <+bridge> [ddnet] ye 01:46 <+bridge> [ddnet] <> oh 01:46 <+bridge> [ddnet] can u still see it? 01:46 <+bridge> [ddnet] <> ye 01:46 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/696868444905144500/unknown.png 01:46 <+bridge> [ddnet] mm 01:46 <+bridge> [ddnet] <> oh 01:46 <+bridge> [ddnet] wtf 01:47 <+bridge> [ddnet] and they got several ips 01:47 <+bridge> [ddnet] its literally 01:47 <+bridge> [ddnet] just some dbag with his home internet doing it 01:47 <+bridge> [ddnet] look up his ip 01:47 <+bridge> [ddnet] the servers with ticks are real tho 01:47 <+bridge> [ddnet] nice feature to have rn 01:47 <+bridge> [ddnet] i dont see the fake servers now 01:47 <+bridge> [ddnet] 😄 01:47 <+bridge> [ddnet] <> same 01:47 <+bridge> [ddnet] sometimes they appear 01:48 <+bridge> [ddnet] rip 01:48 <+bridge> [ddnet] i wonder who is doing this 01:48 <+bridge> [ddnet] <> time to ddos him 01:48 <+bridge> [ddnet] bumm ddoser 01:48 <+bridge> [ddnet] <> jk 01:48 <+bridge> [ddnet] bumm ddoser 01:48 <+bridge> [ddnet] :monkaS: 01:49 <+bridge> [ddnet] <> :justatest: 01:49 <+bridge> [ddnet] but i wouldnt be surprised if someone tried 01:49 <+bridge> [ddnet] <> someone haha :justatest: 01:49 <+bridge> [ddnet] someone aka u 01:49 <+bridge> [ddnet] <> no u 01:49 <+bridge> [ddnet] why me 01:49 <+bridge> [ddnet] lol 01:49 <+bridge> [ddnet] <> cuz u are the biggest troll 01:50 <+bridge> [ddnet] ddosing isnt trolling its just being an ass 01:50 <+bridge> [ddnet] also no im not, no u 01:50 <+bridge> [ddnet] thats what a troll would say 05:41 <+bridge> [ddnet] @Aoe 08:25 <+bridge> [ddnet] @noby well highdpi effectively multiplies the resolution maybe the computer is genuinely having trouble getting out all the pixels 08:26 <+bridge> [ddnet] Does it atleast look sharper? 08:26 <+bridge> [ddnet] ddnet 13.0 error 08:26 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/696969099208818758/unknown.png 08:26 <+bridge> [ddnet] a lot of players got this 08:26 <+bridge> [ddnet] help 08:27 <+bridge> [ddnet] @deen ^^ did you forget to ship sth? 08:33 <+bridge> [ddnet] I would take a look but I dont have anything with windows on it 08:38 <+bridge> [ddnet] DDNet 13.0 error message . 08:38 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/696972025008226324/unknown.png 08:39 <+bridge> [ddnet] Title: entry point not found 08:39 <+bridge> [ddnet] Text: entry point to procedure Crypt... not found in dll library. 08:40 <+bridge> [ddnet] @filodex fixed, sorry about that 08:41 <+bridge> [ddnet] How did that happen? Missing libcurl.dll or sth? 08:41 <+bridge> [ddnet] yes, forgot to update libcurl.dll 08:45 <+bridge> [ddnet] Wow, so fast, Thanks! 09:07 <+bridge> [ddnet] @deen I am also getting the same error popup after updating to 13.0 09:07 <+bridge> [ddnet] I have redownloaded 13.0 twice from the webpage, but still the same thing. 09:08 <+bridge> [ddnet] ok, that's too bad 09:08 <+bridge> [ddnet] I hoped that was the fix 😄 09:10 <+bridge> [ddnet] Well, its not like its rushed or anything, I can just redownload 12.9.2 again while waiting. I was also hoping its just the libcurl.dll 09:14 <+bridge> [ddnet] It's weird that it works for me in Wine 09:15 <+bridge> [ddnet] Could be win10 version or a recent update? 09:15 <+bridge> [ddnet] to windows I mean 09:33 <+bridge> [ddnet] Well that function is deprecated but 12.9.2 works, so it can't be removed yet 09:33 <+bridge> [ddnet] I don't think microsoft does macOS like compatibility hooks checking which version a program is linked on 09:34 <+bridge> [ddnet] @deen how do we get curl for windows? just a download or do we compile that aswell? 09:34 <+bridge> [ddnet] compile myself 09:34 <+bridge> [ddnet] but I guess this has nothing to do with curl 09:34 <+bridge> [ddnet] we use CryptAcquireContext in system.c 09:35 <+bridge> [ddnet] My guess would be a mingw change and I might now have to statically link in advapi32.a 09:35 <+bridge> [ddnet] can you even statically link that? it's part of winapi 09:38 <+bridge> [ddnet] not sure, but the error message seems to say that we're missing that symbol in DDNet.exe, right? 09:39 <+bridge> [ddnet] yep 09:40 <+bridge> [ddnet] @deen does mingw come with dumpbin? 09:41 <+bridge> [ddnet] I think it has something similar, maybe you can check with that whether it links with advapi to begin with 09:45 <+bridge> [ddnet] maybe we are missing a linker flag for it? `-ladvapi` is something we could try (not sure why it worked in the past in that case though) 09:46 <+bridge> [ddnet] @onby can you profile that btw? Instruments is a pretty decent profiler, it'd be nice to see where the CPU is spending much more time 09:49 <+bridge> [ddnet] So maybe it's not our fault but SDL2 is doing something wrong 09:50 <+bridge> [ddnet] in 12.9.2 I see CryptAcquireContextA only imported from ADVAPI32.dll 09:50 <+bridge> [ddnet] but in 13.0 it's also imported from SDL2.dll 09:50 <+bridge> [ddnet] yeah, or it's also possible that the cpu is actually used more for the extra pixels, the fan curve on macbooks are VERY steep 09:50 <+bridge> [ddnet] oh you mean that 😄 09:51 <+bridge> [ddnet] huh, SDL2.dll is exporting CryptAcquireContextA? 09:51 <+bridge> [ddnet] we could also disable hidpi mode somehow maybe? 09:52 <+bridge> [ddnet] @deen can you dumpbin sdl2.dll aswell, see if it exports that? 09:52 <+bridge> [ddnet] it's also possible that minGW is just tripping balls 09:52 <+bridge> [ddnet] objdump -x seems to tell me so: https://ddnet.tw/13.0.objdump https://ddnet.tw/12.9.2.objdump 09:53 <+bridge> [ddnet] it's a bit annoying that I can't repro in wine 09:55 <+bridge> [ddnet] somehow I can't check exported symbols, only imported ones 09:55 <+bridge> [ddnet] well if sdl2.dll doesn't even export the symbol it's a problem in mingw 09:55 <+bridge> [ddnet] eh, that's a setback :/ 09:57 <+bridge> [ddnet] maybe you can't objdump libraries not created by mingw 09:57 <+bridge> [ddnet] ah, winedump works, CryptAcquireContextA is not exported in SDL2.dll 09:58 <+bridge> [ddnet] I wonder why mingws linker thinks it does 09:59 <+bridge> [ddnet] i dont get double export when i compile with mingw 09:59 <+bridge> [ddnet] using objdump 09:59 <+bridge> [ddnet] but are you also cross compiling @Jupstar ✪ 09:59 <+bridge> [ddnet] yes 09:59 <+bridge> [ddnet] what version? maybe I upgraded mingw recently and got a weird update 09:59 <+bridge> [ddnet] debian bullseye 10:00 <+bridge> [ddnet] need to lookup the verison 10:00 <+bridge> [ddnet] I verified with winedump and CryütAcquireContextA & CryptGenRandom are imported from SDL2.dll indeed 10:00 <+bridge> [ddnet] we really should finish up that docker container to compile all this 10:00 <+bridge> [ddnet] 9.3.0-8+22 10:00 <+bridge> [ddnet] my bet is that we are running into a linking bug, because the thunks have the same locations 10:00 <+bridge> [ddnet] so basically the import table for SDL2 and ADVAPI32 overlap 10:01 <+bridge> [ddnet] I guess I should try downgrading mingw and opening a bug 10:08 <+bridge> [ddnet] hopefully a downgrade takes care of it 10:26 <+bridge> [ddnet] @deen are you not using https://github.com/ddnet/ddnet-scripts/blob/master/ddnet-release.sh 10:26 <+bridge> [ddnet] to create the binary? 10:27 <+bridge> [ddnet] I am 10:27 <+bridge> [ddnet] ok 10:27 <+bridge> [ddnet] bcs i build another projects wich adds more aggressive compiler optimizations and i get the double symbol there too 10:28 <+bridge> [ddnet] but when i use that script i dont xd 10:28 <+bridge> [ddnet] hm, weird... 10:32 <+bridge> [ddnet] ok not reltaed to that, but still strange 10:33 <+bridge> [ddnet] are you stripping after the script? 10:33 <+bridge> [ddnet] oh yeah 10:33 <+bridge> [ddnet] or doesn't the script 10:33 <+bridge> [ddnet] i think the script does 10:36 <+bridge> [ddnet] oh yeah its in the cmake txt 😄 10:36 <+bridge> [ddnet] ok, do you think that might be the problem? 10:37 <+bridge> [ddnet] i don't think so, but our linker build differently appearently 10:37 <+bridge> [ddnet] so maybe ur right with linker bug, but i also get the double symbol in another project 10:37 <+bridge> [ddnet] and the other project still works fine? 10:37 <+bridge> [ddnet] I also don't get why it works fine in wine, but no on real windows 10:37 <+bridge> [ddnet] i tested it around a week ago and it worked then 10:37 <+bridge> [ddnet] but i can also just go into windows and check 10:39 <+bridge> [ddnet] well it is quite possible that wine is doing some magic to hook that call to CryptAcquireContextA which fixes the issue 10:45 <+bridge> [ddnet] well i get the error in the other project indeed, but not if not stripped 10:45 <+bridge> [ddnet] then I'll just try not stripping 10:45 <+bridge> [ddnet] and i use -s and hide debug symbols in the other project 10:45 <+bridge> [ddnet] u can build and and send it to me @deen 10:45 <+bridge> [ddnet] i can test it 10:46 <+bridge> [ddnet] how much fat does stripping trim anyway? 10:46 <+bridge> [ddnet] alot 10:46 <+bridge> [ddnet] yea 10:47 <+bridge> [ddnet] ok, the non-stripped one still has CryptAcquireContextA imported from SDL2, but not overlapped now 10:47 <+bridge> [ddnet] this leads me to believe that the overlap is an intentional optimization by strip 10:48 <+bridge> [ddnet] oh looks like i have another cmake warning to look into ```cmake 10:48 <+bridge> [ddnet] CMake Warning (dev) at /usr/share/cmake-3.17/Modules/FindPackageHandleStandardArgs.cmake:272 (message): 10:48 <+bridge> [ddnet] The package name passed to `find_package_handle_standard_args` (FFMPEG) 10:48 <+bridge> [ddnet] does not match the name of the calling package (FFmpeg). This can lead to 10:48 <+bridge> [ddnet] problems in calling code that expects `find_package` result variables 10:48 <+bridge> [ddnet] (e.g., `_FOUND`) to follow a certain pattern. 10:48 <+bridge> [ddnet] Call Stack (most recent call first): 10:48 <+bridge> [ddnet] cmake/FindFFmpeg.cmake:82 (find_package_handle_standard_args) 10:48 <+bridge> [ddnet] CMakeLists.txt:332 (find_package) 10:48 <+bridge> [ddnet] This warning is for project developers. Use -Wno-dev to suppress it. 10:48 <+bridge> [ddnet] ``` 10:48 <+bridge> [ddnet] also this 10:48 <+bridge> [ddnet] Can't set BUILD_RPATH in CMake before 3.8, pass -Wl,-rpath,'$ORIGIN' manually if you wish to emulate this. Or just install a newer version of CMake... 10:48 <+bridge> [ddnet] and i have cmake 3.17 xD 10:48 <+bridge> [ddnet] i fixed that already 10:48 <+bridge> [ddnet] both? 10:48 <+bridge> [ddnet] i see 10:48 <+bridge> [ddnet] only the FFMPEG 10:48 <+bridge> [ddnet] ah 10:49 <+bridge> [ddnet] hmm i dont get it now 10:50 <+bridge> [ddnet] (i got that stuff when compiling the aur package) 10:51 <+bridge> [ddnet] btw i found a cool thing, if you use make typing just "make -j" without specifying any core number uses the max number of cores 10:51 <+bridge> [ddnet] ok, wihtout strip it crashes for me in wine too when loading dlls 10:51 <+bridge> [ddnet] @Ryozuki no it does not 😄 10:51 <+bridge> [ddnet] rly? 10:51 <+bridge> [ddnet] well i used a non stripped and it didnt crash 10:51 <+bridge> [ddnet] that builds with maximum parallelization, so it will run as many processes as possible 10:51 <+bridge> [ddnet] (compiled by me) 10:51 <+bridge> [ddnet] If you have a small project it's usually fine, but on a large process this will kill your system and go OOM 10:51 <+bridge> [ddnet] large project* 10:51 <+bridge> [ddnet] ooo 10:52 <+bridge> [ddnet] xd 10:52 <+bridge> [ddnet] @deen are u using mingw posix? 10:52 <+bridge> [ddnet] i saw a project suggesting it in the readme xD 10:52 <+bridge> [ddnet] I'm using mingw-w64 10:52 <+bridge> [ddnet] ok i think i have posix default, dunno if the runtime can make a difference 10:53 <+bridge> [ddnet] atleast regarding the crash 10:53 <+bridge> [ddnet] the other thing is regarding stripping 😄 10:54 <+bridge> [ddnet] why does it even think SDL2 exports that symbol, maybe it's a malformed import table 10:55 <+bridge> [ddnet] well it creates the same export address 10:55 <+bridge> [ddnet] and in my project its not sdl 10:55 <+bridge> [ddnet] its some weird strip bug 10:55 <+bridge> [ddnet] i wonder if -s flag for mingw uses the linux internal strip 10:57 <+bridge> [ddnet] https://github.com/ddnet/ddnet/issues/2114 10:57 <+bridge> [ddnet] :tee_thinking: 10:59 <+bridge> [ddnet] well @deen said it crashes without the strip too 10:59 <+bridge> [ddnet] but thats something different 10:59 <+bridge> [ddnet] for me it works when i compile it 11:00 <+bridge> [ddnet] does it have the double import? 11:00 <+bridge> [ddnet] btw strip -s DDNet.exe does not remove 11:00 <+bridge> [ddnet] 11:00 <+bridge> [ddnet] symbols stripped 11:00 <+bridge> [ddnet] 11:00 <+bridge> [ddnet] debugging information removed 11:00 <+bridge> [ddnet] 11:00 <+bridge> [ddnet] for me, strangly enough 11:00 <+bridge> [ddnet] but the script uses the linux command doesnt it? 11:03 <+bridge> [ddnet] @deen does it crash for u in debug mode? 11:11 <+bridge> [ddnet] binutils source is a little too huge to dissect for this bug 😛 11:12 <+bridge> [ddnet] might also be an SDL2 bug, I'll try downgrading SDL2 later 11:15 <+bridge> [ddnet] well as i said i dont have sdl in the other project 11:15 <+bridge> [ddnet] still get the error 11:45 <+bridge> [ddnet] @Jupstar ✪ if you are here can you link verbosely? 11:45 <+bridge> [ddnet] what does that mean? 11:45 <+bridge> [ddnet] pass --verbose to ld 11:45 <+bridge> [ddnet] mh ok 11:48 <+bridge> [ddnet] it should output the linker script which I wanted to take a look at 11:49 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/697020228307255366/message.txt 11:49 <+bridge> [ddnet] dunno if that was passed to ld 11:49 <+bridge> [ddnet] i just added it to cmake linker flags 11:51 <+bridge> [ddnet] doesn't seem it made it's way to LD 11:51 <+bridge> [ddnet] mhh 11:51 <+bridge> [ddnet] it made it to gcc though 11:51 <+bridge> [ddnet] how did you add to linker flags btw? 11:52 <+bridge> [ddnet] ok i pass it to release maybe it gets overwritten 11:52 <+bridge> [ddnet] looks better now xd 11:52 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/697021055453233162/message.txt 11:53 <+bridge> [ddnet] but looks same to me 11:53 <+bridge> [ddnet] but it only is verbose when its linking 11:53 <+bridge> [ddnet] so that is mingws linker output 11:54 <+bridge> [ddnet] that's gccs --verbose output 11:54 <+bridge> [ddnet] not ld's 11:54 <+bridge> [ddnet] set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} --verbose") 11:54 <+bridge> [ddnet] what am i doing wrong then 11:55 <+bridge> [ddnet] hm, maybe try -Wl,--verbose 11:56 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/697021898046832730/message.txt 11:56 <+bridge> [ddnet] yeah thats verbose xd 11:56 <+bridge> [ddnet] let's see if there is any glaring issue 😛 11:56 <+bridge> [ddnet] fixing an actual binutils bug is a little beyond me 11:58 <+bridge> [ddnet] but i can just tell to clearify: 11:58 <+bridge> [ddnet] i use another project and have the acquire error there too(it uses -s flag which just strips too). 11:58 <+bridge> [ddnet] i don't get any error when i use either my exe or deens exe without stripping 11:58 <+bridge> [ddnet] but deens crashes, mine does not 11:59 <+bridge> [ddnet] can you dump that exe see if it still has the double import 12:00 <+bridge> [ddnet] it does not, but it also does not strip correctly when i use strip -s DDNet.exe, i'll try with -s gimme a minute 12:02 <+bridge> [ddnet] it does not double it 12:02 <+bridge> [ddnet] but in my other project it does xD 12:03 <+bridge> [ddnet] its just random 12:03 <+bridge> [ddnet] are you stripping with binutils strip btw? 12:03 <+bridge> [ddnet] from mingw? 12:03 <+bridge> [ddnet] it should be included in mingw yeah 12:04 <+bridge> [ddnet] well i used linux internal command and -s 12:04 <+bridge> [ddnet] i dunno if i have a mingwstrip 12:04 <+bridge> [ddnet] but according to cmakelist.txt it also just does strip -s 12:05 <+bridge> [ddnet] there is one in bintools and mingw does come with the entire bintools 12:05 <+bridge> [ddnet] what is the cross compiler called? 12:06 <+bridge> [ddnet] try /usr/bin/x86_64-w64-mingw32-strip 12:06 <+bridge> [ddnet] well i use mingw64 12:08 <+bridge> [ddnet] ok used the mingw strip and it stripped, but no double symbol 12:08 <+bridge> [ddnet] but that doesnt mean its working 12:08 <+bridge> [ddnet] i'll try it in windows now, if it doesnt work, it might be related to windows 12:10 <+bridge> [ddnet] The PE executable format is extremely finicky from what I can see from that linker script and the dozens of workarounds in ld's source 12:10 <+bridge> [ddnet] it is entirely possible that stripping with a different version of binutils or one that might be missing the mingw plugin could cause issues 12:12 <+bridge> [ddnet] yeah for me all versions work: 12:12 <+bridge> [ddnet] no strip 12:12 <+bridge> [ddnet] strip with linux strip 12:12 <+bridge> [ddnet] strip with mingw strip 12:12 <+bridge> [ddnet] and in my other project it does not 12:12 <+bridge> [ddnet] doing nothing else xd 12:13 <+bridge> [ddnet] I guess it is indeed a mingw/binutils issue, the best we can do is create an issue about it 12:13 <+bridge> [ddnet] well the question is does mingw -s use mingw strip or linux strip 12:14 <+bridge> [ddnet] mingw -s? 12:14 <+bridge> [ddnet] compiler flag 12:14 <+bridge> [ddnet] you can pass -s to the compiler too? 12:14 <+bridge> [ddnet] yes 12:14 <+bridge> [ddnet] bcs in the other project it uses -s, maybe deens strip does what mingw's -s does too 12:14 <+bridge> [ddnet] or its random 12:14 <+bridge> [ddnet] xD 12:15 <+bridge> [ddnet] maybe deen should use mingw's strip 12:15 <+bridge> [ddnet] but deens version without stripping worked, except it crashes 12:15 <+bridge> [ddnet] he can try but I doubt that's the issue, even without stripping it was broken 12:15 <+bridge> [ddnet] but not bcs no ddl entry was missing 12:15 <+bridge> [ddnet] why must microsoft make everything so hard for everyone? 12:16 <+bridge> [ddnet] maybe they had a bug or changed it, and mingws wrapper call breaks it currently or smth 12:16 <+bridge> [ddnet] like there is no fall back 12:17 <+bridge> [ddnet] but its too strange that the symbol is double tho 12:17 <+bridge> [ddnet] its pure random 12:18 <+bridge> [ddnet] i also don't find any issue regarding this, but the fact that 2 completly different projects have the same error 12:18 <+bridge> [ddnet] makes me wonder nobody reportet it 12:18 <+bridge> [ddnet] It'd be nice if we could all agree on an executable format, ELF is nice and it can even be extended 12:19 <+bridge> [ddnet] @Jupstar ✪ It probably triggers in pretty specific circumstances 12:20 <+bridge> [ddnet] I saw one more person ask about it on stackoverflow but no one could reply 12:20 <+bridge> [ddnet] well atleast u found smth xd 12:20 <+bridge> [ddnet] https://stackoverflow.com/questions/41885215/dependencies-to-cryptreleasecontext-and-other-advapi32-calls-with-cgo 12:21 <+bridge> [ddnet] This guy seems to have very meh english and doesn't describe the issue well, but there it is advapi functions ending up in different/wrong dlls 12:26 <+bridge> [ddnet] I don't even know if we can write a minimal testcase for this while reporting tbh 12:27 <+bridge> [ddnet] well anyway the other project doesnt work at all without or with strip. 12:27 <+bridge> [ddnet] so you really just have to report its not reproducable 12:28 <+bridge> [ddnet] well except you use the excat versions of deens arch linux and the current ddnet github maybe 12:41 <+bridge> [ddnet] okay I give up, PE is just too complex for my mortal eyes 12:42 <+bridge> [ddnet] also I ran out of music 13:04 <+bridge> [ddnet] its probably about this 13:04 <+bridge> [ddnet] http://mingw.org/wiki/MixingCompilers 13:04 <+bridge> [ddnet] 13:04 <+bridge> [ddnet] i tried changing the linkers to mingw-ld without success 13:04 <+bridge> [ddnet] so I should build SDL2 myself? 13:04 <+bridge> [ddnet] well i dunno 13:04 <+bridge> [ddnet] test it out i have 0 plan xD 13:05 <+bridge> [ddnet] to me this sounds like an unrelated issue 13:05 <+bridge> [ddnet] but i remember i build the libs for my project too a longer time ago, so maybe thats it 13:06 <+bridge> [ddnet] i'll try it out for my project and give u feedback 13:09 <+bridge> [ddnet] @deen it indeed worked now 13:09 <+bridge> [ddnet] so yes 13:09 <+bridge> [ddnet] u need to build sdl for mingw yourself XD 13:10 <+bridge> [ddnet] will probs work 13:10 <+bridge> [ddnet] meh 13:10 <+bridge> [ddnet] ah, sdl2 also provides development libs for mingw 13:11 <+bridge> [ddnet] i always used the VS ones I think? 13:11 <+bridge> [ddnet] yeah i think ddnet uses vs ones 13:12 <+bridge> [ddnet] that's the other problem, if I provide mingw ones, then maybe we get other problems when someone builds in visual studio 13:12 <+bridge> [ddnet] so we need both? 13:12 <+bridge> [ddnet] probably the best way 13:12 <+bridge> [ddnet] @heinrich5991 does adding 13:12 <+bridge> [ddnet] set(CMAKE_LINKER x86_64-w64-mingw32-ld) 13:12 <+bridge> [ddnet] set(CMAKE_AR x86_64-w64-mingw32-gcc-ar) 13:13 <+bridge> [ddnet] set(CMAKE_RANLIB x86_64-w64-mingw32-gcc-ranlib) 13:13 <+bridge> [ddnet] to the mingw toolchain makes sense? 13:24 <+bridge> [ddnet] @deen would this commit help? https://github.com/heinrich5991/ddnet/commit/293e58c932e6ffeb779f07e5b0e44fdfe1a849d8 13:25 <+bridge> [ddnet] I don't know. I can try 13:30 <+bridge> [ddnet] ``` 13:30 <+bridge> [ddnet] /usr/lib/gcc/x86_64-w64-mingw32/9.3.0/../../../../x86_64-w64-mingw32/bin/ld: CMakeFiles/DDNet.dir/objects.a(system.c.obj):system.c:(.text+0x579e): undefined reference to `RtlGenRandom' 13:30 <+bridge> [ddnet] /usr/lib/gcc/x86_64-w64-mingw32/9.3.0/../../../../x86_64-w64-mingw32/bin/ld: CMakeFiles/DDNet.dir/objects.a(system.c.obj):system.c:(.text+0x5878): undefined reference to `RtlGenRandom' 13:30 <+bridge> [ddnet] /usr/lib/gcc/x86_64-w64-mingw32/9.3.0/../../../../x86_64-w64-mingw32/bin/ld: CMakeFiles/DDNet.dir/objects.a(system.c.obj):system.c:(.text+0x58e2): undefined reference to `RtlGenRandom' 13:30 <+bridge> [ddnet] ``` 13:34 <+bridge> [ddnet] interesting. I don't get the error locally 😦 13:35 <+bridge> [ddnet] @deen can you try again? I pushed another version 13:40 <+bridge> [ddnet] @heinrich5991 works for me in wine 13:41 <+bridge> [ddnet] sent exe to Jupstar for windows verification 13:53 <+bridge> [ddnet] in the struct `CSwitchTile` from `mapitems.h` 13:53 <+bridge> [ddnet] ``` 13:53 <+bridge> [ddnet] class CSwitchTile 13:53 <+bridge> [ddnet] { 13:53 <+bridge> [ddnet] public: 13:53 <+bridge> [ddnet] unsigned char m_Number; 13:53 <+bridge> [ddnet] unsigned char m_Type; 13:53 <+bridge> [ddnet] unsigned char m_Flags; 13:53 <+bridge> [ddnet] unsigned char m_Delay; 13:53 <+bridge> [ddnet] }; 13:53 <+bridge> [ddnet] ``` 13:53 <+bridge> [ddnet] where can i find what `m_Flags` contains? 13:54 <+bridge> [ddnet] search how its used 13:54 <+bridge> [ddnet] didnt find anything in reasonable time 13:54 <+bridge> [ddnet] so i figured someone here might either know it or find it a lot faster 13:56 <+bridge> [ddnet] it *looks like* they're the normal tile flags 13:57 <+bridge> [ddnet] i.e. rotation, mirroring 13:57 <+bridge> [ddnet] 👍 thx 14:14 <+bridge> [ddnet] so for Jupstar it crashes, not sure why 14:26 <+bridge> [ddnet] @heinrich5991 any hint for how to best use these paths? 14:48 <+bridge> [ddnet] @deen add them to the `set_extra_dirs_lib` function 14:48 <+bridge> [ddnet] in `CMakeLists.txt` 14:52 <+bridge> [ddnet] ``` 14:52 <+bridge> [ddnet] if(TARGET_BITS AND TARGET_OS) 14:52 <+bridge> [ddnet] set(DIR "ddnet-libs/${NAME}/${TARGET_OS}/lib${TARGET_BITS}") 14:52 <+bridge> [ddnet] if(TARGET_OS STREQUAL) 14:52 <+bridge> [ddnet] if(MSVC) 14:52 <+bridge> [ddnet] list(APPEND DIR "ddnet-libs/${TARGET_OS}/lib${TARGET_BITS}/vc") 14:52 <+bridge> [ddnet] else() 14:52 <+bridge> [ddnet] list(APPEND DIR "ddnet-libs/${TARGET_OS}/lib${TARGET_BITS}/mingw") 14:52 <+bridge> [ddnet] endif() 14:52 <+bridge> [ddnet] endif() 14:52 <+bridge> [ddnet] [...] 14:53 <+bridge> [ddnet] endif() 14:53 <+bridge> [ddnet] ``` 14:53 <+bridge> [ddnet] @deen ^ 14:53 <+bridge> [ddnet] btw, the libsdl.org sdl2.dll for mingw also didn't fix the issue, so now I'm building my own 14:53 <+bridge> [ddnet] thanks 14:55 <+bridge> [ddnet] ``` 14:55 <+bridge> [ddnet] if(TARGET_BITS AND TARGET_OS) 14:55 <+bridge> [ddnet] set(DIR "ddnet-libs/${NAME}/${TARGET_OS}/lib${TARGET_BITS}") 14:55 <+bridge> [ddnet] if(TARGET_OS STREQUAL "windows") 14:55 <+bridge> [ddnet] if(MSVC) 14:55 <+bridge> [ddnet] list(APPEND DIR "ddnet-libs/${TARGET_OS}/lib${TARGET_BITS}/vc") 14:55 <+bridge> [ddnet] else() 14:55 <+bridge> [ddnet] list(APPEND DIR "ddnet-libs/${TARGET_OS}/lib${TARGET_BITS}/mingw") 14:55 <+bridge> [ddnet] endif() 14:55 <+bridge> [ddnet] endif() 14:55 <+bridge> [ddnet] [...] 14:55 <+bridge> [ddnet] endif() 14:55 <+bridge> [ddnet] ``` 15:18 <+bridge> [ddnet] @deen do you understand this error? 15:18 <+bridge> [ddnet] ``` 15:18 <+bridge> [ddnet] W: The repository 'https://dl.yarnpkg.com/debian stable Release' does not have a Release file. 15:18 <+bridge> [ddnet] E: Failed to fetch https://dl.yarnpkg.com/debian/dists/stable/main/binary-amd64/Packages 503 No healthy IP available for the backend 15:18 <+bridge> [ddnet] ``` 15:18 <+bridge> [ddnet] doesn't seem to be related to the change 15:22 <+bridge> [ddnet] yeah, can't reach dl.yarnpkg.com 15:22 <+bridge> [ddnet] infrastructure problem i guess 15:24 <+bridge> [ddnet] Hm, why do I still need SDL2.lib when I build my own SDL2.dll 15:24 <+bridge> [ddnet] windows dynamic linking works like this: you build a library that you statically link that takes care of dynamically linking the actual dll 15:25 <+bridge> [ddnet] (as far as I understand it; I read a bit about it, but I haven't looked into it in that much detail) 15:25 <+bridge> [ddnet] so how do i get the .lib file? 15:25 <+bridge> [ddnet] building SDL2 didn't create it 15:26 <+bridge> [ddnet] hm, it should 15:26 <+bridge> [ddnet] and the official SDL2 mingw version also doesn't contain a .lib file 15:26 <+bridge> [ddnet] hm ok 15:26 <+bridge> [ddnet] it does 15:26 <+bridge> [ddnet] oh? 15:27 <+bridge> [ddnet] \x86_64-w64-mingw32\lib\ 15:27 <+bridge> [ddnet] currently don't have time to investigate, and I'm probably not being helpful, I'll just leave for now 15:27 <+bridge> [ddnet] libSDL2.dll.a 15:27 <+bridge> [ddnet] it probs static links the shared api for the shared library 15:28 <+bridge> [ddnet] (ok its not a .lib file like for vs) but its a normal static lib u have to link against 15:34 <+bridge> [ddnet] make[3]: *** No rule to make target 'ddnet-libs/sdl/windows/lib64/SDL2.lib', needed by 'DDNet.exe'. Stop. 15:38 <+bridge> [ddnet] ok, works now 15:38 <+bridge> [ddnet] had to remove CMakeCache.txt 15:39 <+bridge> [ddnet] finally, runs fine in wine 15:39 <+bridge> [ddnet] please test on Windows if client starts fine: https://ddnet.tw/DDNet-13.0-win64.zip 15:42 <+bridge> [ddnet] @deen works for me 😄 15:42 <+bridge> [ddnet] was it with self compiled libs now?= 15:42 <+bridge> [ddnet] yes 15:42 <+bridge> [ddnet] ok 15:42 <+bridge> [ddnet] thanks for the hint @Jupstar ✪ 15:42 <+bridge> [ddnet] well i looked thorugh SDL and only saw that for WinRT it compiles cpp code too 15:43 <+bridge> [ddnet] dunno if that is enabled by default 15:43 <+bridge> [ddnet] Will release in the evening, see you 15:43 <+bridge> [ddnet] or mingw isnt downward compatible to all types of c dlls 15:43 <+bridge> [ddnet] cya 16:17 <+bridge> [ddnet] could we have the freeze stars for other tees in /showothers also be transparent and maybe hammer / hook sounds could also be less volume? 18:09 <+bridge> [ddnet] hello 18:09 <+bridge> [ddnet] i have problem 18:09 <+bridge> [ddnet] whern i try to convert my demos to .mp4 my ddnet i crash 18:09 <+bridge> [ddnet] whern i try to convert my demos to .mp4 my ddnet crash 18:10 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/697116041867952228/message.txt 18:10 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/697116047907487914/message.txt 18:11 <+bridge> [ddnet] relevant part: 18:11 <+bridge> [ddnet] ``` 18:11 <+bridge> [ddnet] [2020-04-07 18:00:38][datafile]: loading data index=15 size=13883 uncompressed=262144 18:11 <+bridge> [ddnet] [2020-04-07 18:00:38][demo_render]: ppp.demo.mp4 18:11 <+bridge> [ddnet] [NULL @ 0x55978136e520] Requested output format 'mp4' is not a suitable output format 18:11 <+bridge> [ddnet] [2020-04-07 18:00:38][video_recorder]: Failed to create formatcontext for recoding video. 18:11 <+bridge> [ddnet] [2020-04-07 18:00:39][datafile]: loading data index=16 size=88 uncompressed=152 18:11 <+bridge> [ddnet] [...] 18:11 <+bridge> [ddnet] [2020-04-07 18:00:40][datafile]: loading data index=25 size=620 uncompressed=2888 18:11 <+bridge> [ddnet] [2020-04-07 18:01:00][video_recorder]: called 18:11 <+bridge> [ddnet] Erreur de segmentation (core dumped) 18:11 <+bridge> [ddnet] ``` 18:12 <+bridge> [ddnet] and ? 18:12 <+bridge> [ddnet] sorry, wasn't directed at you 18:12 <+bridge> [ddnet] just in case someone wanted to take a quick look without downloading the log 18:12 <+bridge> [ddnet] I don't know about the video recorder tbh 18:13 <+bridge> [ddnet] ah ok 18:13 <+bridge> [ddnet] np 18:13 <+bridge> [ddnet] :3 18:45 <+bridge> [ddnet] lmao just wasted some hours of life the create a bash script that generates bash code using sed to prepare sed statements 18:45 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/697124911063498752/unknown.png 18:46 <+bridge> [ddnet] it s for me ? 18:46 <+bridge> [ddnet] 🤦🏻‍♂️ 18:47 <+bridge> [ddnet] @Ryuken nah its for my vim config 😄 18:47 <+bridge> [ddnet] ur free to use it as well 😄 18:59 <+bridge> [ddnet] @deen does replacing `"${EXTRA_CURL_LIBDIR}/libcurl.dll"` with `"${CURL_LIBRARY}"` in `cmake/FindCurl.cmake` help? 19:59 <+bridge> [ddnet] @heinrich5991 yes, that worked, but now it doesn't find SDL2. Maybe I should just get rid of the VS version 20:00 <+bridge> [ddnet] @deen do the same to the line with SDL2.dll in `cmake/FindSDL2.cmake` 20:00 <+bridge> [ddnet] i did 21:55 <+bridge> [ddnet] sorry, couldn't follow what happened today 21:55 <+bridge> [ddnet] @deen does it work if we compile our own SDL? 22:04 <+bridge> [ddnet] 25 AM] Learath2: @onby well highdpi effectively multiplies the resolution maybe the computer is genuinely having trouble getting out all the pixels 22:04 <+bridge> [ddnet] [1:26 AM] Learath2: Does it atleast look sharper? 22:04 <+bridge> [ddnet] 22:04 <+bridge> [ddnet] it changes the way the font looks i guess 22:04 <+bridge> [ddnet] but the change isnt worth making the fans spin and it would be cool if it could be disabled in newer versions 22:06 <+bridge> [ddnet] if it's just the font we are probably doing sth wrong about highdpi anyway 22:11 <+bridge> [ddnet] Maybe the rest of the texture is not high-res enough, so you don't see a difference. The text renderer automatically uses a bigger font size if resolution goes up 22:13 <+bridge> [ddnet] you can look for the rounded corners on rectangles, those should look better 22:13 <+bridge> [ddnet] but highdpi will definitely increase gpu load, so the fans start spinning 22:13 <+bridge> [ddnet] high dpi cant be turned off on new ddnet it seems? 22:13 <+bridge> [ddnet] @Learath2 maybe we can give an option to disable hidpi? 22:13 <+bridge> [ddnet] yess 22:13 <+bridge> [ddnet] pls 22:15 <+bridge> [ddnet] definitely, we should also steal some of the highdpi code from 0.7 we are doing some things wrong, but 0.7 has most of it right 22:15 <+bridge> [ddnet] (atleast for macOS 22:15 <+bridge> [ddnet] ) 22:18 <+bridge> [ddnet] nonononono 22:18 <+bridge> [ddnet] no 22:18 <+bridge> [ddnet] 0.7 is ass 22:18 <+bridge> [ddnet] it makes my fans spin so hard 22:19 <+bridge> [ddnet] and it doesnt even let me change the resolution atall 22:19 <+bridge> [ddnet] it has an option to turn off highdpi 22:19 <+bridge> [ddnet] does it come at the expense of those two things^ 22:19 <+bridge> [ddnet] try `gfx_highdpi 0` in 0.7 22:20 <+bridge> [ddnet] ohhhhh wow 22:20 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/697179107988865085/Screen_Shot_2020-04-07_at_3.20.23_PM.png 22:20 <+bridge> [ddnet] etf 22:21 <+bridge> [ddnet] etf? 22:21 <+bridge> [ddnet] wtf* 22:21 <+bridge> [ddnet] cus its cutting off 22:21 <+bridge> [ddnet] half of the ui 22:24 <+bridge> [ddnet] the fuck 22:24 <+bridge> [ddnet] looks like the clipping is broken 22:24 <+bridge> [ddnet] Thats looks not good 22:25 <+bridge> [ddnet] open an issue in tw repo 22:25 <+bridge> [ddnet] i kindof gave up on that 22:25 <+bridge> [ddnet] after the resolution bug has been a "known bug" for like a year 22:25 <+bridge> [ddnet] and noones tried to fix it 22:25 <+bridge> [ddnet] i literally cant see what i type in the chat lol 22:25 <+bridge> [ddnet] I know what's brokebn 22:26 <+bridge> [ddnet] wait I'm not sure anymore, I patched SDL so much to get 0.7 working 22:26 <+bridge> [ddnet] @onby is your resolution set to the desktop resolution? 22:26 <+bridge> [ddnet] no 22:26 <+bridge> [ddnet] i want windowed 1024x768 like i have on ddnet 22:27 <+bridge> [ddnet] put it to desktop resolution so we can test it 22:27 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/697180715409997824/Screen_Shot_2020-04-07_at_3.27.08_PM.png 22:27 <+bridge> [ddnet] oh here are my settings 22:27 <+bridge> [ddnet] uuh 22:27 <+bridge> [ddnet] that will make me lag even more but ok 22:27 <+bridge> [ddnet] or you can patch the source to allow arbitrary resolution 22:27 <+bridge> [ddnet] but don't toggle fullscreen 22:28 <+bridge> [ddnet] how 22:28 <+bridge> [ddnet] backend_sdl.cpp L703, put 702 703 704 in a block 22:29 <+bridge> [ddnet] what the hell 22:29 <+bridge> [ddnet] it changes my resolution back to 1280x800 22:29 <+bridge> [ddnet] no matter what i do 22:29 <+bridge> [ddnet] even if i change it in settings.cfg 22:29 <+bridge> [ddnet] those are the lines that forces it 22:29 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/697181356488392704/unknown.png 22:30 <+bridge> [ddnet] put it in a block like this and it'll stop forcing the resolution 22:30 <+bridge> [ddnet] @Learath2 yes, compiling our own SDL fixed it 22:30 <+bridge> [ddnet] @deen guess cross-compiling with mingw is different from compiling with mingw on windows 22:31 <+bridge> [ddnet] XDD 22:31 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/697181796806164560/Screen_Shot_2020-04-07_at_3.31.23_PM.png 22:31 <+bridge> [ddnet] interesting 22:32 <+bridge> [ddnet] @onby is that with highdpi off? 22:32 <+bridge> [ddnet] yesr 22:32 <+bridge> [ddnet] you can turn it on and off right? 22:32 <+bridge> [ddnet] idk if it needs restart but yes 22:32 <+bridge> [ddnet] it shouldn't break anything 22:32 <+bridge> [ddnet] great 22:32 <+bridge> [ddnet] does it change gpu/cpu usage? 22:33 <+bridge> [ddnet] yes i get noticeably better fps in 0.7 with it turned off aha 22:34 <+bridge> [ddnet] but it clips the ui 22:34 <+bridge> [ddnet] that is another thing that suggests that ddnet *needs* the option to turn off high dpi support 22:35 <+bridge> [ddnet] on windows there's a built-in way to disable hidpi, didn't know mac osx doesn't have one 22:38 <+bridge> [ddnet] oh cmon, @onby why didn't you say it broke clipping 22:38 <+bridge> [ddnet] yessssss thank u xd 22:38 <+bridge> [ddnet] uh 22:39 <+bridge> [ddnet] because i just found out about it two seconds ago? 22:39 <+bridge> [ddnet] i hadnt thought to try gfx_highdpi 0 22:39 <+bridge> [ddnet] https://github.com/ddnet/ddnet/pull/2120/files#diff-0667e4904b6955f3b3b84482d7167d0eR568 22:39 <+bridge> [ddnet] until u suggested it 22:39 <+bridge> [ddnet] I think there's a way to get the logical vs physical resolution 22:39 <+bridge> [ddnet] i dont have that problem on 06 22:39 <+bridge> [ddnet] which is probably what we need here 22:40 <+bridge> [ddnet] The resolution you get from GL_Drawablesize is the resolution we should be rendering in 22:40 <+bridge> [ddnet] we already do that 22:40 <+bridge> [ddnet] @onby is that 0.7 master btw? 22:40 <+bridge> [ddnet] no 22:40 <+bridge> [ddnet] i can try 22:41 <+bridge> [ddnet] its on my test client based on 0.7.3.1 22:41 <+bridge> [ddnet] can you try master instead, just so I don't try to hunt down a bug that no longer exists 22:41 <+bridge> [ddnet] ye 22:41 <+bridge> [ddnet] #if defined(CONF_PLATFORM_MACOSX) // Todo SDL: remove this when fixed (game freezes when losing focus in fullscreen) 22:41 <+bridge> [ddnet] { 22:41 <+bridge> [ddnet] SdlFlags |= SDL_WINDOW_FULLSCREEN_DESKTOP; // always use "fake" fullscreen 22:41 <+bridge> [ddnet] *pWindowWidth = *pDesktopWidth; 22:41 <+bridge> [ddnet] *pWindowHeight = *pDesktopHeight; 22:41 <+bridge> [ddnet] } 22:41 <+bridge> [ddnet] just that? 22:41 <+bridge> [ddnet] yeah, that should get you windowed gfx at any resolution 22:41 <+bridge> [ddnet] OK 22:42 <+bridge> [ddnet] btw are you on catalina? 22:43 <+bridge> [ddnet] The clipping issue happens when m_ScreenWidth and m_ScreenHeight in the graphics backend don't match the actual resolution 22:43 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/697184765496918066/Screen_Shot_2020-04-07_at_3.43.13_PM.png 22:43 <+bridge> [ddnet] yes same issue on master 22:43 <+bridge> [ddnet] @onby are you on macOS Catalina? 22:43 <+bridge> [ddnet] @Learath2 should I merge the hidpi change and my sdl2 and try to release a new one? Or do you still need to fix the clipping thing? 22:43 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/697184839597686794/Screen_Shot_2020-04-07_at_3.43.32_PM.png 22:43 <+bridge> [ddnet] @deen nah, clipping thing is not an issue in ddnet 22:43 <+bridge> [ddnet] "Use Highdpi" → "Use high DPI" 22:45 <+bridge> [ddnet] @onby can you add a debug print before and after SDL_GL_GetDrawableSize and print the resolution? 22:45 <+bridge> [ddnet] SDL_GL_GetDrawableSize(m_pWindow, pScreenWidth, pScreenHeight); // drawable size may differ in high dpi mode 22:45 <+bridge> [ddnet] OK 22:45 <+bridge> [ddnet] `*pScreenWidth` and `*pScreenHeight` that is 22:45 <+bridge> [ddnet] ye 22:46 <+bridge> [ddnet] there are like 4 resolutions right there so it can get a little confusing which one to print 😛 22:46 <+bridge> [ddnet] 0 -1 -1 22:46 <+bridge> [ddnet] 1 2560 1600 22:46 <+bridge> [ddnet] lol 22:46 <+bridge> [ddnet] SDL_GetWindowSize(m_pWindow, pWindowWidth, pWindowHeight); 22:46 <+bridge> [ddnet] printf("0 %d %d\n", *pScreenWidth, *pScreenHeight); 22:46 <+bridge> [ddnet] 22:46 <+bridge> [ddnet] SDL_GL_GetDrawableSize(m_pWindow, pScreenWidth, pScreenHeight); // drawable size may differ in high dpi mode 22:46 <+bridge> [ddnet] printf("1 %d %d\n", *pScreenWidth, *pScreenHeight); 22:46 <+bridge> [ddnet] is that with gfx_highdpi 1 or 0? 22:46 <+bridge> [ddnet] 0 22:46 <+bridge> [ddnet] try with 1 too please 22:47 <+bridge> [ddnet] gfx_highdbi for ddnet? 22:47 <+bridge> [ddnet] [2020-04-07 15:47:10][client]: starting... 22:47 <+bridge> [ddnet] 0 -1 -1 22:47 <+bridge> [ddnet] 1 2560 1600 22:47 <+bridge> [ddnet] [2020-04-07 15:47:11][sdl]: SDL version 2.0.8 (dll = 2.0.8) 22:47 <+bridge> [ddnet] [2020-04-07 15:47:11][render]: opengl max texture sizes: 16384, 2048(3D) 22:47 <+bridge> [ddnet] mm 22:47 <+bridge> [ddnet] highdpi fixes the clipping 22:47 <+bridge> [ddnet] but causes fan spinning 22:47 <+bridge> [ddnet] the "better font but horrible cpu usage" seems to be the same problem for ddnet high dpi and also 0.7 22:47 <+bridge> [ddnet] ooh, i bet this is some macOS issue again 22:48 <+bridge> [ddnet] great xd 22:48 <+bridge> [ddnet] see how you set gfx_highdpi and the drawable resolution doesn't change? 22:48 <+bridge> [ddnet] that's not supposed to happen 22:48 <+bridge> [ddnet] ooo 22:48 <+bridge> [ddnet] i see 22:49 <+bridge> [ddnet] `[2020-04-07 23:49:03][DEBUG]: 1920 1200` when I set gfx_highdpi 0 22:49 <+bridge> [ddnet] do u have high dpi screen? do u have osx 22:50 <+bridge> [ddnet] `[2020-04-07 23:49:43][DEBUG]: 3840 2400` when I set gfx_highdpi 1 22:50 <+bridge> [ddnet] o_o 22:50 <+bridge> [ddnet] macOS catalina on a macbook pro, with a high dpi screen 22:50 <+bridge> [ddnet] nice 22:50 <+bridge> [ddnet] so it works for u? 22:50 <+bridge> [ddnet] yeah it works fine for me, but I'm wondering whether it's my patched up SDL 22:50 <+bridge> [ddnet] nope, doesn't seem to be 22:52 <+bridge> [ddnet] @deen you can merge highdpi if you want, but I'm not certain how it behaves on windows 22:52 <+bridge> [ddnet] the default is what we have right now, so I guess it's ok even if it doesn't work great 22:52 <+bridge> [ddnet] @onby I have an idea, can you try sth for me? 22:53 <+bridge> [ddnet] `git cherry-pick 013682a6` 22:54 <+bridge> [ddnet] where 22:55 <+bridge> [ddnet] on master 22:55 <+bridge> [ddnet] well i'm not sure if that commit will resolve though, so you might need to add my repo as a remote 22:55 <+bridge> [ddnet] -.7? 22:55 <+bridge> [ddnet] yeah on 0.7 22:55 <+bridge> [ddnet] nobody1s-MacBook-Pro:teeworlds-master nobody1$ git cherry-pick 013682a6 22:55 <+bridge> [ddnet] fatal: Not a git repository (or any of the parent directories): .git 22:55 <+bridge> [ddnet] 22:56 <+bridge> [ddnet] ..xd 22:56 <+bridge> [ddnet] i suppose i should clone it 22:56 <+bridge> [ddnet] no no 22:56 <+bridge> [ddnet] well its not a git repo 22:56 <+bridge> [ddnet] i just downloaded zip 22:56 <+bridge> [ddnet] nobody1s-MacBook-Pro:teeworlds nobody1$ git cherry-pick 013682a6 22:56 <+bridge> [ddnet] fatal: bad revision '013682a6' 22:56 <+bridge> [ddnet] o 22:56 <+bridge> [ddnet] just add https://github.com/Learath2/teeworlds.git 22:57 <+bridge> [ddnet] how 22:57 <+bridge> [ddnet] `git remote add l2 https://github.com/Learath2/teeworlds.git` 22:57 <+bridge> [ddnet] `git fetch l2` 22:57 <+bridge> [ddnet] that should fetch my commits, then you can cherry-pick 22:57 <+bridge> [ddnet] oo ok 22:58 <+bridge> [ddnet] nobody1s-MacBook-Pro:teeworlds nobody1$ git cherry-pick 013682a6 22:58 <+bridge> [ddnet] error: could not apply 013682a68... Implement runtime resizing 22:58 <+bridge> [ddnet] hint: after resolving the conflicts, mark the corrected paths 22:58 <+bridge> [ddnet] hint: with 'git add ' or 'git rm ' 22:58 <+bridge> [ddnet] hint: and commit the result with 'git commit' 22:58 <+bridge> [ddnet] nobody1s-MacBook-Pro:teeworlds nobody1$ 22:58 <+bridge> [ddnet] conflicts?? 22:58 <+bridge> [ddnet] * [new branch] fngspoof -> l2/fngspoof 22:58 <+bridge> [ddnet] 22:58 <+bridge> [ddnet] what is this o_o 22:58 <+bridge> [ddnet] huh no idea, I wonder what I was doing with that branch 22:58 <+bridge> [ddnet] same 22:58 <+bridge> [ddnet] lol 22:58 <+bridge> [ddnet] yeah idk why conflicts 22:58 <+bridge> [ddnet] i just git clone https://github.com/teeworlds/teeworlds 22:59 <+bridge> [ddnet] then did the commands u otld me 22:59 <+bridge> [ddnet] @onby well do a git status 22:59 <+bridge> [ddnet] nobody1s-MacBook-Pro:teeworlds nobody1$ git status 22:59 <+bridge> [ddnet] On branch master 22:59 <+bridge> [ddnet] Your branch is up to date with 'origin/master'. 22:59 <+bridge> [ddnet] 22:59 <+bridge> [ddnet] You are currently cherry-picking commit 013682a68. 22:59 <+bridge> [ddnet] (fix conflicts and run "git cherry-pick --continue") 22:59 <+bridge> [ddnet] (use "git cherry-pick --abort" to cancel the cherry-pick operation) 22:59 <+bridge> [ddnet] 22:59 <+bridge> [ddnet] Changes to be committed: 22:59 <+bridge> [ddnet] 22:59 <+bridge> [ddnet] modified: src/engine/client/graphics_threaded.cpp 22:59 <+bridge> [ddnet] modified: src/engine/client/input.cpp 22:59 <+bridge> [ddnet] modified: src/engine/graphics.h 22:59 <+bridge> [ddnet] modified: src/game/client/components/menus_scrollregion.cpp 23:00 <+bridge> [ddnet] modified: src/game/client/components/menus_settings.cpp 23:00 <+bridge> [ddnet] 23:00 <+bridge> [ddnet] Unmerged paths: 23:00 <+bridge> [ddnet] (use "git add ..." to mark resolution) 23:00 <+bridge> [ddnet] 23:00 <+bridge> [ddnet] both modified: src/engine/client/backend_sdl.cpp 23:00 <+bridge> [ddnet] both modified: src/engine/client/backend_sdl.h 23:00 <+bridge> [ddnet] both modified: src/engine/client/graphics_threaded.h 23:00 <+bridge> [ddnet] yeah no, do a git cherry-pick --abort 23:00 <+bridge> [ddnet] ok 23:00 <+bridge> [ddnet] do i have to manually fix the sdl include paths again 23:00 <+bridge> [ddnet] then do a `git checkout 013682a68` instead 23:00 <+bridge> [ddnet] src/engine/client/input.cpp:3:10: fatal error: 'SDL.h' file not found 23:00 <+bridge> [ddnet] #include "SDL.h" 23:00 <+bridge> [ddnet] ^~~~~~~ 23:00 <+bridge> [ddnet] 1 error generated. 23:00 <+bridge> [ddnet] heh 23:00 <+bridge> [ddnet] looks like i do 23:00 <+bridge> [ddnet] is that after you aborted the cherry pick and checked out that commit? 23:00 <+bridge> [ddnet] after abort, before checkout 23:01 <+bridge> [ddnet] i typed bam 23:01 <+bridge> [ddnet] and 23:01 <+bridge> [ddnet] im not surprised, i always have to fix this manally 23:01 <+bridge> [ddnet] I wonder what's wrong with your installation of SDL 23:01 <+bridge> [ddnet] by changing "SDL.h" to 23:01 <+bridge> [ddnet] I think cmake should handle that better noby 23:01 <+bridge> [ddnet] oh 23:02 <+bridge> [ddnet] i use abm 23:02 <+bridge> [ddnet] … 23:02 <+bridge> [ddnet] ill try cmake 23:02 <+bridge> [ddnet] well it works with bam too for me, but your computer is weird 😛 23:02 <+bridge> [ddnet] xd 23:02 <+bridge> [ddnet] that runtime resizing patch should allow AppKit to resize the window to the correct size 23:03 <+bridge> [ddnet] sounds good 23:03 <+bridge> [ddnet] im compiling rn 23:03 <+bridge> [ddnet] if that doesn't work either, then idk why that surface is larger then it reports 23:03 <+bridge> [ddnet] oh can you also add a dbg_msg? 23:03 <+bridge> [ddnet] lol it worked wit cmake wtf 23:03 <+bridge> [ddnet] where 23:04 <+bridge> [ddnet] SDL_WINDOWEVENT_SIZE_CHANGED 23:04 <+bridge> [ddnet] xd 23:04 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/697190078589435956/Screen_Shot_2020-04-07_at_4.04.18_PM.png 23:04 <+bridge> [ddnet] uh 23:04 <+bridge> [ddnet] right below there, just add a dbg_msg("DEBUG", "o/"); 23:04 <+bridge> [ddnet] what was i supposed to test again 23:05 <+bridge> [ddnet] well that clipping issue happens when the client is using the wrong resolution, but opengl reports a smaller drawable resolution 23:05 <+bridge> [ddnet] so I was thinking maybe that window is resized by AppKit after launch when it knows we can do highdpi 23:05 <+bridge> [ddnet] the debug msgs never comes up 23:06 <+bridge> [ddnet] can you try one last thing? then I need to go sleep anyway 😛 23:06 <+bridge> [ddnet] i tried 23:06 <+bridge> [ddnet] case SDL_WINDOWEVENT: 23:06 <+bridge> [ddnet] dbg_msg("DEBUG", "asdfasdf"); 23:06 <+bridge> [ddnet] 23:06 <+bridge> [ddnet] and that msg came up 23:06 <+bridge> [ddnet] but not the one u said 23:06 <+bridge> [ddnet] ok 23:06 <+bridge> [ddnet] oh I wonder what event we get 23:07 <+bridge> [ddnet] can you print Event.window.event before we try the other thing? 23:07 <+bridge> [ddnet] [2020-04-07 16:07:18][client]: WARNING: netversion hash differs 23:07 <+bridge> [ddnet] EVENT 4 23:07 <+bridge> [ddnet] EVENT 1 23:07 <+bridge> [ddnet] EVENT 3 23:07 <+bridge> [ddnet] EVENT 10 23:07 <+bridge> [ddnet] EVENT 12 23:07 <+bridge> [ddnet] [2020-04-07 16:07:18][engine/mastersrv]: saving addresses 23:07 <+bridge> [ddnet] [2020-04-07 16:07:18][datafile]: loading data index=6 size=76 uncompressed=152 23:07 <+bridge> [ddnet] [2020-04-07 16:07:18][datafile]: loading data index=7 size=57 uncompressed=152 23:07 <+bridge> [ddnet] [2020-04-07 16:07:18][datafile]: loading data index=8 size=76 uncompressed=152 23:07 <+bridge> [ddnet] [2020-04-07 16:07:18][datafile]: loading data index=9 size=4831 uncompressed=31768 23:07 <+bridge> [ddnet] EVENT 11 23:07 <+bridge> [ddnet] EVENT 13 23:07 <+bridge> [ddnet] on launch 23:07 <+bridge> [ddnet] on clicking the "play" button: 23:07 <+bridge> [ddnet] EVENT 12 23:07 <+bridge> [ddnet] EVENT 10 23:07 <+bridge> [ddnet] [2020-04-07 16:07:27][pFont]: memory usage: 3145728 23:07 <+bridge> [ddnet] [2020-04-07 16:07:27][pFont]: memory usage: 4194304 23:07 <+bridge> [ddnet] EVENT 11 23:07 <+bridge> [ddnet] EVENT 13 23:08 <+bridge> [ddnet] on joining a server: 23:08 <+bridge> [ddnet] EVENT 12 23:08 <+bridge> [ddnet] EVENT 10 23:08 <+bridge> [ddnet] EVENT 11 23:08 <+bridge> [ddnet] EVENT 13 23:08 <+bridge> [ddnet] EVENT 12 23:08 <+bridge> [ddnet] EVENT 10 23:08 <+bridge> [ddnet] [2020-04-07 16:07:40][client]: disconnecting. reason='unknown' 23:08 <+bridge> [ddnet] 23:08 <+bridge> [ddnet] ... 23:08 <+bridge> [ddnet] [2020-04-07 16:07:41][client/network]: loading done 23:08 <+bridge> [ddnet] EVENT 11 23:08 <+bridge> [ddnet] EVENT 13 23:08 <+bridge> [ddnet] [2020-04-0 23:08 <+bridge> [ddnet] that's for windowevent only right? 23:08 <+bridge> [ddnet] case SDL_WINDOWEVENT: 23:08 <+bridge> [ddnet] printf("EVENT %d\n", Event.window.event); 23:08 <+bridge> [ddnet] 23:08 <+bridge> [ddnet] ye 23:09 <+bridge> [ddnet] okay, this will be horrible for performance but lets see if it helps 23:09 <+bridge> [ddnet] instead of that printf put `Graphics()->UpdateScreenSize();` there 23:10 <+bridge> [ddnet] windowevent? 23:10 <+bridge> [ddnet] yep 23:10 <+bridge> [ddnet] on any window event, just update the screensize 23:10 <+bridge> [ddnet] printf or just update 23:10 <+bridge> [ddnet] don't printf, just that graphics command 23:11 <+bridge> [ddnet] ok 23:11 <+bridge> [ddnet] fanspinny but clipping goes awy 23:11 <+bridge> [ddnet] :o 23:11 <+bridge> [ddnet] Who would have guessed a race condition in macOS SDL 23:12 <+bridge> [ddnet] can you check which version of SDL it's running with? 23:12 <+bridge> [ddnet] it should be output in the console 23:12 <+bridge> [ddnet] [2020-04-07 16:07:17][sdl]: SDL version 2.0.8 (dll = 2.0.8) 23:13 <+bridge> [ddnet] I'd ask you to test with a newer version, but the pain of doing library work on macOS is not something I'd wish on anyone 23:13 <+bridge> [ddnet] shouldnt it ultimately work no matter what version someone has 23:13 <+bridge> [ddnet] well no, if SDL itself has bugs 23:14 <+bridge> [ddnet] there are many macOS bugs in SDL 23:14 <+bridge> [ddnet] ripp 23:15 <+bridge> [ddnet] I wonder when SDL_GL_GetDrawable starts returning the correct size 23:15 <+bridge> [ddnet] if we knew at which point the resolution is correct, we could make it work for all versions 23:16 <+bridge> [ddnet] how to test 23:16 <+bridge> [ddnet] @onby what is in client.m in that patch btw? 23:16 <+bridge> [ddnet] is it my updated version? that might help 23:16 <+bridge> [ddnet] https://cdn.discordapp.com/attachments/293493549758939136/697193227672879144/client.m 23:17 <+bridge> [ddnet] ah yes, that's what's breaking things 23:18 <+bridge> [ddnet] wait oy merged my pull 23:18 <+bridge> [ddnet] how do you still have that client.m? 😄 23:18 <+bridge> [ddnet] oh that commit has that client.m 23:18 <+bridge> [ddnet] I'm stupid 23:19 <+bridge> [ddnet] lol 23:19 <+bridge> [ddnet] `git cherry-pick 7f602078e` 23:19 <+bridge> [ddnet] I think that should work 23:19 <+bridge> [ddnet] nobody1s-MacBook-Pro:teeworlds nobody1$ git cherry-pick 7f602078e 23:19 <+bridge> [ddnet] error: commit 7f602078e4d9543f4182ed63654ad4dd20b528f0 is a merge but no -m option was given. 23:19 <+bridge> [ddnet] fatal: cherry-pick failed 23:19 <+bridge> [ddnet] uhm wait are we on a detached head? 23:20 <+bridge> [ddnet] `git checkout -b testbranch` 23:20 <+bridge> [ddnet] `git merge l2/tw_pr_sdldelegate` 23:23 <+bridge> [ddnet] wat am i testing here 23:23 <+bridge> [ddnet] @onby remove the updatescreensize thing on WINDOWEVENT 23:23 <+bridge> [ddnet] add a dbg message to WINDOWEVENT_RESIZED THINGOMAJIG 23:24 <+bridge> [ddnet] SDL_WINDOWEVENT_SIZE_CHANGED this one 23:24 <+bridge> [ddnet] highdpi 0 and 1 both should work 23:24 <+bridge> [ddnet] hopefully 23:24 <+bridge> [ddnet] or I'll just cry 23:25 <+bridge> [ddnet] case SDL_WINDOWEVENT: 23:25 <+bridge> [ddnet] Graphics()->UpdateScreenSize(); 23:25 <+bridge> [ddnet] printf("EVENT %d\n", Event.window.event); 23:25 <+bridge> [ddnet] switch(Event.window.event) 23:25 <+bridge> [ddnet] { 23:25 <+bridge> [ddnet] #if defined(CONF_PLATFORM_MACOSX) // Todo SDL: remove this when fixed (mouse state is faulty on start) 23:25 <+bridge> [ddnet] case SDL_WINDOWEVENT_MAXIMIZED: 23:25 <+bridge> [ddnet] MouseModeAbsolute(); 23:25 <+bridge> [ddnet] MouseModeRelative(); 23:25 <+bridge> [ddnet] break; 23:25 <+bridge> [ddnet] #endif 23:25 <+bridge> [ddnet] case SDL_WINDOWEVENT_SIZE_CHANGED: 23:25 <+bridge> [ddnet] dbg_msg("DEBUG", "o/"); 23:25 <+bridge> [ddnet] Graphics()->UpdateScreenSize(); 23:25 <+bridge> [ddnet] break; 23:25 <+bridge> [ddnet] } 23:25 <+bridge> [ddnet] break; 23:25 <+bridge> [ddnet] this is already there 23:25 <+bridge> [ddnet] should i just comment out the updatescreensize 23:25 <+bridge> [ddnet] comment out the first updatescreensize 23:25 <+bridge> [ddnet] the rest can stay the same 23:26 <+bridge> [ddnet] mm 23:26 <+bridge> [ddnet] clipping was broken for you when highdpi was 0 right? 23:26 <+bridge> [ddnet] yes 23:26 <+bridge> [ddnet] and either i fcked something up or its still broken 23:26 <+bridge> [ddnet] is it still broken like this? 23:26 <+bridge> [ddnet] wait 23:26 <+bridge> [ddnet] also did we get a o/ in the log? 😄 23:27 <+bridge> [ddnet] no :X 23:27 <+bridge> [ddnet] okay so we don't get a o/ which means we didn't get lucky 23:28 <+bridge> [ddnet] I hoped it was just the broken event loop in that client.m which was not allowing the resize event through 23:28 <+bridge> [ddnet] cant it just updatescreensize less often 23:28 <+bridge> [ddnet] but I guess not, debugging any further is just too hard 23:28 <+bridge> [ddnet] since that seemed to fix it 23:28 <+bridge> [ddnet] rip 23:29 <+bridge> [ddnet] @onby well I know one more place we can check 23:29 <+bridge> [ddnet] uncomment that top updatescreensize 23:29 <+bridge> [ddnet] find updatescreensize, and add a dbg msg above and below GL_GetDrawablesize 23:30 <+bridge> [ddnet] wait wut 23:30 <+bridge> [ddnet] case SDL_WINDOWEVENT: 23:30 <+bridge> [ddnet] Graphics()->UpdateScreenSize(); 23:30 <+bridge> [ddnet] printf("EVENT %d\n", Event.window.event); 23:30 <+bridge> [ddnet] switch(Event.window.event) 23:30 <+bridge> [ddnet] so do that again in input 23:30 <+bridge> [ddnet] and put back the dbg messages in backend_sdl? 23:30 <+bridge> [ddnet] void CGraphicsBackend_SDL_OpenGL::GetScreenSize(int *pScreenWidth, int *pScreenHeight) 23:30 <+bridge> [ddnet] { 23:30 <+bridge> [ddnet] SDL_GL_GetDrawableSize(m_pWindow, pScreenWidth, pScreenHeight); 23:30 <+bridge> [ddnet] } 23:30 <+bridge> [ddnet] 23:30 <+bridge> [ddnet] void CCommandProcessorFragment_SDL::Cmd_Resize(const CCommandBuffer::SCommand_Resize *pCommand) 23:30 <+bridge> [ddnet] { 23:30 <+bridge> [ddnet] SDL_SetWindowSize(m_pWindow, pCommand->m_Width, pCommand->m_Height); 23:30 <+bridge> [ddnet] int Width, Height; 23:30 <+bridge> [ddnet] SDL_GL_GetDrawableSize(m_pWindow, &Width, &Height); 23:30 <+bridge> [ddnet] glViewport(0, 0, Width, Height); 23:31 <+bridge> [ddnet] } 23:31 <+bridge> [ddnet] 23:31 <+bridge> [ddnet] 23:31 <+bridge> [ddnet] SDL_GetWindowSize(m_pWindow, pWindowWidth, pWindowHeight); 23:31 <+bridge> [ddnet] SDL_GL_GetDrawableSize(m_pWindow, pScreenWidth, pScreenHeight); // drawable size may differ in high dpi mode 23:31 <+bridge> [ddnet] there are three instances of getdrawablesize 23:31 <+bridge> [ddnet] oh wait there is no GL_Getdrawablesize there 23:31 <+bridge> [ddnet] which one should i add dbg to 23:31 <+bridge> [ddnet] okay it's a little different then I remembered, go to L929 instead 23:32 <+bridge> [ddnet] put the dbg above and below m_pBackend->GetScreenSize 23:32 <+bridge> [ddnet] wut 23:32 <+bridge> [ddnet] in which file? 23:32 <+bridge> [ddnet] neither one are that long 23:32 <+bridge> [ddnet] graphics_threaded.cpp 23:32 <+bridge> [ddnet] oh 23:33 <+bridge> [ddnet] this really is the last thing we can check, after this we need to hook into or debug sdl, or go even further and debug AppKit, which is not fun nor something I can really walk you through given I wouldn't even know where I myself would begin 😛 23:34 <+bridge> [ddnet] [2020-04-07 16:33:34][client]: WARNING: netversion hash differs 23:34 <+bridge> [ddnet] 0: 2560 1600 23:34 <+bridge> [ddnet] 1: 1280 800 23:34 <+bridge> [ddnet] EVENT 4 23:34 <+bridge> [ddnet] 0: 1280 800 23:34 <+bridge> [ddnet] 1: 1280 800 23:34 <+bridge> [ddnet] EVENT 1 23:34 <+bridge> [ddnet] 0: 1280 800 23:34 <+bridge> [ddnet] 1: 1280 800 23:34 <+bridge> [ddnet] EVENT 3 23:34 <+bridge> [ddnet] this is with highdpi 0 23:34 <+bridge> [ddnet] oh that's interesting 23:34 <+bridge> [ddnet] how/when did it become 2560 1600? 23:35 <+bridge> [ddnet] ¯\_(ツ)_/¯ 23:35 <+bridge> [ddnet] thats the size of my screen 23:35 <+bridge> [ddnet] I think that's highdpi resolution 23:35 <+bridge> [ddnet] try setting the window smaller, see if it's 2x again 23:36 <+bridge> [ddnet] or if it's the desktop resolution 23:36 <+bridge> [ddnet] the window cant be set smaller thats the whole problem i thoughtr 23:36 <+bridge> [ddnet] didn't we patch that first? 23:37 <+bridge> [ddnet] no 23:37 <+bridge> [ddnet] i change in settings.cfg 23:37 <+bridge> [ddnet] 1024x768 23:37 <+bridge> [ddnet] and it changes it back to 1280x800 23:37 <+bridge> [ddnet] remember that block patch in backend_sdl.cpp? 23:37 <+bridge> [ddnet] L701 702 703 23:37 <+bridge> [ddnet] you can do that again 23:37 <+bridge> [ddnet] that only breaks fullscreen toggle which we won't do 23:38 <+bridge> [ddnet] oh the line numbers might have changed with the patches 23:38 <+bridge> [ddnet] ohhh 23:38 <+bridge> [ddnet] wow 23:38 <+bridge> [ddnet] finally 23:39 <+bridge> [ddnet] what does the console look like? 23:39 <+bridge> [ddnet] 0: 1600 1200 23:39 <+bridge> [ddnet] 1: 800 600 23:39 <+bridge> [ddnet] EVENT 1 23:39 <+bridge> [ddnet] 0: 800 600 23:39 <+bridge> [ddnet] 1: 800 600 23:39 <+bridge> [ddnet] EVENT 3 23:39 <+bridge> [ddnet] 0: 800 600 23:39 <+bridge> [ddnet] 1: 800 600 23:39 <+bridge> [ddnet] see that 1600 1200 is verywrong 23:39 <+bridge> [ddnet] this is with highdpi 0 right? 23:39 <+bridge> [ddnet] i put 800x600 23:39 <+bridge> [ddnet] yes 23:40 <+bridge> [ddnet] that initial 1600 1200 is what breaks the clipping 23:40 <+bridge> [ddnet] if we figure out where that comes from, that's a bug fixed 23:41 <+bridge> [ddnet] You can check the return of SDL_GL_GetDrawableSize in Init, hopefully it's there 23:42 <+bridge> [ddnet] where 23:43 <+bridge> [ddnet] in backend_sdl.cpp 23:45 <+bridge> [ddnet] /Users/nobody1/Downloads/teeworlds/src/engine/client/backend_sdl.cpp:754:26: error: cannot pass expression of type 'void' to variadic function; expected type from format string 23:45 <+bridge> [ddnet] was 'int' 23:45 <+bridge> [ddnet] printf("returned %d\n", SDL_GL_GetDrawableSize(m_pWindow, pScreenWidth, pScreenHeight)); // drawable size may differ in high dpi mode 23:45 <+bridge> [ddnet] it doenst return anything 23:45 <+bridge> [ddnet] lol not like that 23:46 <+bridge> [ddnet] that function returns in pScreenWidth, pScreenHeight 23:46 <+bridge> [ddnet] oh just 23:46 <+bridge> [ddnet] ok 23:46 <+bridge> [ddnet] so print *pScreenWidth, *pScreenHeight after calling it 23:46 <+bridge> [ddnet] [2020-04-07 16:46:12][client]: starting... 23:46 <+bridge> [ddnet] 800 600 23:47 <+bridge> [ddnet] see at some point, the resolution becomes 1600 1200, idk how that is possible 23:48 <+bridge> [ddnet] we only seem to modify m_ScreenHeight and m_ScreenWidth there, except for code I added for the runtime resizing 23:48 <+bridge> [ddnet] and all of my modifications of it go through UpdateScreenSize 23:48 <+bridge> [ddnet] Well if you are up for it we can try a watchpoint tomorrow 23:49 <+bridge> [ddnet] ok 23:49 <+bridge> [ddnet] if u tell me how 23:50 <+bridge> [ddnet] watchpoints are very nice debugging tools 23:50 <+bridge> [ddnet] they watch a section of memory for change and if it changes it breaks so you can see what changed it 23:52 <+bridge> [ddnet] oo cool 23:52 <+bridge> [ddnet] that sounds like it would be useful for this 23:52 <+bridge> [ddnet] sleep is required 23:52 <+bridge> [ddnet] cya @onby, thanks for sticking around and debugging this 23:53 <+bridge> [ddnet] np :D 23:53 <+bridge> [ddnet] leme know if u want to try more tmrw