DDraceNetwork - developer

DDraceNetwork

Development / developer

Development discussion. Logged to https://ddnet.tw/irclogs/ Connected with DDNet's IRC channel, Matrix room and GitHub repositories — IRC: #ddnet on Quakenet | Matrix: #ddnet-developer:matrix.org GitHub: https://github.com/ddnet

Between 2018-03-19 00:00:00Z and 2018-03-20 00:00:00Z

[quakenet] <deen> Learath2: The idea is that by restrictring your own power you have fewer things to think about and thus make fewer bugs. If you see a value is immutable you don't have to reason about how it will change during the function execution

05:04

[quakenet] <deen> And of course, like heinrich5911 said, the compiler can make use of it for optimizations

05:06

[quakenet] <deen> 5991*

05:09

[quakenet] <deen> See also: "Using control structures instead of goto", "Using scope instead of globals", "Static instead of dynamic typing", DSLs, etc

05:10

[quakenet] <deen> https://en.wikipedia.org/wiki/Rule_of_least_power

Rule of least power

In programming, the rule of least power is a design principle that "suggests choosing the least powerful [computer] language suitable for a given purpose". Stated alternatively, given a choice among computer languages, classes of which range from...

Interesting

[freenode] <nemo> immutability is used for reliability/performance too

12:36

[freenode] <nemo> like Java (and others) making strings immutable

[quakenet] <Learath2> or you can just pay attention? The optimization gains are nice which is why it'd be a nice language feature like in javascript

13:57

[quakenet] <Learath2> but having it by default or forcing it on all variables is just overkill, been using redux for the last month and everyone keeps shouting at you to not mutate the state but mutating the state is much quicker then making a new copy of the entire app state and modifying that

14:02

[quakenet] <deen> Well, if you have to modify the entire state all the time, then it sounds like you still want to do imperative programming. I guess redux is trying to enforce functional style. You have some example?

14:05

[quakenet] <Learath2> I want the language to let me do whatever I want without fancy keywords. If I want safety features or to mark something immutable I should be the one to decide that

14:05

[quakenet] <deen> So you also want to type check by default?

14:06

[quakenet] <Learath2> I guess you mean "(disable) type check..."

14:07

[quakenet] <deen> yes

14:07

[quakenet] <deen> And in my experience people keep forgetting to mark values as immutable

14:07

[quakenet] <Learath2> As long as it doesn't hurt the language's structure I don't mind, if you are going to make it a mess like js you better keep the type checking

14:10

[quakenet] <Learath2> I feel a strongly typed language helps very much in reducing bugs, but ask many people who use js, py, ruby and they might feel it slows them down

14:13

[quakenet] <deen> Yeah, and you sound similar about immutability :D

14:14

[quakenet] <Learath2> I kinda enjoyed the fact that javascript had all the tools needed to write in a functional way but also let's you write in an imperative way

14:14

[quakenet] <Learath2> What's wrong with freedom? Opionated languages and frameworks make us boring and predictable.

14:15

[quakenet] <deen> What's wrong with freedom? Building codes and fire exists make us boring and predictable.

14:16

[quakenet] <deen> So the idea is that you want your code to be boring and predictable in order to work together in big teams and easily read and understand each other's code

14:16

[quakenet] <deen> Preferring a simple solution over an interesting but harder to understand one

14:17

[quakenet] <Learath2> off-topic: deen: where did we mount the test server maps? servers/test/maps or servers/test/data/maps?

14:17

[quakenet] <deen> Go language tries very hard to do that

14:17

[quakenet] <deen> dunno :D

14:17

[quakenet] <deen> run-all.sh says test/data/maps

14:18

[quakenet] <deen> Nim is quite the opposite, you can write fully imperative code, do unsafe stuff everywhere, have macros that change the language in some places to something totally different. So that's a whole lot of fun, but I wouldn't recommend it to a large team for mission critical software

14:20

[quakenet] <Learath2> mission critical software has been written for decades, standards and coding simply not fancily already solve these issues

14:20

[freenode] <nemo> heh. there was a dude on hedgewars server the other day who was scoffing at rust, 'cause it wasn't coq

14:20

[freenode] <nemo> "How can you trust any code that hasn't been formally verified"

14:21

[freenode] <nemo> at some point, gotta actually just get stuff done

14:21

[freenode] <nemo> (this was because we were discussing the rust port in that room)

14:21

[quakenet] <Learath2> functional programming surely makes it easier, which is why it's a nice paradigm, shoving it down peoples throats is just, meh

14:21

[quakenet] <deen> yeah, I tried running a SAT solver on some of our code at work, ran out of memory after 48 hours :D

14:21

[quakenet] <deen> not much success with proving for me

14:22

[quakenet] <Learath2> what is a coq even?

14:22

[quakenet] <deen> https://en.wikipedia.org/wiki/Coq

In computer science, Coq is an interactive theorem prover. It allows the expression of mathematical assertions, mechanically checks proofs of these assertions, helps to find formal proofs, and extracts a certified program from the constructive pr...

14:23

[freenode] <nemo> did you guys see how Sir Robin was discovered using a SAT? (the conway knightship)

14:23

[quakenet] <deen> yeah

14:24

[freenode] <nemo> also... apparently some dude out there is busy formally verifying fundamental rust libs, so at least there's that, if it makes dude-on-our-gameserver-feel better.

14:24

[freenode] <nemo> 'course he misunderstood me at time and thought I was suggesting this was a normal thing to do in rust. but then, he also thought rust only had memory bounds checks, nothing more

14:25

[quakenet] <deen> Unit testing would already be a step in the right direction :D

14:25

[quakenet] <deen> At least prove that your code works for one case, not for all

14:25

[quakenet] <Learath2> currently learning rust, it is pretty different

14:26

[quakenet] <Learath2> as in I've never quite seen anything like it

14:28

[quakenet] <deen> anything in specific?

14:30

[quakenet] <Learath2> the syntax is really new, i keep finding myself thinking in if/else if/else where I should probably be thinking in terms of match blocks

[quakenet] <heinrich5991> deen: coq is different, it's not an automatic prover but a proof assistent – it won't run out of memory, it'll tell you to specify the proof ^^

[quakenet] <deen> heinrich5991: I know, I've used it a bit. Wanted to try out something new though

[quakenet] <heinrich5991> o_be_one / o_tee_one: I understand your name now. it hit me in the shower :D

pituh?

[quakenet] <jxsl13> xD "It hit me in the shower"

21:33

[quakenet] <jxsl13> anyone online :D?

[quakenet] <laxadedi> bam support deprecated for ddnet ?

22:06

[quakenet] <laxadedi> Is there still different build for server / server-sql / clients ? If yes, I think it should be added to the README.md

[quakenet] <heinrich5991> laxadedi: it's no longer deprecated, it's removed

23:11

[quakenet] <heinrich5991> there's something in the readme about mysql support:

23:11

[quakenet] <heinrich5991> laxadedi: MySQL (or MariaDB) support in the server is not included in the binary releases but it can be built by specifying -DMYSQL=ON, like cmake -DMYSQL=ON ... It requires libmariadbclient-dev, libmysqlcppconn-dev and libboost-dev, which are also bundled for the common platforms.

23:12

[quakenet] <laxadedi> I hate cmake, cannot find my local libs...

23:13

[quakenet] <heinrich5991> what do you mean by "local"? self-compiled? in the package manager? in the repository?

23:13

[quakenet] <laxadedi> package manager

23:14

[quakenet] <heinrich5991> mh. should be able to find these

23:14

[quakenet] <heinrich5991> mind sharing the output of the initial cmake run?

23:14

[quakenet] <laxadedi> -- * MySQL not found (using bundled version)

23:14

[quakenet] <laxadedi> # dpkg-query -L libmysqlcppconn-dev

23:14

[quakenet] <laxadedi> /usr/lib/libmysqlcppconn-static.a

23:15

[quakenet] <heinrich5991> do you have the tool mysql_config? which mysql_config

23:15

[quakenet] <laxadedi> https://pastebin.com/f3QptLiP

$ cmake . -- Could NOT find GLEW (missing: GLEWINCLUDE_DIR GLEW...

23:15

[quakenet] <laxadedi> I do

23:15

[quakenet] <laxadedi> /usr/bin/mysql_config

23:16

[quakenet] <laxadedi> It's up-to-date debian 8.10

23:16

[quakenet] <heinrich5991> and also libmariadbclient-dev (or the equivalent for mysql, if that exists)? I guess you previously compiled with bam and it worked?

23:16

[quakenet] <laxadedi> cmake 3.10.3

23:16

[quakenet] <laxadedi> yes, I had no issue with bam, libmariadblcient-dev does not exist on debian

23:17

[quakenet] <heinrich5991> libmysqlclient-dev then I guess?

23:17

[quakenet] <laxadedi> already installed :)

23:18

[quakenet] <heinrich5991> good

23:18

[quakenet] <laxadedi> /usr/lib/x86_64-linux-gnu/libmysqlclient.a

23:19

[quakenet] <heinrich5991> can you pastebin the CMakeCache.txt?

23:20

[quakenet] <laxadedi> pmed the link

23:21

[quakenet] <heinrich5991> output of mysql_config --libs_r

23:21

[quakenet] <laxadedi> -L/usr/lib/x86_64-linux-gnu -lmysqlclient_r -lpthread -lz -lm -ldl

23:21

[quakenet] <heinrich5991> hmmm

23:24

[quakenet] <laxadedi> Is there a way to say to cmake where to get the library ?

23:24

[quakenet] <heinrich5991> yes

23:25

[quakenet] <laxadedi> I guess it's not a proper fix, but I'd take it for now

23:25

[quakenet] <heinrich5991> -DMYSQL_LIBRARY=/usr/lib/x86_64-linux-gnu/libmysqlclient.a

23:25

[quakenet] <laxadedi> ok building, going to take 2 minutes at least, crappy server :D

23:25

[quakenet] <heinrich5991> but you also need MYSQL_INCLUDEDIR, MYSQL_CPPCONN_LIBRARY and MYSQL_CPPCONN_INCLUDEDIR then

23:26

[quakenet] <heinrich5991> you could apply this patch and tell me what it outputs

23:26

[quakenet] <laxadedi> according to CMAkeCache, the include dirs are good

23:26

[quakenet] <heinrich5991> https://gist.github.com/heinrich5991/005382b95c6e3e65a91022bb2ec74d9a

23:26

[quakenet] <heinrich5991> ok

23:26

[quakenet] <laxadedi> only the libs are wrong, which could explain the "undefined ref" when compiling

23:26

[quakenet] <heinrich5991> yes

23:30

[quakenet] <laxadedi> So slow :(, 65%

23:31

[quakenet] <laxadedi> And I am stupid, compiling everything instead of just the DDNet-Server

23:31

[quakenet] <heinrich5991> make DDNet-Server

23:31

[quakenet] <heinrich5991> you can safely ctrl-c

23:31

[quakenet] <heinrich5991> progress is kept

23:31

[quakenet] <laxadedi> Now building the server x)

[ddnet/ddnet] Pull request closed: #1050 Pointless to hide the ip. Fixes #907

[quakenet] <laxadedi> beh, failed

23:34

[quakenet] <laxadedi> tons of undefined reference

23:35

[quakenet] <heinrich5991> grep 'MYSQL.*_LIBRARY' CMakeCache.txt

23:36

[quakenet] <laxadedi> MYSQL_CPPCONN_LIBRARY:FILEPATH=/usr/lib/libmysqlcppconn-static.a

23:36

[quakenet] <laxadedi> MYSQL_LIBRARY:FILEPATH=/usr/lib/x86_64-linux-gnu/libmysqlclient.a

23:37

[quakenet] <heinrich5991> hm

23:37

[quakenet] <heinrich5991> what kind of undefined references do you see?

[ddnet/ddnet] Pull request closed: #1001 added '/mods' command which shows a list of moderators

[quakenet] <laxadedi> /usr/lib/gcc/x86_64-linux-gnu/4.9/../../../../lib/libmysqlcppconn-static.a(mysql_client_api.cpp.o): In function ql::mysql::NativeAPI::LibmysqlStaticProxy::get_server_version(st_mysql*)':

23:40

[quakenet] <laxadedi> trying a fresh build after cleaning cmake + objs files

23:40

[quakenet] <heinrich5991> I hope that doesn't fix it

23:40

[quakenet] <heinrich5991> cmake/make should be able to track that

23:41

[quakenet] <laxadedi> Well, I have no idea, I never liked CMake, too much "magic"

23:41

[quakenet] <laxadedi> A giant pain in the ass to debug that

23:42

[quakenet] <heinrich5991> mh. I just recently learned it and I don't think it's that bad

23:42

[quakenet] <laxadedi> it worked

23:42

[quakenet] <heinrich5991> ahh :(

[ddnet/ddnet] Pull request closed: #1000 Add /modhelp chat command

[quakenet] <heinrich5991> laxadedi: could you try the patch and check what it outputs?

23:43

[quakenet] <laxadedi> I guess it's because the only difference was the includes, afaik, make/cmake dont check timestamps on those files

23:44

[quakenet] <heinrich5991> for our own includes, it certainly does

23:44

[quakenet] <heinrich5991> (and otherwise you'd get serious miscompiles)

23:45

[quakenet] <laxadedi> What steps exactly you want me to do ? I tried to apply your patch but I had no different output when doing cmake .

23:47

[quakenet] <heinrich5991> mhmhmhmh

23:47

[quakenet] <heinrich5991> there isn't an extra line on top of Checking for â€¦?

23:49

[quakenet] <laxadedi> No

23:49

[quakenet] <heinrich5991> mhhmhmmh

23:50

[quakenet] <heinrich5991> → will go to sleep now :(

23:50

[quakenet] <heinrich5991> good night

23:50

[quakenet] <laxadedi> Will do the same, thanks for the hlpe

23:50

[quakenet] <laxadedi> you can pm me later if you want me to help you toubleshot this

yep, thank you

Exported 136 message(s)