Guild icon
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 2021-08-23 00:00:00Z and 2021-08-24 00:00:00Z
Avatar
GitHub BOT 2021-08-23 04:30:26Z
Author icon
ddnet-maps
[ddnet-maps:master] 1 new commit
da87618 A Broken World - ddnet-maps
Avatar
gerdoe 2021-08-23 07:05:04Z
Image attachment
Avatar
Deleted User 2021-08-23 09:29:30Z
@deen do you compile SDL yourself?
09:29
for ddnet-libs
Avatar
deen 2021-08-23 09:29:48Z
yes
09:29
Should I update it to 2.0.16?
Avatar
Deleted User 2021-08-23 09:30:01Z
ok, not yet, but i made a pr on sdl
Avatar
deen 2021-08-23 09:30:06Z
I heard some people had problems
Avatar
Deleted User 2021-08-23 09:30:08Z
if that goes in we can give it a try
Avatar
deen 2021-08-23 09:30:09Z
yeah, I can add your PR
Avatar
Deleted User 2021-08-23 09:30:34Z
yeah but i'd prefer the SDL devs atleast give a small feedback, who knows if that works on every OS XD
09:30
only tested it on windows and linux
09:31
@nori all SDL problems solved under wayland?
Avatar
nori 2021-08-23 09:40:54Z
for me it's works on sway
09:41
i don't know how it is for gnome/kde
Avatar
Deleted User 2021-08-23 09:42:06Z
ok i'll try it too in a few minutes, and check if glews pr works to convince the glew guy to merge https://github.com/nigels-com/glew/pull/216 Then we also wouldn't need EGL hacks probably
Extend GLEW API to support glewContextInit - GL without implied GLX...
For Issue #172 Documentation needs updating
Thumbnail
Avatar
GitHub BOT 2021-08-23 09:52:23Z
Author icon
Jupeyy
[ddnet/ddnet] Issue closed: #3953 LCTRL equals ALT GR
09:53
Author icon
Jupeyy
[ddnet/ddnet] Issue closed: #4004 Can't build client using MSVS
09:59
Author icon
Jupeyy
[ddnet/ddnet] Issue closed: #3871 Windows XP build with discord SDK fails to open
10:06
Author icon
Jupeyy
[ddnet/ddnet] Pull request opened: #4110 Remove Windows XP code

Checklist

  • [ ] Tested the change ingame
  • [ ] Provided screenshots if it is a visual change
  • [ ] Tested in combination with possibly related configuration options
  • [ ] Written a unit test if it works standalone, system.c especially
  • [ ] Considered possible null pointers and out of bounds array indexing
  • [ ] Changed no physics that affect existing maps
  • [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--und...
Avatar
GitHub BOT 2021-08-23 10:19:44Z
Author icon
bors[bot]
[ddnet:master] 2 new commits
c7580b3 Remove Windows XP code - Jupeyy d51f205 Merge #4110 - bors[bot]
10:19
Author icon
bors[bot]
[ddnet/ddnet] Pull request closed: #4110 Remove Windows XP code
Avatar
Ryozuki 2021-08-23 10:22:36Z
can someone merge/review? #4103 😀
Avatar
DDNet BOT 2021-08-23 10:22:36Z
https://github.com/ddnet/ddnet/pull/4103
Fix swap resets switches. by edg-l · Pull Request #4103 · ddnet/ddn...
Fixes #3806 Checklist Tested the change ingame Provided screenshots if it is a visual change Tested in combination with possibly related configuration options Written a unit test if it works ...
Thumbnail
Avatar
Avatar
GitHub
Click to see attachment 🖼️
gerdoe 2021-08-23 10:29:00Z
unsupport macos when xd
Avatar
Deleted User 2021-08-23 10:49:40Z
i cant run macos builds from github :c
Avatar
Avatar
Ryozuki
can someone merge/review? #4103 😀
Deleted User 2021-08-23 10:53:35Z
the problem is, you cannot really test every case, it will be a trial and error again anyway the switch bug was already not really visible
Avatar
Ryozuki 2021-08-23 10:54:26Z
idk
10:54
but this fixes a known issue
Avatar
Deleted User 2021-08-23 10:55:24Z
just sayan
10:55
saiyan
Avatar
Ryozuki 2021-08-23 10:55:44Z
super sayan
Avatar
Deleted User 2021-08-23 10:55:51Z
<-
Avatar
Ryozuki 2021-08-23 10:56:06Z
xD
Avatar
GitHub BOT 2021-08-23 11:11:28Z
Author icon
Robyt3
[ddnet/ddnet] Pull request opened: #4111 Fix filesystem/IO with unicode filenames on windows
It was previously not possible to load a map or really any file with a filename containing unicode, e.g. german umlauts. Now all filenames are converted from multibyte to widechar on windows, and the correct functions and structs (with the W suffix) are used consistently. Some changes are made to synchronize ddnet's implementation with upstream (once https://github.com/teeworlds/teeworlds/pull/2936 has been merged there):
  • The unused and not working IOFLAG_RANDOM is removed (on up...
Avatar
Deleted User 2021-08-23 11:15:09Z
^ this is why we need c++20 xd
11:15
u8"path"
11:15
finished
Avatar
Learath2 2021-08-23 11:16:12Z
I love that Robyt is contributing to ddnet too
11:16
He is great at refactors, would be great to have him around
Avatar
Ryozuki 2021-08-23 11:16:41Z
who is robyt?
Avatar
Deleted User 2021-08-23 11:16:44Z
thats for sure
Avatar
Ryozuki 2021-08-23 11:16:49Z
is he someone famous
Avatar
Deleted User 2021-08-23 11:17:01Z
he's the last man standing vanilla contributor xd
Avatar
Ryozuki 2021-08-23 11:17:05Z
xd
Avatar
Avatar
Ryozuki
is he someone famous
Learath2 2021-08-23 11:17:21Z
he has been doing some great work on 0.7 very clean commits, not clinging to staying merge-safe like others
11:17
Downstream client authors haven't been very happy with it but teeworlds is a mess in need of cleaning
Avatar
Ryozuki 2021-08-23 11:18:52Z
ye
11:18
but i dont see his prs merged
Avatar
Deleted User 2021-08-23 11:19:19Z
now comes the oy drama
Avatar
Ryozuki 2021-08-23 11:19:23Z
ah well some are
11:19
Image attachment
11:19
monkalaugh
Avatar
Learath2 2021-08-23 11:20:17Z
Nah, oy has been very receptive to his stuff. I honestly expected most to never get merged from how extensive the diffs were
Avatar
Deleted User 2021-08-23 11:20:45Z
i guess aslong as they aren't changing physics or opengl or cpu architecture he is fine xD
11:20
or c++1
11:20
1
11:20
or stl
11:21
so basiucally stay at 1999
11:22
robyt3 should overtake teeworlds vanilla, and just program as he wants
11:22
then you get a nice teeworlds after 1-2 years
Avatar
Learath2 2021-08-23 11:22:53Z
Seeing as I'm having trouble focusing on math maybe I should finally get rid of base/tl
Avatar
Deleted User 2021-08-23 11:23:08Z
😄
11:23
c++20 lets go
Avatar
Learath2 2021-08-23 11:23:37Z
I think we are doing 14 in ddnet for now, right?
Avatar
Deleted User 2021-08-23 11:23:44Z
11
11:23
17 would add filesystem support
11:23
c++20 adds char8_t
11:24
then the pr above would be 20 lines
11:24
instead of so many
11:24
just saiyan
Avatar
Learath2 2021-08-23 11:24:53Z
Maybe we can get to 17 now so I can finally get rid of the ugly void_t mess in https://github.com/ddnet/ddnet/blob/master/datasrc/seven/compile.py#L121-L134
Avatar
Deleted User 2021-08-23 11:25:09Z
wtf
11:25
never noticed it yet
Avatar
Learath2 2021-08-23 11:26:07Z
only instance of template magic in ddnet I guess 😛
11:26
c++ template<class T, typename std::enable_if<!protocol7::is_sixup<T>::value, int>::type = 0> inline int SendPackMsg(T *pMsg, int Flags, int ClientID)
11:27
Allows this mess
Avatar
Ryozuki 2021-08-23 11:28:01Z
char8_t?
Avatar
Deleted User 2021-08-23 11:28:50Z
char8_t - type for UTF-8 character representation, required to be large enough to represent any UTF-8 code unit (8 bits). It has the same size, signedness, and alignment as unsigned char (and therefore, the same size and alignment as char and signed char), but is a distinct type.
11:28
https://en.cppreference.com/w/cpp/language/types#char8_t
cpp/language/types
Avatar
Ryozuki 2021-08-23 11:29:09Z
omaigat
11:29
unsigned char, char, signed char, char8_t
Avatar
GitHub BOT 2021-08-23 11:29:56Z
Author icon
def-
[ddnet-web:master] 1 new commit
13b76ba Add donation by SP | Someone - def-
Avatar
Deleted User 2021-08-23 11:29:57Z
but its nice bcs std::is_same can be used with constrains to differ between math and characters
Avatar
GitHub BOT 2021-08-23 11:30:05Z
Author icon
Learath2
[ddnet/ddnet] Pull request opened: #4112 Clean up CDoor

Checklist

  • [x] Tested the change ingame
  • [ ] Provided screenshots if it is a visual change
  • [ ] Tested in combination with possibly related configuration options
  • [ ] Written a unit test if it works standalone, system.c especially
  • [ ] Considered possible null pointers and out of bounds array indexing
  • [ ] Changed no physics that affect existing maps
  • [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--u...
Avatar
Ryozuki 2021-08-23 11:32:16Z
why are "string" literals in c char and not unsigned char?
Avatar
Deleted User 2021-08-23 11:32:28Z
good question
11:32
next question xd
Avatar
Ryozuki 2021-08-23 11:32:45Z
in rust u have strings they are made up of u8
11:32
iirc
11:32
xd
11:33
and char type
11:33
The char type represents a single character. More specifically, since ‘character’ isn’t a well-defined concept in Unicode, char is a ‘Unicode scalar value’, which is similar to, but not the same as, a ‘Unicode code point’.
Avatar
Deleted User 2021-08-23 11:33:43Z
fancy
11:33
thats the benefit of rust, clean designed (edited)
Avatar
Ryozuki 2021-08-23 11:33:57Z
ye
11:34
i always feel weird casting a char* to unsigned char*
Avatar
Avatar
Ryozuki
why are "string" literals in c char and not unsigned char?
Learath2 2021-08-23 11:36:05Z
To allow implementation flexibility I guess
Avatar
Ryozuki 2021-08-23 11:36:39Z
whathever that means
11:36
wathever
11:36
idk how to write
Avatar
Deleted User 2021-08-23 11:36:57Z
char is unsigned under arm xd
Avatar
Learath2 2021-08-23 11:37:18Z
e.g. afaik older ARM machines didn't have signed support, so you get to avoid a sign extend if your chars are unsigned there
Avatar
Deleted User 2021-08-23 11:37:48Z
i just recently discovered that the ddnet github wiki even mentions how to build for arm32 https://github.com/ddnet/ddnet/wiki/armv6h-server
armv6h server · ddnet/ddnet Wiki
DDraceNetwork, a cooperative racing mod of Teeworlds - armv6h server · ddnet/ddnet Wiki
Thumbnail
Avatar
Learath2 2021-08-23 11:41:06Z
Working with unicode strings is absolute pain in C and C++
11:42
You get used to it but after using modern languages with stdlibs designed in this decade it's always a bit of a bummer when I need to iterate through utf8 code points by hand
Avatar
Deleted User 2021-08-23 11:43:00Z
honestly i generally dont like working with utf8 if you dont program a text editor you can also just use an decoded utf8 string, so directly work in utf32 or whatever that is and dont care about the few extra bytes lost in your RAM
Avatar
Learath2 2021-08-23 11:43:16Z
EW
Avatar
Deleted User 2021-08-23 11:43:22Z
therefore you have direct access in your memory array
11:43
the offsets are all equal
11:43
and only encode to utf8 when needed
11:43
e.g. networking etc.
Avatar
Learath2 2021-08-23 11:43:45Z
the absolute wastefulness
Avatar
Deleted User 2021-08-23 11:43:49Z
as said
11:43
if you dont work on text editor
11:43
it doesnt matter
Avatar
Learath2 2021-08-23 11:43:59Z
You are quadrupling memory use 😄
Avatar
Deleted User 2021-08-23 11:44:06Z
wchar_t is 32bit under linux
11:44
ok
Avatar
Learath2 2021-08-23 11:44:39Z
Which is why you wouldn't find any wchars in any code I write 😛
Avatar
Deleted User 2021-08-23 11:45:07Z
max ddnet chat message = 256 chars, maybe 10 total 2560 bytes * 4 = 10 240
11:45
its nothing
11:45
10kb and you have no disadvantes of utf8
11:45
decoding utf8 on fly wastes cpu performance
11:45
and freetype only takes decoded anyway
Avatar
Avatar
Deleted User
decoding utf8 on fly wastes cpu performance
Learath2 2021-08-23 11:46:09Z
I bet prefetch and using the cache line more efficiently more than makes up for it
Avatar
Deleted User 2021-08-23 11:46:11Z
so you decode every code point every time
11:46
i doubt at that small numbers
11:47
make a 256byte unicode string and get a random offset
11:47
and see the performance
11:48
utf8 vs 32
11:48
then we'll know
Avatar
Learath2 2021-08-23 11:49:30Z
It's absolutely trivial to craft a testcase that would make either look worse
Avatar
Deleted User 2021-08-23 11:49:46Z
then make on that makes utf32 look worse
11:49
but in a realistic ddnet task
11:50
bcs that was my condition
Avatar
Learath2 2021-08-23 11:52:09Z
Sure, since you didn't mention a metric it is trivial. 1) Receive any message that is just ascii (most of our messages) 2) Notice that your 256 character message is now taking 1k of memory 3) Notice memory does not grow on trees
Avatar
Deleted User 2021-08-23 11:52:29Z
most players are chinese arent they
Avatar
Learath2 2021-08-23 11:52:52Z
Sure, for them notice your 512b message is now taking 1k, similar
Avatar
Deleted User 2021-08-23 11:52:55Z
so you probs often have 4byte utf8
11:53
yes
11:53
not going to deny there are disadvantages (edited)
11:53
if you proof me wrong, i accept it
Avatar
Learath2 2021-08-23 11:53:37Z
The disregard with which you treat memory usage actually makes me so annoyed
11:53
This attitude is how we ended up with browsers that use the entire system memory
Avatar
Deleted User 2021-08-23 11:54:03Z
i dunno
11:54
i think its unrelated
11:54
browers use utf8 probably
Avatar
Learath2 2021-08-23 11:54:42Z
No not utf32, the attitude of "it's just a few bytes" is what I take issue with
Avatar
Deleted User 2021-08-23 11:54:55Z
but its true
11:55
do you always shrink your std::vector?
11:55
probably no
11:55
so that wastes memory too
Avatar
Learath2 2021-08-23 11:55:33Z
I extensively use linked lists in my code so it's usually not an issue for me
Avatar
Deleted User 2021-08-23 11:55:54Z
well i prefer performance over ram, there is a sweetspot sure
11:56
but lists in many cases have too many disadvantages bcs of the heap allocations
Avatar
Learath2 2021-08-23 11:56:21Z
Modern arena based allocators are extremely good at putting linked list nodes in the same bucket for me
11:57
But if I'm using a vector I do try to reserve if at all possible and if I'm doing bulk insertions I do shrink it afterwards
Avatar
Deleted User 2021-08-23 11:57:44Z
it depends on the usecase anyway, but i want a 2d game as use case
11:58
and there the hotpath should stay away from as much allocations or decoding etc as possible
11:58
but if you want to write a test case and proof me wrong, i'd be glad bcs its better to learn from such things than not 😄
Avatar
Learath2 2021-08-23 11:59:38Z
I mean you constrained me to cases where utf32 is strictly better, so I guess you win. There is no way in hell O(n) indexing is better than O(1) indexing in a 256b string
12:00
Try the same in a 4Mb string and you won't be very glad with your now 16Mb string with now O(n) indexing because it can't fit in even L3 cache
Avatar
Deleted User 2021-08-23 12:01:38Z
possible, i dunno honestly
12:01
depends of if the processor even loads it into l3 cache at all
Avatar
Learath2 2021-08-23 12:04:04Z
Besides indexing is an operation hardly used at all in string manipulation, atleast not without first scanning through the string anyway
Avatar
GitHub BOT 2021-08-23 12:04:47Z
Author icon
bors[bot]
[ddnet:master] 2 new commits
4014385 Clean up CDoor - Learath2 b851d74 Merge #4112 - bors[bot]
12:04
Author icon
bors[bot]
[ddnet/ddnet] Pull request closed: #4112 Clean up CDoor
Avatar
Avatar
Learath2
Downstream client authors haven't been very happy with it but teeworlds is a mess in need of cleaning
gerdoe 2021-08-23 12:05:14Z
yeah and ddnet too
12:05
every tw project xd
Avatar
GitHub BOT 2021-08-23 12:33:00Z
Author icon
fokkonaut
[ddnet/ddnet] Issue opened: #4113 DDNet client crashes
It seems that ddnet client doesnt like my server, as it sometimes crashes there. Just like https://github.com/ddnet/ddnet/issues/3695 there is another crash, someone told me the program showed "free(): invalid size" and then "Aborted" before the crash. Also there is anothet problem where the client hangs and if you dont wait for a few seconds and press nothing, it would crash aswell. I cant reproduce this myself, so I have no idea where these crashes could come from. If anyone here could try...
Avatar
Comrade 2021-08-23 12:34:19Z
what's the point of using char8_t instead of char?
12:34
i'm reading its papers and can't find anything
12:34
https://en.cppreference.com/w/cpp/string/char_traits/cmp
12:34
For the char specialization, eq and lt are defined identically to the built-in operators == and < for type unsigned char (not char).
Avatar
Anime.pdf 2021-08-23 12:36:59Z
giftee_green
Avatar
Deleted User 2021-08-23 12:38:29Z
mathematically speaking there probably isnt any You could code your codebase to distinct between utf8 and ansii template <typename T> concept IsUtf = std::is_same_v<T, char8_t> || std::is_same_v<T, char16_t>; //.... template<typename T> void do_smth(T ...) requires IsUtf<T> { .. do smth for utf } template<typename T> void do_smth(T ...) requires !IsUtf<T> { .. do smth for other types } (edited)
Avatar
Anime.pdf 2021-08-23 12:39:01Z
🧐
Avatar
Comrade 2021-08-23 12:39:55Z
isn't handling everything as UTF-8 the correct way?
Avatar
Deleted User 2021-08-23 12:40:09Z
i doubt
12:40
does utf8 even exist in c
12:40
it exist in c++ <20 but only pretty annoying, and they removed it iirc in c++20
12:41
https://en.cppreference.com/w/cpp/locale/codecvt_utf8
cpp/locale/codecvt utf8
12:41
ok not removed
12:41
deprecated in c++17
12:43
i guess they remove it as soon as there are better utf8 string functions
12:43
however, for the api in the std it might already be useful now
12:45
i guess someone under windows could try what happens if you use a utf8 path with std::filesystem
Avatar
Comrade 2021-08-23 12:45:58Z
std::filesystem::path uses std::wstring on Windows
12:46
so yes, some conversion is needed
12:47
but i still don't see any reason why every char string shouldn't be considered as UTF-8 string 😄
Avatar
Deleted User 2021-08-23 12:47:44Z
yeah but apparently windows doesnt
12:47
have u seen the pr on ddnet?
Avatar
Comrade 2021-08-23 12:47:56Z
which one?
Avatar
Deleted User 2021-08-23 12:48:07Z
#4111
Avatar
DDNet BOT 2021-08-23 12:48:07Z
https://github.com/ddnet/ddnet/pull/4111
Fix filesystem/IO with unicode filenames on windows by Robyt3 · Pul...
It was previously not possible to load a map or really any file with a filename containing unicode, e.g. german umlauts. Now all filenames are converted from multibyte to widechar on windows, and t...
Thumbnail
Avatar
Deleted User 2021-08-23 12:49:55Z
try that pr with char8_t xd i dont know, but i could imagine msvc added support for it
12:50
bcs devs are annoyed by these legacy things
Avatar
Comrade 2021-08-23 12:51:27Z
it looks fine
12:53
i don't think char8_t can fix WinAPI awfulness in any way
Avatar
Deleted User 2021-08-23 12:53:33Z
well u8"my_pathüöäÜÖÄ.txt" is easier
12:53
than converting
12:54
what msvc does under the hood doesnt matter for the dev then
Avatar
Comrade 2021-08-23 12:54:28Z
"my_pathüöäÜÖÄ.txt" => convert it to UTF-16 => pass it to some W WinAPI function
12:54
it's always like that with or without char8_t
Avatar
Deleted User 2021-08-23 12:55:01Z
yeah but since its a distinct type windows can handle it internally
12:55
it knows it expects an utf8 string now
Avatar
Comrade 2021-08-23 12:56:27Z
WinAPI heavily relies on UTF-16 and it cannot be changed for sure 🤔
Avatar
Deleted User 2021-08-23 12:56:32Z
yes
12:56
not winapi must change
12:56
filesystem can implement everything
12:56
under the hood
12:57
imagine it like this u8"ÜÖÄÜÖÄ.txt" works on every system
12:57
standard
Avatar
Comrade 2021-08-23 12:58:25Z
ah so the point is that it's guaranteed to be UTF-8 string, so std::filesystem stuff is forced to do valid UTF-8 => UTF-16 conversion always?
Avatar
Deleted User 2021-08-23 12:58:38Z
i dunno if that stands like that in the standard
12:58
but i guess thats the idea yes
Avatar
Comrade 2021-08-23 13:00:54Z
what if std::filesystem::path file = "ÜÖÄÜÖÄ.txt"; already works as expected? (edited)
Avatar
Deleted User 2021-08-23 13:01:24Z
uff, thats also possible yeah, really dunno
Avatar
Comrade 2021-08-23 13:01:37Z
and if not, it's definitely a bug
Avatar
Deleted User 2021-08-23 13:01:45Z
i could imagine c++ prepares for string functions that automatically convert such stuff to other types
13:01
and makes it a requirement for alot of the API
13:02
and char just doesnt have this requirement
13:02
bcs c didnt gave it it
Avatar
Comrade 2021-08-23 13:03:17Z
so everyone should now pollute their code with char8_t and u8"" junk?
13:03
that's bad
Avatar
Deleted User 2021-08-23 13:03:37Z
depends, how often do you use unicode?
13:03
but if you use, i'd do it yes
13:04
i mean you can probs just use char* and put (char8_t*) before it xd
13:04
for IO stuff
13:04
what you use internally is your choice anyway
Avatar
Comrade 2021-08-23 13:05:04Z
you know, even #pragma once is not standardized, but it just works correctly everywhere
13:05
and code should be readable!
Avatar
Deleted User 2021-08-23 13:05:18Z
yes
13:05
i use pragma once always
Avatar
Comrade 2021-08-23 13:05:56Z
the same applies to char
Avatar
Deleted User 2021-08-23 13:06:05Z
but thats different still
13:06
pragma once is an addition
13:06
not a new way
13:06
its more like you say header guards should work differently than they are designed
13:06
char just isnt designed for it
Avatar
Comrade 2021-08-23 13:08:19Z
hmmmm maybe storing arbitrary byte sequence in char array actually violates some standards, but again, it just works 😄
Avatar
Deleted User 2021-08-23 13:08:27Z
its not like i dont understand your point
13:08
but thats the disadvantage of an old language probably
13:08
not designed for the future 😄
Avatar
Comrade 2021-08-23 13:08:51Z
yes
13:09
it seems that sometimes people who design C++ and people who actually use it don't understand each other
Avatar
Comrade 2021-08-23 13:47:24Z
#include <filesystem> #ifdef _WIN32 #include <locale.h> #endif int main() { #ifdef _WIN32 setlocale(LC_ALL, ".UTF-8"); #endif const bool success = std::filesystem::create_directory("ÜÖÄÜÖÄ"); return (success) ? 0 : 1; }
13:47
this works as expected on both Linux and Windows @Deleted User
13:47
setlocale is needed on Windows
13:48
and it doesn't work with MinGW for some reason
13:48
MSVC is needed
Avatar
Deleted User 2021-08-23 13:48:30Z
mhh
13:48
if that has no side effects.. locale always changes so much stuff
13:48
printf and stuff like that
13:49
but good to know anyway
Avatar
Comrade 2021-08-23 13:49:03Z
but no stupid char8_t 😄
Avatar
Deleted User 2021-08-23 13:50:05Z
but does it work with char8_t?
13:50
@Comrade
13:50
i am curious about that xxd
Avatar
Comrade 2021-08-23 13:50:25Z
it does probably
13:50
let's try
13:50
u8"ÜÖÄÜÖÄ" is so ugly 😄
Avatar
Deleted User 2021-08-23 13:50:45Z
xD
13:51
at some point i got used to suffix and prefix of strings and numbers, so idc anymore xD
13:51
6llu * 3llu xd
13:51
just to silence warnings
Avatar
Comrade 2021-08-23 13:54:31Z
it still works well on Linux, but it doesn't work on Windows 😄
13:54
even setlocale doesn't help
Avatar
Deleted User 2021-08-23 13:54:42Z
ui, thats unexpected
13:54
well thats really sad, guess i need to wait for c++26 then xDD
Avatar
Comrade 2021-08-23 13:55:14Z
or just use good old char 😄
13:56
my MSVC installation is few months old so maybe that's why it doesn't work
Avatar
Deleted User 2021-08-23 13:56:42Z
i dunno, could also be that they dont care yet
13:56
aslong its no requirement
Avatar
Comrade 2021-08-23 13:57:26Z
and MinGW still doesn't work
13:57
tested both Wine and native Windows
Avatar
Deleted User 2021-08-23 13:57:46Z
what gcc does ur mingw use?
13:58
last time i tested c++20 features alot of stuff was still missing
Avatar
Comrade 2021-08-23 13:58:22Z
x86_64-w64-mingw32-g++ (GCC) 10-win32 20210110
13:58
looks quite old
Avatar
Deleted User 2021-08-23 13:58:42Z
yeah, but probs still newer than mine xd
Avatar
Comrade 2021-08-23 13:58:51Z
but it doesn't complain about -std=c++20 at least
Avatar
Deleted User 2021-08-23 13:59:19Z
https://en.cppreference.com/w/cpp/compiler_support#C.2B.2B20_features
cpp/compiler support
13:59
msvc is actually really good
13:59
implemented most stuff
Avatar
Comrade 2021-08-23 14:01:00Z
yes, and STL source code is available finally
14:01
https://github.com/microsoft/STL/blob/472161105d596192194d4715ccad307c6c163b4a/stl/src/filesystem.cpp#L266
Avatar
Deleted User 2021-08-23 14:01:27Z
ah nice
Avatar
Ryozuki 2021-08-23 14:17:43Z
im now fully vaccinated 2/2
14:17
monkalaugh
14:17
greenthing
Avatar
Avatar
Comrade
but i still don't see any reason why every char string shouldn't be considered as UTF-8 string 😄
Ryozuki 2021-08-23 14:20:01Z
in rust a string has to be valid utf8
14:20
but c doesnt have strings anyway
14:20
just buffers of data that end with a \0
14:20
feelsbadman
Avatar
Avatar
Ryozuki
in rust a string has to be valid utf8
Comrade 2021-08-23 14:21:17Z
nice, that's how it should be
Avatar
Ryozuki 2021-08-23 14:21:57Z
thats also why the api for the filesystem doesnt use strings in rust iirc, since strings have to be valid utf8 buth unix paths dont
14:21
xd
Avatar
Learath2 2021-08-23 14:22:14Z
I am here to be the cranky old man, what are we talking about this time?
Avatar
Ryozuki 2021-08-23 14:22:25Z
that i can sense 5g waves
14:22
and that everythiung should be utf8
👍 1
Avatar
Avatar
Ryozuki
that i can sense 5g waves
Learath2 2021-08-23 14:22:56Z
14 days passed after your 2nd vaccine?
Avatar
Ryozuki 2021-08-23 14:23:07Z
no i got it today
14:23
im developing my senses
14:23
greenthing
Image attachment
14:23
imagine having a modern std
14:25
isnt out there some more modernish libc or something
Avatar
Comrade 2021-08-23 14:25:45Z
hmmmm PathBuf and Path 🤔
Avatar
Learath2 2021-08-23 14:25:59Z
Facebook, google and boost all have moderner implementations of std stuff
14:27
What don't you like about std anyway? It is getting fairly modern nowadays, albeit very slowly
Avatar
Ryozuki 2021-08-23 14:27:31Z
iirc the C std doesnt have filesystem stuff
14:27
c++20 does
Avatar
Learath2 2021-08-23 14:28:02Z
Ah, C, yeah C pretty much requires companion libraries
Avatar
Ryozuki 2021-08-23 14:28:43Z
C is so "simple" but thats why it requires either too make the wheel or use libs, and managing libs is a pain
14:28
xd
14:29
did u figure out the cdoor collision stuff?=
Avatar
Learath2 2021-08-23 14:29:45Z
I know how it works now
Avatar
Ryozuki 2021-08-23 14:29:49Z
:o
Avatar
Learath2 2021-08-23 14:30:49Z
It creates some weird pseudo layer, it's extremely odd how it's implemented
Avatar
Ryozuki 2021-08-23 14:49:40Z
oddly i always name the files lol when testing stuff
Image attachment
14:49
just noticed lol
Avatar
Robyt3 2021-08-23 14:50:41Z
are you that one guy hosting a server with "sv_hostname lol"? lol
Avatar
Ryozuki 2021-08-23 14:50:53Z
not hosting a sv rn lol xd
Avatar
Deleted User 2021-08-23 14:51:13Z
maintest.cpp mh.cpp moduletest.o test.c test.cpp test.txt a.out2 compile.sh main.cpp mh2.cpp moduletest.cppm moduletest.pcm test.c.obj test.exe tmp seems like i like mh or *test
Avatar
Learath2 2021-08-23 14:51:14Z
I always have a ~/test 😛
Avatar
Ryozuki 2021-08-23 14:51:28Z
@Learath2
Image attachment
Avatar
Learath2 2021-08-23 14:52:04Z
Image attachment
Avatar
Ryozuki 2021-08-23 14:52:17Z
@Deleted User moduletest
14:52
where u testing c++20 modules?
Avatar
Deleted User 2021-08-23 14:52:27Z
yeah
14:52
with clang
Avatar
Ryozuki 2021-08-23 14:52:34Z
i think they look nice
Avatar
Deleted User 2021-08-23 14:52:37Z
but thats like 1 year old xd
Avatar
Ryozuki 2021-08-23 14:52:40Z
xd
Avatar
Avatar
Robyt3
are you that one guy hosting a server with "sv_hostname lol"? lol
Deleted User 2021-08-23 14:53:29Z
welcome to ddnet discord! 😄
Avatar
Robyt3 2021-08-23 14:53:46Z
hey!
Avatar
Ryozuki 2021-08-23 14:53:46Z
oh he joined
14:53
thats him
14:53
or her
Avatar
Robyt3 2021-08-23 14:54:01Z
I'm usually just lurking around
Avatar
Ryozuki 2021-08-23 14:54:44Z
bluekitty
Avatar
Avatar
Learath2
Click to see attachment 🖼️
Ryozuki 2021-08-23 14:55:16Z
is this zoomed or u rly have big font
14:59
L"rb" monkalaugh
Image attachment
14:59
windows monkalaugh
Avatar
Robyt3 2021-08-23 14:59:48Z
maybe we'll add a mode L"fröüß" later lol
Avatar
Avatar
Ryozuki
is this zoomed or u rly have big font
Learath2 2021-08-23 15:00:25Z
retina display, so hidpi
Avatar
Ryozuki 2021-08-23 15:00:46Z
Image attachment
15:00
i actually dislike hiding the pointer stuff behind typedefs
15:00
its so confusing
Avatar
Learath2 2021-08-23 15:01:18Z
winapi should really just get a reset
Avatar
Ryozuki 2021-08-23 15:01:21Z
this is also a problem with auto iirc
Avatar
Learath2 2021-08-23 15:01:42Z
sometimes it's okay not to support 16bit executables anymore
Avatar
Ryozuki 2021-08-23 15:01:49Z
xd
Avatar
Learath2 2021-08-23 15:02:40Z
I've sent this a few times, but https://devblogs.microsoft.com/commandline/windows-command-line-unicode-and-utf-8-output-text-buffer/ this blogpost really illustrates how insane winapi can get
Rich Turner
Windows Command-Line: Unicode and UTF-8 Output Text Buffer
In this post, we’ll discuss the improvements we’ve been making to the Windows Console’s internal text buffer, enabling it to better store and handle Unicode and UTF-8 text. Posts in the Windows Command-Line series: This list will be updated as more posts are published: Command-Line Backgrounder The Evolution of the Windows Command-Line Inside th...
Thumbnail
Avatar
Ryozuki 2021-08-23 15:02:45Z
@Robyt3 btw ur filesystem io pr doesnt conform with our naming conventions monkaS
Avatar
Robyt3 2021-08-23 15:03:03Z
you mean outside of system.c?
Avatar
Deleted User 2021-08-23 15:03:03Z
typedef has one advantage @Ryozuki char_ptr a,b; are both pointers
Avatar
Robyt3 2021-08-23 15:03:14Z
I though system.c was not following the naming conventions anyway
Avatar
Deleted User 2021-08-23 15:03:15Z
char* a,b; not
Avatar
Ryozuki 2021-08-23 15:03:39Z
iirc we dont have system.c
15:03
we are modern
15:03
we got system.cpp
Avatar
Learath2 2021-08-23 15:03:52Z
"modern" 😄
Avatar
Ryozuki 2021-08-23 15:03:56Z
and i think new code should conform
Avatar
Robyt3 2021-08-23 15:03:57Z
yes, this caused me great pain during rebase
15:04
lol
Avatar
Ryozuki 2021-08-23 15:04:02Z
idk what others think
Avatar
Deleted User 2021-08-23 15:04:16Z
c++23 lets go
Avatar
Avatar
Deleted User
char* a,b; not
Comrade 2021-08-23 15:04:45Z
this should be illegal in both C and C++
Avatar
Avatar
Comrade
this should be illegal in both C and C++
Deleted User 2021-08-23 15:04:55Z
i agree
15:04
i dislike it too
Avatar
Ryozuki 2021-08-23 15:05:01Z
xD
Avatar
Avatar
Ryozuki
idk what others think
Learath2 2021-08-23 15:05:04Z
Hm, this is a good question, I actually didn't think about what to do in system.cpp
Avatar
Ryozuki 2021-08-23 15:05:15Z
imho new code should conform
15:05
i am also for correcting stuff
Avatar
Avatar
Comrade
this should be illegal in both C and C++
Learath2 2021-08-23 15:05:32Z
It is completely sane syntax if you learn C/C++ properly
Avatar
Deleted User 2021-08-23 15:05:43Z
but makes it harder to read
👍 1
Avatar
Robyt3 2021-08-23 15:05:53Z
I wanted to have ddnet and upstream somewhat stay in sync, so I didn't update the style on upstream
Avatar
Ryozuki 2021-08-23 15:05:58Z
char *a, *b; this is valid?
Avatar
Deleted User 2021-08-23 15:06:03Z
yes
Avatar
Robyt3 2021-08-23 15:06:05Z
but I agree that new style would be better to have everywhere
Avatar
Ryozuki 2021-08-23 15:06:07Z
i would always put it in 2 lines anyway
Avatar
Deleted User 2021-08-23 15:06:13Z
yes
Avatar
Ryozuki 2021-08-23 15:06:46Z
Image attachment
15:06
what is this magic number? xd
Avatar
Avatar
Deleted User
but makes it harder to read
Learath2 2021-08-23 15:06:58Z
I will say I never have any trouble with it, you will give me no evidence at all about your claim, you will go on with your day, I'll be pissed for the entire night, so let's agree to disagree
Avatar
Deleted User 2021-08-23 15:07:13Z
wtf xD
Avatar
Avatar
Ryozuki
Click to see attachment 🖼️
Learath2 2021-08-23 15:07:14Z
Windows epoch is before unix epoch
Avatar
Robyt3 2021-08-23 15:07:29Z
1 January 1601
Avatar
Ryozuki 2021-08-23 15:07:32Z
lel
Avatar
Robyt3 2021-08-23 15:07:32Z
according to wikipedia
Avatar
Ryozuki 2021-08-23 15:07:40Z
windows wants to be special always
Avatar
Learath2 2021-08-23 15:08:11Z
Btw I don't remember if we clamped that, what happens if the time is before unix epoch?
Avatar
Ryozuki 2021-08-23 15:08:20Z
monkaS
Avatar
Learath2 2021-08-23 15:08:59Z
Ah I do, t == li.QuadPart
15:09
if it underflows this will fail
Avatar
Comrade 2021-08-23 15:11:17Z
and time_t is still 32-bit on some platforms even today 😐
Avatar
Deleted User 2021-08-23 15:11:31Z
ah yeah, the mingw trouble
Avatar
Comrade 2021-08-23 15:11:39Z
and signed for some reason
Avatar
Learath2 2021-08-23 15:12:00Z
@Robyt3 I think your pr looks fine, I'll give it a more thorough eye over tonight
Avatar
Deleted User 2021-08-23 15:12:24Z
@Learath2 can you also review #4103
Avatar
DDNet BOT 2021-08-23 15:12:24Z
https://github.com/ddnet/ddnet/pull/4103
Fix swap resets switches. by edg-l · Pull Request #4103 · ddnet/ddn...
Fixes #3806 Checklist Tested the change ingame Provided screenshots if it is a visual change Tested in combination with possibly related configuration options Written a unit test if it works ...
Thumbnail
Avatar
Deleted User 2021-08-23 15:13:40Z
@Ryozuki btw sv_swap's default value must also be changed then
Avatar
Ryozuki 2021-08-23 15:14:03Z
aye
15:14
should i do it?
Avatar
Deleted User 2021-08-23 15:14:16Z
i dunno
Avatar
Learath2 2021-08-23 15:14:20Z
oooof, 4103 looks scary, I'll take a look
Avatar
Deleted User 2021-08-23 15:14:35Z
ask deen and learath if they are ready for potential trouble xD
Avatar
Ryozuki 2021-08-23 15:14:53Z
monkaS
15:14
i tested it with switches and teams
Avatar
Learath2 2021-08-23 15:15:03Z
Changing defaults shouldn't break anything
Avatar
Ryozuki 2021-08-23 15:15:05Z
test it with team of 2 and 2+
Avatar
Deleted User 2021-08-23 15:15:11Z
when i tested it, i found the same function as cause
Avatar
Learath2 2021-08-23 15:15:13Z
if you changed the meaning of a value feel free to update the default
Avatar
Deleted User 2021-08-23 15:15:20Z
so yeah, it probably is the right patch
15:15
but who knows, just look at the "new" startline skip xD
15:15
the players always find weird stuff xD
Avatar
Ryozuki 2021-08-23 15:16:02Z
xD
Avatar
Deleted User 2021-08-23 15:16:06Z
SetForceCharacterTeam
15:16
this was the problem wasnt it
Avatar
Ryozuki 2021-08-23 15:16:12Z
ye its the cause
15:16
maybe the if doesnt need to wrap the other 2 functions
Avatar
Deleted User 2021-08-23 15:16:30Z
yeah also dunno xd
Avatar
Learath2 2021-08-23 15:16:51Z
@Ryozuki about your switchstate netobj, do you think maybe we should send it per team instead? With 64p and delta compression it's not that important and allows more flexibility. fokkonaut seemed a bit concerned
Avatar
Ryozuki 2021-08-23 15:18:25Z
i dont understand
15:18
int Status = (int)GameServer()->Collision()->m_pSwitchers[i].m_Status[Team()];
15:18
isnt it alreayd per team? more or less
15:18
xd
Avatar
Learath2 2021-08-23 15:18:52Z
Yes, but you send a snap object per character
Avatar
Ryozuki 2021-08-23 15:19:01Z
ah
15:19
feel free to make the pr
Avatar
Learath2 2021-08-23 15:19:16Z
And you send all players switch states to all players
Avatar
Ryozuki 2021-08-23 15:19:27Z
monkaS
Avatar
Learath2 2021-08-23 15:20:20Z
Mh, not sure if I want to make the change, I feel it's good as is. Do you see any reason?
Avatar
Ryozuki 2021-08-23 15:20:39Z
i dont know
15:20
maybe fokkonaut should explain himself
15:20
lately his comments are "not like this"
15:20
without any explanation xd
15:21
Image attachment
15:21
it triggers me so much xdd
Avatar
Learath2 2021-08-23 15:21:13Z
Yeah, hm
15:21
@fokkonaut what was your concern with the switch state object?
Avatar
chillerdragon BOT 2021-08-23 15:47:13Z
because you do not have an macos? do you want me to test something? (@Deleted User)
Author icon
Jupstar ✪
i cant run macos builds from github :c
2021-08-23 10:52:24Z
Avatar
Deleted User 2021-08-23 15:47:42Z
a friend of a friend has a friend who has macos on an emulator which is totally not me
15:48
but nice reply
15:48
better than the one from discord xd
15:48
but yeah, actually you can try smth chiller
15:49
https://github.com/Jupeyy/ddnet/runs/3399295343
Build software better, together
GitHub is where people build software. More than 65 million people use GitHub to discover, fork, and contribute to over 200 million projects.
Thumbnail
15:49
if you compile yourself, you need to use bundled libs, i tried the glew pr, which we also need
15:51
and if your mac supports it, especially try opengl 2.1 as this loads the most extensions
Avatar
chillerdragon BOT 2021-08-23 15:51:53Z
thats nice :) (@Robyt3)
Author icon
Robyt3#2709
I wanted to have ddnet and upstream somewhat stay in sync, so I didn&#x27;t update the style on upstream
2021-08-23 15:06:09Z
Avatar
Deleted User 2021-08-23 15:52:02Z
if it starts it is already enough, tahts all i need
Avatar
chillerdragon BOT 2021-08-23 15:52:27Z
is mac even supported at all? I can not get any builds working.
Avatar
Deleted User 2021-08-23 15:52:44Z
you mean, if ddnet supports mac?
Avatar
chillerdragon BOT 2021-08-23 15:53:03Z
ya well ok to be fair only tested teeworlds
Avatar
Deleted User 2021-08-23 15:53:17Z
i wouldn't know that macos isnt supported
15:53
but i hope with new SDL it gets less buggy
15:53
bcs the version we have is buggier than the one tsfreddie is testing on
Avatar
[quakenet] ChillerDragon BOT 2021-08-23 15:55:37Z
I am not sure if its me not using mac at all anymore but I feel like apple gets more and more trash.
Avatar
Deleted User 2021-08-23 15:56:03Z
just bcs they spy you and hide it under child porn as excuse?
15:56
no but i cannot judge
Avatar
[quakenet] ChillerDragon BOT 2021-08-23 15:56:34Z
Got so many bugs with the new macbooks at work. And omagawd im not sure who? maybe zsh or the terminal someone hijacks my alt presses so i can not change irc buffers anymore aaa
Avatar
Deleted User 2021-08-23 15:56:39Z
on the emulator it ran normally
15:56
just the github builds dont work
15:56
only from ddnet.tw
Avatar
[quakenet] ChillerDragon BOT 2021-08-23 15:56:53Z
can you compile on the emulator?
Avatar
Deleted User 2021-08-23 15:56:58Z
no
Avatar
[quakenet] ChillerDragon BOT 2021-08-23 15:56:59Z
lemme try to compile ddnet real quick
Avatar
Deleted User 2021-08-23 15:57:03Z
and i also dont want that trouble
15:57
its enough is see learath struggle once per month
Avatar
[quakenet] ChillerDragon BOT 2021-08-23 15:57:24Z
hrhrhr learath the macos guy
15:58
its more than just the spying its usability and availibility that feels worse than i remembered it. I feel like those two things are the main reason one uses mac.
15:58
"it just works"
Avatar
Deleted User 2021-08-23 15:59:06Z
its like with android, users like it, bcs its easy but devs hate it
15:59
bcs its annoying
Avatar
[quakenet] ChillerDragon BOT 2021-08-23 15:59:42Z
ouu android i feel like i will never try it
15:59
im planning to replace my iphone by a pinephone in the near future
16:00
only good thing about mac terminals is that cmd+c and cmd+v works :D
16:00
not like ctrl+shift+c like in linux
16:00
yea ddnet does not build on my machine latest macos
Avatar
[quakenet] deen BOT 2021-08-23 16:01:15Z
ChillerDragon: you see PMs?
Avatar
[quakenet] ChillerDragon BOT 2021-08-23 16:01:43Z
i see that i got one
16:02
but i need my archlinux desktop to open it :D
16:02
https://paste.zillyhuhn.com/0z
16:02
fakin mac
16:02
i use arch btw
Avatar
Ryozuki 2021-08-23 16:04:33Z
i use arch btw
Avatar
Deleted User 2021-08-23 16:05:21Z
hi, im jupstar i have an iphone btw xd
Avatar
[quakenet] ChillerDragon BOT 2021-08-23 16:05:24Z
how can github even build
16:05
also im vegan
Avatar
Deleted User 2021-08-23 16:05:46Z
nice
Avatar
[quakenet] ChillerDragon BOT 2021-08-23 16:05:53Z
im not :D
16:05
i just like saying it
Avatar
Deleted User 2021-08-23 16:06:00Z
i am most of the time xd
Avatar
[quakenet] ChillerDragon BOT 2021-08-23 16:06:01Z
good meme
Avatar
Robyt3 2021-08-23 16:06:02Z
even github macos build fails like 10% of the time randomly
kek 1
Avatar
Deleted User 2021-08-23 16:06:04Z
except when i go with friends
Avatar
[quakenet] ChillerDragon BOT 2021-08-23 16:07:12Z
running ddnet/teeworlds on macOS is cursed anyways isnt it?
16:07
i dont even know what the current ddnet state on my mac is
16:07
https://paste.zillyhuhn.com/0z
Avatar
Deleted User 2021-08-23 16:07:29Z
i dunno, cant be that hard
Avatar
[quakenet] ChillerDragon BOT 2021-08-23 16:07:33Z
oops wrong paste
Avatar
Deleted User 2021-08-23 16:07:41Z
if it comes hard on hard static link everything
Avatar
[quakenet] ChillerDragon BOT 2021-08-23 16:07:53Z
oh no it launches
16:08
but i mean the actual expirience
Avatar
Deleted User 2021-08-23 16:08:12Z
oh, i assume it shouldnt be bad
16:08
except that the hardware maybe is
Avatar
[quakenet] ChillerDragon BOT 2021-08-23 16:08:25Z
tabbing in and out didnt work iirc fonts not being rendered resolution being wrong or zoomed
Avatar
Avatar
Ryozuki
Click to see attachment 🖼️
fokkonaut 2021-08-23 16:08:34Z
Isnt it self explaining? The layer is always switch for switchstates.
Avatar
Deleted User 2021-08-23 16:08:41Z
didnt we fix all this stuff xd
16:08
but tabbing in out sound like SDL bug
16:08
that might be fixed with newer sdl versions
Avatar
Avatar
fokkonaut
Isnt it self explaining? The layer is always switch for switchstates.
gerdoe 2021-08-23 16:09:32Z
doesn't switch layer change it's state like everytime you interact with it
Avatar
chillerdragon BOT 2021-08-23 16:09:44Z
ima_25f8ed5.jpeg
Image attachment
Avatar
gerdoe 2021-08-23 16:09:48Z
best
16:09
XD
16:09
wtf is that
Avatar
Deleted User 2021-08-23 16:09:56Z
is HiDPI on chiller?
Avatar
Ryozuki 2021-08-23 16:10:06Z
chiller when new vid showing u traveling with a laptop with tw
Avatar
gerdoe 2021-08-23 16:10:17Z
today
Avatar
chillerdragon BOT 2021-08-23 16:10:20Z
Yes
Avatar
gerdoe 2021-08-23 16:10:26Z
^^
Avatar
Deleted User 2021-08-23 16:10:26Z
wow weird
Avatar
chillerdragon BOT 2021-08-23 16:10:29Z
On
Avatar
Ryozuki 2021-08-23 16:10:30Z
https://www.youtube.com/watch?v=wKV18A1DluU
Prudrugtiq
Teeworlds in 2020
Avatar
Deleted User 2021-08-23 16:10:35Z
must be the old SDL version too
16:10
i think TSfreddie had no problems
16:10
he tested HiDPI and non
16:11
or broken opengl drivers xD
Avatar
chillerdragon BOT 2021-08-23 16:11:03Z
ima_4e7a1a3.jpeg
Image attachment
16:11
Time to get sdlvm
Avatar
fokkonaut 2021-08-23 16:11:12Z
https://github.com/ddnet/ddnet/pull/4084#issuecomment-903225761 It is not required to predict players from other teams anyways, so it makes sense to only send one set of switch states per character @Learath2
Send switch state to the client by edg-l · Pull Request #4084 · ddn...
This PR is related to #3990 Checklist Tested the change ingame Provided screenshots if it is a visual change Tested in combination with possibly related configuration options Written a unit te...
Thumbnail
Avatar
chillerdragon BOT 2021-08-23 16:11:16Z
Lemme try latest official release to compare
Avatar
Robyt3 2021-08-23 16:11:24Z
I love how drivers just mess everything up anyway
16:11
like NVIDIA somehow makes it impossible to actually disable anti-aliasing
Avatar
Deleted User 2021-08-23 16:12:07Z
are you in windows @Robyt3 ?
16:12
OpenGL + windows = true love
Avatar
Robyt3 2021-08-23 16:12:47Z
yes, windows (edited)
Avatar
Deleted User 2021-08-23 16:12:55Z
https://gist.github.com/Jupeyy/ca2ac80fb9765a9c4370bb9a6b88fb32
OpenGL on DirectX 12 for DDNet
OpenGL on DirectX 12 for DDNet. GitHub Gist: instantly share code, notes, and snippets.
Thumbnail
16:13
you can try this, it works on intel, nvidia and amd i think
16:13
the latest version is nice, but you only have like 60% of the native performance (edited)
Avatar
chillerdragon BOT 2021-08-23 16:13:30Z
im actually not planning a new one was always meant as a duology (@Ryozuki)
Author icon
Ryozuki
chiller when new vid showing u traveling with a laptop with tw
2021-08-23 16:10:07Z
16:15
i also feel like every mac is different :D
16:15
@Deleted User: ok latest ddnet release looks pretty similar
16:16
what differences am i searching for?
16:16
what did you try to fix?
Avatar
Deleted User 2021-08-23 16:16:11Z
i just want to know if it starts
Avatar
chillerdragon BOT 2021-08-23 16:16:21Z
ye it starts
16:16
:D
Avatar
Deleted User 2021-08-23 16:16:25Z
so the glew mantainer knows it works, bcs he asked ppl to test his pr
Avatar
chillerdragon BOT 2021-08-23 16:17:06Z
owo
banhammer 1
banhammer 1
ban 1
ban 1
ban 1
ban 1
ban 1
banhammer 1
bann 1
bann 1
banhammer 1
16:17
can i comment on the pr?
Avatar
Deleted User 2021-08-23 16:17:19Z
https://github.com/nigels-com/glew/pull/216
Extend GLEW API to support glewContextInit - GL without implied GLX...
For Issue #172 Documentation needs updating
Thumbnail
16:17
if you have something to add xD
16:17
spam his emails so he merges it xD
Avatar
chillerdragon BOT 2021-08-23 16:17:36Z
can i write "developing graphic drivers" on my CV now?
Avatar
Deleted User 2021-08-23 16:17:43Z
glew has no compile time version, thats annoying
Avatar
chillerdragon BOT 2021-08-23 16:17:53Z
i liked ur comment
16:18
that should work
Avatar
Deleted User 2021-08-23 16:18:10Z
ez
Avatar
Blade 2021-08-23 16:18:41Z
Anyone know how I can fix this?
Image attachment
Avatar
chillerdragon BOT 2021-08-23 16:18:57Z
i love how i keep getting python crash reports randomly on my mac
16:19
someone trys to run python i dont even know who :D
16:19
maybe vim? idk
16:19
ah yes its vim :D
Avatar
Learath2 2021-08-23 16:20:35Z
I'm finally removing base/tl, Q: How do you think I should handle sorted_array?
16:21
In cases where inserts happen bulk, it's much more efficient to just insert everything and then sort, which is what I did
Avatar
Ryozuki 2021-08-23 16:21:29Z
seems fine
Avatar
Learath2 2021-08-23 16:21:43Z
but for CUuidManager inserts can always happen but currently they happen in bulk
16:22
How should I handle that case? Just bite the bullet and do insertion sort?
Avatar
chillerdragon BOT 2021-08-23 16:27:02Z
@Deleted User: btw ddnet 13.1 works fullscreen
Avatar
Deleted User 2021-08-23 16:27:21Z
nice, then its not the driver
16:27
maybe turn hidpi off xd
Avatar
chillerdragon BOT 2021-08-23 16:27:47Z
xxd
16:28
i love how i get 3 blackscreens
16:28
then a red
16:28
:D
Avatar
Deleted User 2021-08-23 16:29:01Z
4 windows or what?
Avatar
chillerdragon BOT 2021-08-23 16:29:10Z
now it blinks :D
16:29
no*
Avatar
Blade 2021-08-23 16:29:24Z
Would anyone kindly help me fix the "Windows headers require the default packing option. Changing this can lead to memory corruption. This diagnostic can be disabled by building with WINDOWS_IGNORE_PACKING_MISMATCH defined." issue? I'm not familiar with VS so I'm unsure how to change the build variables
Avatar
Deleted User 2021-08-23 16:29:40Z
define WINDOWS_IGNORE_PACKING_MISMATCH
16:29
xD
Avatar
Blade 2021-08-23 16:30:00Z
You'd think I just said I didn't know how
Avatar
Deleted User 2021-08-23 16:30:00Z
where did you even change the packaging option
16:30
did you use pragma pack?
Avatar
Blade 2021-08-23 16:30:24Z
I just did what it said on the git, nothing more, nothing less
Avatar
Deleted User 2021-08-23 16:30:36Z
oh its on ddnet git?
Avatar
Blade 2021-08-23 16:30:42Z
Yes
Avatar
Learath2 2021-08-23 16:31:17Z
Did we forget to pragma pop somewhere?
Avatar
Deleted User 2021-08-23 16:31:25Z
i doubt we use it
16:31
heinrich killed it, when i used it once xxD
Avatar
Avatar
Learath2
How should I handle that case? Just bite the bullet and do insertion sort?
Deleted User 2021-08-23 16:32:38Z
just use std::lower_bound or smth
16:32
but yeah
Avatar
Blade 2021-08-23 16:32:44Z
Is there a stable I can use instead?
Avatar
Deleted User 2021-08-23 16:32:46Z
its still insertion sort
Avatar
Avatar
Blade
Is there a stable I can use instead?
Deleted User 2021-08-23 16:33:12Z
update msvc
16:33
i don't see why this should happening
Avatar
Blade 2021-08-23 16:34:10Z
Updating the build tools should suffice?
Avatar
Deleted User 2021-08-23 16:37:16Z
update everything, delete cmake cache
16:37
the classic
Avatar
Blade 2021-08-23 16:37:45Z
"The classic" first time working with cmake, I'm used to Bam
Avatar
chillerdragon BOT 2021-08-23 16:39:18Z
bam master race
16:39
I use bam btw
Avatar
Deleted User 2021-08-23 16:39:33Z
does bam even support clang?
Avatar
Blade 2021-08-23 16:39:38Z
Is it still possible to build with bam?
Avatar
Deleted User 2021-08-23 16:39:40Z
no
Avatar
chillerdragon BOT 2021-08-23 16:39:44Z
ddnet no
Avatar
Deleted User 2021-08-23 16:39:45Z
and pls dont add it back
Avatar
Blade 2021-08-23 16:39:52Z
Sadge
Avatar
Avatar
Deleted User
and pls dont add it back
gerdoe 2021-08-23 16:39:56Z
troll
Avatar
chillerdragon BOT 2021-08-23 16:39:59Z
petition to add back bam
f3 2
Avatar
Blade 2021-08-23 16:40:05Z
I'm all for it
Avatar
chillerdragon BOT 2021-08-23 16:40:15Z
and ./configure autgen
16:40
and ./configure autogen
Avatar
Deleted User 2021-08-23 16:40:18Z
pls
Avatar
Blade 2021-08-23 16:40:31Z
It was clunky and weird, but it worked
Avatar
Deleted User 2021-08-23 16:40:44Z
cmake works too
16:40
our CI works with MSVC
16:41
i dunno, just use cmake-gui
Avatar
Blade 2021-08-23 16:41:12Z
I am
Avatar
Deleted User 2021-08-23 16:41:22Z
does the configure work?
16:41
no errors
Avatar
Blade 2021-08-23 16:41:29Z
Yeah
16:41
Worked flawlessly, just compiling threw that one error
Avatar
Deleted User 2021-08-23 16:41:46Z
well then its a MSVC problem i guess, we don't change packaging i think
16:41
does it name a file?
16:42
where the microsoft header is included from
Avatar
Blade 2021-08-23 16:42:10Z
I'm updating msvc right now, hopefully it fixes itself (edited)
Avatar
Robyt3 2021-08-23 16:42:11Z
anything that includes windows.h would trigger this warning
Avatar
Blade 2021-08-23 16:42:41Z
What Roby said
16:47
Everything updated, same error. Seems to come from winnt.h
16:47
Line 2482
Avatar
Deleted User 2021-08-23 16:48:21Z
mhh, also updated the windows sdk?
Avatar
Blade 2021-08-23 16:48:32Z
Windows sdk?
Avatar
Deleted User 2021-08-23 16:48:44Z
its probably an option in the MSVC updater
16:48
what even is your MSVC version?
Avatar
Blade 2021-08-23 16:49:20Z
15.9.38
Avatar
Deleted User 2021-08-23 16:49:35Z
"As 1201ProgramAlarm mentioned above, the solution was disabling the /Zp (structure packing) compiler option. "
16:49
https://stackoverflow.com/questions/56570681/static-assertion-failed-with-windows-headers-require-the-default-packing-option
Static assertion failed with "Windows headers require the default p...
When I'm trying to compile my C++ project in Visual Studio, I keep getting the 2 following errors: E1574: Static assertion failed with "Windows headers require the default packing option. Changin...
Thumbnail
16:49
right click ddnet and search that option
Avatar
Blade 2021-08-23 16:52:15Z
Assuming you mean in CMake, I don't see any Zp option
Avatar
Deleted User 2021-08-23 16:52:21Z
in msvc
Avatar
gerdoe 2021-08-23 16:52:39Z
do you click on ALL_BUILD?
Avatar
Blade 2021-08-23 16:53:03Z
No, I build only DDNet
Avatar
gerdoe 2021-08-23 16:53:51Z
kekw
16:53
strange thing
Avatar
Deleted User 2021-08-23 16:54:09Z
why dont u use visual studio 2019 btw
Avatar
gerdoe 2021-08-23 16:55:11Z
2022 incoming xd
Avatar
Deleted User 2021-08-23 16:55:32Z
why dont u use visual studio 2022 btw
Avatar
Blade 2021-08-23 16:57:26Z
Embedded image
16:57
I have no clue what I'm doing wrong
Avatar
Deleted User 2021-08-23 16:57:38Z
disable
16:57
maybe default is on
Avatar
Blade 2021-08-23 16:58:01Z
I can't, I can only choose a value or "default"
16:58
Embedded image
Avatar
Deleted User 2021-08-23 16:58:46Z
mh i dunno then
16:58
google your problem or use a modern version
Avatar
Learath2 2021-08-23 16:59:22Z
Okay, what kind of crackhead at the standards committee decided that std::binary_search should return a boolean?
16:59
It takes a dimwit of monumental stupidness to not return an index or at the very least a damn iterator
17:00
The person who got it in should just quit programming, taking along the people that voted yes to it
justatest 1
Avatar
Blade 2021-08-23 17:03:11Z
Aight, weird. Using VS2019 worked
Avatar
Learath2 2021-08-23 17:05:59Z
I have seriously seen smarter toasters than this person... I'll have to litter the code with lower_bound which is far less descriptive now
Avatar
Deleted User 2021-08-23 17:22:13Z
Return value true if an element equal to value is found, false otherwise.
17:22
does that not sound reasonable?
Avatar
Learath2 2021-08-23 17:30:45Z
It is correct, I have no problem with that, but it's also useless except for checking if something is already in a vector
17:31
A sane implementation would return an iterator to the found value, and the end iterator if not found
Avatar
Deleted User 2021-08-23 17:31:46Z
ah yeah
17:31
you are right xxd
17:32
didnt even notice xxD
Avatar
Learath2 2021-08-23 17:32:58Z
Q: I'm wrapping std::vector to get a sorted vector of our own. Should I inherit and delete the methods that can lead to the vector getting unsorted or should I keep the vector as a private member and only expose methods of my own?
Avatar
gerdoe 2021-08-23 17:35:13Z
second i think
Avatar
Robyt3 2021-08-23 17:35:53Z
yep, composition > inheritance
Avatar
Deleted User 2021-08-23 17:37:09Z
inheritance best xd
17:40
why don't you use boost headers @Learath2 ? most of these template stuff can be used in header only mode
Avatar
gerdoe 2021-08-23 17:42:33Z
replace /base with boost xdd
Avatar
Deleted User 2021-08-23 17:44:34Z
could probably xD, boost has everything
17:44
but header only are safer, bcs they work everywhere
Avatar
gerdoe 2021-08-23 17:46:54Z
don't we include boost for mysql support?
Avatar
trml 2021-08-23 17:53:52Z
looking at the server code, it doesn't look like m_Layer in entity.h is used for much except to check if an entity is in switch. would be ok to remove/refactor out m_Layer, and just use m_Number == 0 when it's not in switch?
Avatar
gerdoe 2021-08-23 18:08:20Z
make an issue or pr
18:08
this chat is fast af xd
Avatar
Deleted User 2021-08-23 18:09:35Z
yes looks safe to me
Avatar
trml 2021-08-23 18:09:54Z
yeah 😛
18:10
the only thing is that reserving Number == 0 is very clean (but not sure if using something like -1 would be better?)
18:10
not very clean*
Avatar
Deleted User 2021-08-23 18:11:07Z
is it even so bad to send the layer?
18:11
or if its always a switch layer just set it when the packet comes in?
Avatar
trml 2021-08-23 18:12:06Z
I don't think so, since this won't be sent often
18:13
but if we are certain that m_Layer will never be used for anything else in the future (and can always be inferred from m_Number) it's perhaps not needed
Avatar
Deleted User 2021-08-23 18:13:49Z
you can never be certain xd
18:14
if somebody ever creates a new switch layer or smth
Avatar
trml 2021-08-23 18:14:21Z
yeah, thats my thought:P
Avatar
Deleted User 2021-08-23 18:16:34Z
then argument like this in the pr, and its fine, either the ppl comment on that give better explaination or it doesn't help xD
Avatar
trml 2021-08-23 18:17:29Z
yes, will comment
Avatar
Learath2 2021-08-23 18:36:37Z
Hm, if only we were up to C++20 this would be so much prettier
Avatar
Learath2 2021-08-23 18:55:36Z
@Deleted User https://github.com/ddnet/ddnet/commit/0e71a4770 why did you use a sorted array here? You only insert unsorted into it and you never sort it
Add customization · ddnet/ddnet@0e71a47
Thumbnail
Avatar
gerdoe 2021-08-23 18:56:36Z
xd
Avatar
Deleted User 2021-08-23 18:56:41Z
doesnt it automatically sort?
Avatar
Learath2 2021-08-23 18:56:51Z
Not if you insert unsorted 😛
Avatar
Deleted User 2021-08-23 18:57:23Z
shitty library then xd
Avatar
gerdoe 2021-08-23 18:57:37Z
u
Avatar
Learath2 2021-08-23 18:58:01Z
Ehe? You find a function literally called add_unsorted and you expect it to sort somehow? How? by magic? 😄
Avatar
Ryozuki 2021-08-23 18:58:09Z
monkalaugh
Avatar
Avatar
Learath2
Ehe? You find a function literally called add_unsorted and you expect it to sort somehow? How? by magic? 😄
gerdoe 2021-08-23 18:58:25Z
XDD
Avatar
Deleted User 2021-08-23 18:58:26Z
but where is that even called
Avatar
Learath2 2021-08-23 18:58:44Z
Um, since no one ever noticed that these arrays are unsorted I guess they can be replaced by vectors
Avatar
Deleted User 2021-08-23 18:59:09Z
im not so sure
Avatar
Avatar
Deleted User
but where is that even called
Learath2 2021-08-23 18:59:14Z
menus_settings_assets.cpp:L386 e.g.
Avatar
Deleted User 2021-08-23 18:59:15Z
i insert unsorted FROM a sorted array
Avatar
Learath2 2021-08-23 18:59:52Z
So I can replace it with a simple vector
Avatar
Deleted User 2021-08-23 19:00:18Z
yes if they dont have any call to add then yes
19:01
and you can also remove static while u at it xdd
19:01
ok no
19:01
the temlates to lambdas is too much work
Avatar
Learath2 2021-08-23 19:01:23Z
I think you are by far the biggest used of sorted_array in the entire source 😄
Avatar
Deleted User 2021-08-23 19:01:50Z
its not unuseful xd
19:01
if std would have flat_set
19:02
i wouldnt use it xD
Avatar
Learath2 2021-08-23 19:31:08Z
I have a diff so large it hurts my head
Avatar
Learath2 2021-08-23 19:39:40Z
Storing all the indexes as int was a rather large mistake on our part :/
Avatar
Deleted User 2021-08-23 19:41:44Z
instead of size_t or instead of what
Avatar
Learath2 2021-08-23 19:42:14Z
instead of size_t
19:43
I hope everything still works
19:44
It does, wtf
19:44
oh nvm, no such luck failed to open file. filename='assets/entities/comfort/blockworlds.png'
Avatar
Deleted User 2021-08-23 19:45:07Z
its ok
19:45
that is not a bug
19:45
that just means that comfort asset pack still misses other mods
Avatar
Avatar
trml
looking at the server code, it doesn't look like m_Layer in entity.h is used for much except to check if an entity is in switch. would be ok to remove/refactor out m_Layer, and just use m_Number == 0 when it's not in switch?
fokkonaut 2021-08-23 20:07:13Z
No, its also for Layer_game and layer_front
Avatar
Avatar
trml
but if we are certain that m_Layer will never be used for anything else in the future (and can always be inferred from m_Number) it's perhaps not needed
fokkonaut 2021-08-23 20:08:04Z
we dont need anything to determine this, number 0 is always on and also used for game and front layer
20:08
Thats why its num_switchers + 1, because 0 is reserved as always on, for game, front and also switch layer
20:09
you only need to send which switches are active for your current team, no need to know about others' switch states, as other teams shouldnt be predicted anyways (using /showothers)
20:09
and when you are spectating or in pause, no tees are predicted anyways
20:10
So the client also does not need to know any numswitchers, as it always receives every switch state (0-255) anyways
Avatar
gerdoe 2021-08-23 20:10:28Z
huh
Avatar
fokkonaut 2021-08-23 20:10:53Z
and the layer doesnt need to be sent because it doesnt matter, gane and front layer switches are determined by switch number 0, which is always on when sent
20:11
(if m_Status[0] is not set to true, it has to be done, thats a server issue then because i think loops start at 1)
20:12
I worked a lot with this code in my mod, i think i know it good
Avatar
gerdoe 2021-08-23 20:12:37Z
switches should be refactored... after teams core refactor
Avatar
fokkonaut 2021-08-23 20:13:02Z
Yes, they should, but current system also works fine.
20:13
And makes sense to me, after understanding ut
20:13
it*
Avatar
gerdoe 2021-08-23 20:13:27Z
its literally spaghetti code
20:14
yeah its understandable but bruh you should check every line that works with switches
Avatar
trml 2021-08-23 20:15:16Z
about the first, entities are currently only added to layer_switch or layer_game (layer_front would currently have no special effect for entities anyway), but this is basically the same as what you say in the next message
20:16
I commented about this on the pr too btw
20:16
(entities as in actual objects in the gameworld, not the tiles)
Avatar
gerdoe 2021-08-23 20:18:04Z
stupid Q: can we get std::bad_alloc while adding all teleouts?
Avatar
fokkonaut 2021-08-23 20:19:31Z
I will make a pr soon, fixing the switchstate server message and i will also look at the client
20:20
Dont release a new version with the just added switchstates, new messages should always be checked carefully and well thought, according to heinrich
20:20
We dont need more of compability code
20:20
Especially in such a case
Avatar
gerdoe 2021-08-23 20:21:20Z
Q: how many bytes does server send about one player with character?
20:21
to another player
Avatar
Deleted User 2021-08-23 20:26:38Z
just use vanilla as fallback
Avatar
Avatar
fokkonaut
you only need to send which switches are active for your current team, no need to know about others' switch states, as other teams shouldnt be predicted anyways (using /showothers)
trml 2021-08-23 20:26:39Z
that may reduce the network traffic in some cases (when there are many switches). on maps with few switches it might increase it though, especially if switching to dummy or speccing often. (and you'd also want to send this while speccing to allow adding more features on top of it)
Avatar
Deleted User 2021-08-23 20:26:49Z
outdated ddnet = vanilla
20:26
simple as that xd
Avatar
trml 2021-08-23 20:30:56Z
besides to only send it for the characters own team you dont need to change the pr, just add a couple of lines in the server to skip the message unless character_id == snapping_id
Avatar
fokkonaut 2021-08-23 20:31:18Z
fits better to gamecontroller::snap
Avatar
Saiborg 2021-08-23 20:31:58Z
Hey guys. Ive just started uni and if i translate it to english it would be Computer Engineer. In reality its more of a mix between CE and CS. Since i love this game, i thought id ask if there could be any possibility of me maybe helping out a bit in the future when ive studied a bit more? dont need to be anything huge, but if the opportunity arises i would gladly help a bit. Would give me some RL experience as well 🙂
Avatar
fokkonaut 2021-08-23 20:32:07Z
Yeah okay, sending numswitchers might actually be cool, but i dont think sending all teams is neccessary
Avatar
Avatar
Saiborg
Hey guys. Ive just started uni and if i translate it to english it would be Computer Engineer. In reality its more of a mix between CE and CS. Since i love this game, i thought id ask if there could be any possibility of me maybe helping out a bit in the future when ive studied a bit more? dont need to be anything huge, but if the opportunity arises i would gladly help a bit. Would give me some RL experience as well 🙂
gerdoe 2021-08-23 20:32:35Z
just read the code + issues
Avatar
trml 2021-08-23 20:33:16Z
there is a lot of code:P
Avatar
Avatar
Saiborg
Hey guys. Ive just started uni and if i translate it to english it would be Computer Engineer. In reality its more of a mix between CE and CS. Since i love this game, i thought id ask if there could be any possibility of me maybe helping out a bit in the future when ive studied a bit more? dont need to be anything huge, but if the opportunity arises i would gladly help a bit. Would give me some RL experience as well 🙂
lynn 2021-08-23 20:33:24Z
accounts
20:33
jk but someone wanted animated skins a while ago
Avatar
Avatar
trml
there is a lot of code:P
gerdoe 2021-08-23 20:33:54Z
skip everything unless src/game ahah
Avatar
Saiborg 2021-08-23 20:34:02Z
Ah, yeah that would be a good idea xD Just wanted to know if you guys would be okay with someone helping out a bit.
Avatar
gerdoe 2021-08-23 20:34:06Z
especially src/game/server
Avatar
Avatar
Saiborg
Ah, yeah that would be a good idea xD Just wanted to know if you guys would be okay with someone helping out a bit.
lynn 2021-08-23 20:34:31Z
maybe ask Learath or TsFreddie if there's any small stuff they want to fix but don't have time for
Avatar
gerdoe 2021-08-23 20:35:00Z
as heinrich said to you
20:35
"good first issue"
20:35
i forgot it zd
Avatar
Avatar
lynn
maybe ask Learath or TsFreddie if there's any small stuff they want to fix but don't have time for
Saiborg 2021-08-23 20:35:28Z
Thanks, i will do that as soon as i feel i have a bit more knowledge... Still too newb at the moment to be of any help 😛
Avatar
gerdoe 2021-08-23 20:36:24Z
just try to do some shit with ddnet code to introduce another features you want to see in a game
20:36
good train, rly xd
Avatar
Saiborg 2021-08-23 20:37:13Z
I will give it a go 😄
Avatar
Deleted User 2021-08-23 20:37:59Z
true, starting is the hardest 😄
Avatar
Avatar
gerdoe
"good first issue"
lynn 2021-08-23 20:39:30Z
ptsd
Avatar
Deleted User 2021-08-23 20:39:33Z
i think #4104 also fixed picking too huge resolutions i looked in the SDL code and it always checks the resolution requested first
Avatar
DDNet BOT 2021-08-23 20:39:34Z
https://github.com/ddnet/ddnet/pull/4104
Fix fullscreen resolution setting on SDL 2.0.16 by vainiovano · Pul...
SDLSetWindowSize&#39;s documentation tells that SDL_SetWindowDisplayMode should be used to set the sizes of fullscreen windows. SDL commit 72d812852023450b7c3850b8e87b63575df041ee changed SDL...
Thumbnail
Avatar
lynn 2021-08-23 20:39:34Z
i dont know cpp dont talk to me
Avatar
Avatar
lynn
i dont know cpp dont talk to me
Deleted User 2021-08-23 20:39:56Z
bcs u don't "just" start 😄
Avatar
gerdoe 2021-08-23 20:40:05Z
xd
20:40
c/c++ da best
20:40
you can do anything you want
Avatar
lynn 2021-08-23 20:40:21Z
i will in 5 years
20:40
irl has been too fucked up
Avatar
Deleted User 2021-08-23 20:40:42Z
yeah you can even hack yourself ez
Avatar
gerdoe 2021-08-23 20:40:51Z
asshack ez
20:41
or hackass
20:41
nvm
Avatar
Ryozuki 2021-08-23 20:47:19Z
@Saiborg dont wait for uni to learn programming
20:47
ofc any contribution is welcome
20:47
ah i misread u already started
Avatar
Deleted User 2021-08-23 20:49:19Z
would be funny if he already had an pr up
Avatar
Avatar
Ryozuki
@Saiborg dont wait for uni to learn programming
Saiborg 2021-08-23 20:55:18Z
Yeah, started today 🙂 I have some basic knowledge in Java from before, but nothing fancy
Avatar
Deleted User 2021-08-23 21:02:05Z
ddnet and being able to read discord
Image attachment
21:02
single monitor real life hacks
Avatar
gerdoe 2021-08-23 21:02:24Z
kekw
Avatar
murpi 2021-08-23 21:03:32Z
ontopreplica?
Avatar
Deleted User 2021-08-23 21:04:09Z
no
21:04
linux just allows it xD
21:04
to set the window half transparent
21:04
SDL_SetWindowOpacity
Avatar
GitHub BOT 2021-08-23 21:20:17Z
Author icon
fokkonaut
[ddnet/ddnet] Pull request opened: #4114 Send switch states per team and cleanup
"Combines" https://github.com/ddnet/ddnet/pull/4084/files and https://github.com/ddnet/ddnet/pull/4094/files aswell as making sure no unallocated memory gets accessed, thus making it safe to change NumSwitchers mid-game. It's better to send switch state per team than per character in my opinion, also this directly writes to CCollision instead of saving the switch state per character in m_SwitchStates[].

Checklist

  • [ ] Tested the change ingame
  • [ ] Provided screenshots if it is a ...
Avatar
fokkonaut 2021-08-23 21:20:25Z
@trml @Ryozuki
Avatar
fokkonaut 2021-08-23 21:33:00Z
also, i dont fully understand what https://github.com/ddnet/ddnet/pull/4108 is for
Send item switch info by trml · Pull Request #4108 · ddnet/ddnet
For prediction of bullets and moving pickups in switch layer (#3990). This sends the info in a separate netobject (as proposed in #4102), as an alternative to adding new netobjects/fields for each ...
Thumbnail
Avatar
Learath2 2021-08-23 21:35:46Z
it's to associate entities with switches
Avatar
fokkonaut 2021-08-23 21:36:22Z
And how does the client know where its connected to?
Avatar
Learath2 2021-08-23 21:37:16Z
The id field within the netobj, the same way it's done with character netobjs
Avatar
trml 2021-08-23 21:37:42Z
I have a pr ready for pickups and projectiles based on this, but it requires this netobj or something similar (also one for laser walls, but will test it a little first)
Avatar
fokkonaut 2021-08-23 21:37:43Z
ah, yea
Avatar
Avatar
Learath2
The id field within the netobj, the same way it's done with character netobjs
fokkonaut 2021-08-23 21:38:59Z
ty
Avatar
Avatar
Deleted User
https://github.com/nigels-com/glew/pull/216
nori 2021-08-23 21:50:53Z
hmm, 2 line diff waiting for 2 years ? :d
Avatar
trml 2021-08-23 21:52:15Z
@fokkonaut sending per team seems ok to me. btw I'd replace the block with the "delete"/"new" stuff with just "continue"
Avatar
fokkonaut 2021-08-23 21:53:11Z
no, thats inacceptable :D
Avatar
lynn 2021-08-23 21:53:14Z
Image attachment
21:53
xd
Avatar
Avatar
trml
@fokkonaut sending per team seems ok to me. btw I'd replace the block with the "delete"/"new" stuff with just "continue"
fokkonaut 2021-08-23 21:53:26Z
I think it makes sense to do it like this
21:53
Really
21:53
otherwise 1. mod makers wonder why their stuff isnt working
21:54
and 2. it is just a good way, i think, because otherwise memory that is not allocated could be accessed
21:54
see how SSwitchers is built in collision.cpp
21:54
if it would be a 256 array, it would not be an issue anyways
21:54
But like this, i think this is a good way
Avatar
Avatar
nori
hmm, 2 line diff waiting for 2 years ? :d
Deleted User 2021-08-23 21:56:23Z
xD
Avatar
fokkonaut 2021-08-23 21:56:32Z
i think #4108 is actually okay like it is
Avatar
DDNet BOT 2021-08-23 21:56:32Z
https://github.com/ddnet/ddnet/pull/4108
Send item switch info by trml · Pull Request #4108 · ddnet/ddnet
For prediction of bullets and moving pickups in switch layer (#3990). This sends the info in a separate netobject (as proposed in #4102), as an alternative to adding new netobjects/fields for each ...
Thumbnail
Avatar
Avatar
trml
@fokkonaut sending per team seems ok to me. btw I'd replace the block with the "delete"/"new" stuff with just "continue"
Learath2 2021-08-23 22:02:09Z
I thought about this, I think I like it better per player, it'd be more flexible that way
Avatar
fokkonaut 2021-08-23 22:02:33Z
Thats the same, mod makers can change it depending on the snappingclient
Avatar
Learath2 2021-08-23 22:03:25Z
It's only the same if you send a single switch-list, which is fine with me too, if @trml thinks there is no value in the switch state of other tees
Avatar
fokkonaut 2021-08-23 22:04:00Z
It is, if something is needed when there is no character
Avatar
trml 2021-08-23 22:04:04Z
for other tees I think there is
Avatar
Learath2 2021-08-23 22:04:22Z
yeah, I thought so too, we can predict other tees using it
Avatar
trml 2021-08-23 22:04:51Z
for example, when switching back and forth between dummy or in/out of spec, then you wouldn't have a delay
Avatar
fokkonaut 2021-08-23 22:05:09Z
i dont think we need prediction of players in another team, anyways its not a big difference when sending it per player or per team, just per team makes more sense as the client has all information it needs then
22:05
(all the time)
Avatar
trml 2021-08-23 22:05:43Z
in a way I also agree about doing it per character, if that could open up modding possibilities (if looking at only current ddnet source team is technically enough)
Avatar
Learath2 2021-08-23 22:06:28Z
@trml you can modify it per snappingclient as fokkonaut said, so it's not really any less flexible per team
Avatar
fokkonaut 2021-08-23 22:07:02Z
(just better, there could also be a client that displays all team states)
22:07
then it is more flexible when its per team
Avatar
Learath2 2021-08-23 22:07:20Z
I honestly would prefer per character as the less ddnet specific the client is the better imho
Avatar
fokkonaut 2021-08-23 22:07:28Z
huh?
Avatar
Learath2 2021-08-23 22:07:39Z
Teams are a concept of ddnet after all
Avatar
fokkonaut 2021-08-23 22:07:50Z
I dont get it, it is as you would send gameinfoex per character
22:08
Its not even possible to set players into another team without them being transparent
22:08
(when speaking of modding etc)
Avatar
Learath2 2021-08-23 22:08:46Z
Okay, let's split the discussion, I'm talking about sending all switch states, not just one
Avatar
Avatar
fokkonaut
Its not even possible to set players into another team without them being transparent
fokkonaut 2021-08-23 22:08:49Z
for color purpose
Avatar
trml 2021-08-23 22:09:04Z
I'm not sure I follow the part about the snappingclient
Avatar
fokkonaut 2021-08-23 22:09:20Z
Server can send different switch states to different clients
Avatar
trml 2021-08-23 22:09:44Z
was that still about per-team vs per-character?
Avatar
fokkonaut 2021-08-23 22:09:58Z
Yea, about more flexibility, which in this case is the same
Avatar
trml 2021-08-23 22:10:22Z
I see those two as the only two options, but don't see how you can do much with the snappingclient there
Avatar
fokkonaut 2021-08-23 22:10:47Z
on serverside
22:11
depending on some player specific values you can choose to send number 1 as closed, for exmaple, while its opened for another client
Avatar
Learath2 2021-08-23 22:11:35Z
@trml when sending per-team you can still fake as if you are sending per-character
Avatar
fokkonaut 2021-08-23 22:11:45Z
Yea
Avatar
Learath2 2021-08-23 22:12:12Z
When snapping switch state for team n for client x you can pretend as if team n has a different switch state
Avatar
fokkonaut 2021-08-23 22:12:19Z
(and i dont see how it is not better to send it per team, because when sending per character the client doesnt always have correct information
22:12
When there is no character around for example
Avatar
trml 2021-08-23 22:12:42Z
oh, I see
22:12
so the "subjective" switch state can be different for each character in a team
Avatar
fokkonaut 2021-08-23 22:13:18Z
This is a general information the Client should have, regardless of anything else, also its not character specific so it doesnt make sense at all to me
Avatar
Avatar
fokkonaut
When there is no character around for example
Learath2 2021-08-23 22:13:25Z
This is an interesting edge case
Avatar
fokkonaut 2021-08-23 22:13:38Z
Yea, and stuff like this can be easily avoided :D
Avatar
Learath2 2021-08-23 22:14:03Z
"easily" just need to introduce the concept of teams to the protocol, no biggie 😛
Avatar
fokkonaut 2021-08-23 22:14:08Z
xd
Avatar
Learath2 2021-08-23 22:14:12Z
We don't snap anything for teams yet
Avatar
fokkonaut 2021-08-23 22:15:00Z
I wanted to put this into teams::snap, knowing there is none xD then i thought it would fit (ddrace)controller
22:15
so i put it to the gameinfoex, which is also something that has to be available all the time (and/or should be)
Avatar
Learath2 2021-08-23 22:15:59Z
I guess we should create a CTeams::Snap and snap it there
Avatar
fokkonaut 2021-08-23 22:16:23Z
From where we send it is not from matter for now, but yes
Avatar
Learath2 2021-08-23 22:16:32Z
And the client will need a CTeams of it's own with an array of switch states
Avatar
fokkonaut 2021-08-23 22:16:40Z
Why?
22:16
they are in collision
22:17
Or do you mean for a full refactor of teams including different gameworlds? xd
Avatar
Learath2 2021-08-23 22:17:03Z
collision knows nothing of teams
Avatar
Avatar
Learath2
collision knows nothing of teams
fokkonaut 2021-08-23 22:17:11Z
Doesnt need to
Avatar
trml 2021-08-23 22:17:20Z
I tried putting a Snap in CTeams when first looking at this and the code for some reason ended up looking very ugly
Avatar
fokkonaut 2021-08-23 22:17:23Z
but the switchers are saved there
Avatar
Cipy29 2021-08-23 22:17:36Z
hey guys I just moved from windows to debian and did the basic $ apt-get install ddnet, but that means im on older version
Avatar
Learath2 2021-08-23 22:17:38Z
You are still thinking of sending a single switch state object
Avatar
Cipy29 2021-08-23 22:17:47Z
is it possible to do something to be on newest version? (edited)
Avatar
fokkonaut 2021-08-23 22:17:56Z
CTeams::Snap(int) for (int i = 0; i < 64; i++) SnapTeam(i, int);
Avatar
Avatar
Learath2
You are still thinking of sending a single switch state object
fokkonaut 2021-08-23 22:18:03Z
Me?
Avatar
Learath2 2021-08-23 22:18:38Z
Yes, CCollision has space for one state, we are sending one per team
Avatar
fokkonaut 2021-08-23 22:18:58Z
no? it has space for 64 teams
22:19
states*
Avatar
trml 2021-08-23 22:19:23Z
how about when a character is in solo? that would perhaps be a case where you have to send different states to different team members (when doing per-team instead of per-character)
Avatar
Learath2 2021-08-23 22:19:29Z
Oh, that's so ugly, I missed that 😛
Avatar
Avatar
Learath2
Oh, that's so ugly, I missed that 😛
fokkonaut 2021-08-23 22:19:40Z
xd
22:19
Thats how its saved on the serverside
Avatar
Learath2 2021-08-23 22:20:02Z
@trml you'd think so, solo people in teams share switches, it leads to some very INSANE behaviour
Avatar
trml 2021-08-23 22:20:11Z
haha
Avatar
fokkonaut 2021-08-23 22:20:13Z
xddd
22:20
i still dont think people of other teams or solo players should be predicted at all
22:20
because you dont interact with them
Avatar
trml 2021-08-23 22:20:41Z
yeah, when thinking about it I never saw any place where it would separate the players
Avatar
Avatar
fokkonaut
because you dont interact with them
fokkonaut 2021-08-23 22:20:57Z
then, while watching, it would also be smooth, if you just have /showothers on
22:21
same as we dont predict while spectating
Avatar
trml 2021-08-23 22:21:40Z
we dont predict other teams now
Avatar
fokkonaut 2021-08-23 22:21:56Z
no? thats good, should also be done for solo players
22:22
(if not super)
Avatar
Avatar
trml
we dont predict other teams now
Learath2 2021-08-23 22:22:10Z
Only if unseen though, no?
Avatar
trml 2021-08-23 22:22:14Z
they are also considered as "other team"
Avatar
Avatar
Learath2
Only if unseen though, no?
fokkonaut 2021-08-23 22:22:24Z
I think so too, not sure
Avatar
trml 2021-08-23 22:22:26Z
I removed the physics prediction as well
Avatar
Learath2 2021-08-23 22:22:33Z
If showothers is on and predict others are on it should still be predicted?
Avatar
trml 2021-08-23 22:23:51Z
@Learath2 I thought it would be worth trying it out to see how it worked
Avatar
Learath2 2021-08-23 22:24:22Z
@trml and how do you have the team information?
Avatar
trml 2021-08-23 22:25:33Z
we have it somewhere I think, I just remember using IsOtherTeam
Avatar
fokkonaut 2021-08-23 22:25:49Z
svteamsstate
Avatar
Learath2 2021-08-23 22:26:08Z
Anyway, if you don't think you'd use the switchstate of other teams for anything we can also send just a single one
Avatar
trml 2021-08-23 22:26:19Z
but we can use it
22:26
since it means less delay when switching to/from spec/dummy
Avatar
Learath2 2021-08-23 22:26:53Z
How much less though? We snap fairly often
Avatar
trml 2021-08-23 22:27:09Z
yes, would just be limited to ping
Avatar
Learath2 2021-08-23 22:27:48Z
Hm, wouldn't the dummy also get its own switch state object actually?
Avatar
fokkonaut 2021-08-23 22:27:58Z
yes
Avatar
trml 2021-08-23 22:28:12Z
actually yes, but the other snapshot is never used
22:28
or
22:28
hmm, yeah
Avatar
Learath2 2021-08-23 22:28:44Z
It should be used at the very least to get the dummy client id 😛
22:29
anyway, I think it'd only help when switching to spec if we don't predict other teams
Avatar
trml 2021-08-23 22:29:03Z
so the delay is not actually the whole ping, just the time until the next snapshot is read
Avatar
Learath2 2021-08-23 22:29:27Z
in that case we can just send one I guess
22:30
I just had an idea for a very ugly optimization, if tees tell the server whose dummy they are, they can receive even lighter deltas 😛
Avatar
trml 2021-08-23 22:30:25Z
another possibility would perhaps be to only update the snapped state for the current player. but that would require some hacky code on the server
Avatar
Avatar
trml
another possibility would perhaps be to only update the snapped state for the current player. but that would require some hacky code on the server
Learath2 2021-08-23 22:31:48Z
I don't follow this
Avatar
trml 2021-08-23 22:33:01Z
the state would be sent for each team/character, but the server would not update the state that is sent for all the other characters other than the snapping client, and would just send the last cached state
Avatar
fokkonaut 2021-08-23 22:33:05Z
I think its okay to send it only for your own team. @Learath2, should m_Status from the ccollision array for other teams be reset, or can we just leave old information in there as they get updated when it needs?
Avatar
Avatar
trml
the state would be sent for each team/character, but the server would not update the state that is sent for all the other characters other than the snapping client, and would just send the last cached state
fokkonaut 2021-08-23 22:33:41Z
Hm? every character has their own snap
22:34
SnappingClient iterates through every client
Avatar
trml 2021-08-23 22:34:19Z
you mean each client has, not each character
Avatar
fokkonaut 2021-08-23 22:34:41Z
yes, ofc
Avatar
Avatar
trml
the state would be sent for each team/character, but the server would not update the state that is sent for all the other characters other than the snapping client, and would just send the last cached state
Learath2 2021-08-23 22:34:46Z
This would be a very minor optimization though, since deltas are already extremely light
22:35
An update of up to 8 switches will only cause 1 byte to be transmitted
Avatar
Avatar
fokkonaut
I think its okay to send it only for your own team. @Learath2, should m_Status from the ccollision array for other teams be reset, or can we just leave old information in there as they get updated when it needs?
fokkonaut 2021-08-23 22:35:54Z
clientside
Avatar
trml 2021-08-23 22:36:38Z
yes, you could cache it clientside as well
Avatar
Avatar
fokkonaut
I think its okay to send it only for your own team. @Learath2, should m_Status from the ccollision array for other teams be reset, or can we just leave old information in there as they get updated when it needs?
Learath2 2021-08-23 22:36:49Z
Hm, I guess we can keep garbage in there, the implementer just needs to keep in mind only the current team is valid
Avatar
fokkonaut 2021-08-23 22:36:50Z
I mean the info gets outdated fast
Avatar
Avatar
Learath2
Hm, I guess we can keep garbage in there, the implementer just needs to keep in mind only the current team is valid
fokkonaut 2021-08-23 22:36:58Z
good
Avatar
Learath2 2021-08-23 22:42:01Z
Sorry if I'm being annoying with the questions btw. Until someone gets in some compatibility code we only get one shot at netobjs, so I want to get it right
Avatar
fokkonaut 2021-08-23 22:43:41Z
Same as i think
22:44
updated my pr
22:44
#4114
Avatar
DDNet BOT 2021-08-23 22:44:12Z
https://github.com/ddnet/ddnet/pull/4114
Send switch states only for own team by fokkonaut · Pull Request #4...
&quot;Combines&quot; https://github.com/ddnet/ddnet/pull/4084/files and https://github.com/ddnet/ddnet/pull/4094/files aswell as making sure no unallocated memory gets accessed, thus making...
Thumbnail
Avatar
fokkonaut 2021-08-23 22:44:58Z
This even only changes the server code
22:45
thats quite good i think
22:47
If our own character doesnt exist we dont predict anything, so this seems to be okay
Avatar
Learath2 2021-08-23 22:47:45Z
mind if I review this tomorrow? I'm exhausted
Avatar
fokkonaut 2021-08-23 22:47:55Z
Sure, not much to review tho
22:48
really small pr
22:48
just geting netobj correct, i use item id as team
Avatar
trml 2021-08-23 22:48:49Z
seems correct to use team
Exported 959 message(s)