Teeworlds - bridge

[quakenet] <Dune> hello

13:25

[quakenet] <heinrich5991> we could collect the source files using a glob in cmake, but is there a big benefit to that?

13:25

[quakenet] <heinrich5991> as Dune said, adding/removing source files is rare, so that doesn't really matter

13:26

[quakenet] <Dune> heinrich5991, well that means you need to know that you have to modify the cmake file everytime you commit a file addition/removal

13:26

[quakenet] <heinrich5991> yes. how often does that happen?

13:26

[quakenet] <Dune> infrequently

13:26

[quakenet] <heinrich5991> I've been bitten by files being compiled by bam although they were only left-overs from a stash or so

13:26

[quakenet] <heinrich5991> the inverse problem exists as well (with bam)

13:26

[quakenet] <Dune> hm :/

13:27

[quakenet] <Dune> I think both positions are legit

13:27

[quakenet] <heinrich5991> I think adding/removing a line for files is not such a big deal as it doesn't happen very often

13:27

[quakenet] <Dune> I personally think consistency with the behavior of bam would be nice

13:27

[quakenet] <Dune> in both cases it's not a huge deal yes

13:28

[quakenet] <heinrich5991> but cmake should also work without translations etc. in place

13:28

[quakenet] <heinrich5991> I should fix that I guess

13:28

[quakenet] <Dune> you could also argue having to deal with potential problems of left-over files is rare too

13:28

[quakenet] <heinrich5991> yes

13:28

[quakenet] <heinrich5991> it is

13:28

[quakenet] <Dune> ah yeah that one is important

13:28

[quakenet] <Dune> by the way https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=921274

13:28

[quakenet] <Dune> there is a debate about that SSE stuff Oy

13:31

[quakenet] <Dune> heinrich5991, tbh I'm annoyed to have to carefully maintain and test cmake, because I don't like having to deal with cmake. so I'm biased in that regard ^^

13:37

[quakenet] <Dune> I'm surprised default filters aren't properly localized btw

[quakenet] <Dune> some people on transifex have requested for ngettext support because some languages don't play so nice with using plural for everything (e.g. %d players online) :/

13:50

[quakenet] <heinrich5991> yea, we should probably do that at some point

13:50

[quakenet] <heinrich5991> e.g. russian has different plural for 2-5, 21-25, 31-35, etc.

13:52

[quakenet] <Dune> I don't think there is an open issue

13:55

[quakenet] <heinrich5991> https://github.com/teeworlds/teeworlds/issues/2030

Handle languages with non-germanic plural rules · Issue #2030 · ...

For English and German, it's easy, 1: singular, everything else: plural. For other languages, there are sometimes multiple plural forms, e.g. in Russian, we have 1: singular; 2-5, 21-25, 31-35,...

13:55

[quakenet] <heinrich5991> fixed

13:55

[quakenet] <Dune> :D

[quakenet] <Oy> Dune: it didn't build without the flag?

14:03

[quakenet] <Dune> not on linux 32 bit iirc

14:04

[quakenet] <Dune> (nor 64 bit?)

14:04

[quakenet] <Dune> hold on I try

14:05

[quakenet] <Oy> that discussion/patch suggest otherwise

14:05

[quakenet] <heinrich5991> 64bit worked without the flag

/usr/lib/gcc/i586-linux-gnu/4.9/include/xmmintrin.h:1238:1: error: inlining failed in call to always_inline 'void _mm_pause()': target specific option mismatch
 _mm_pause (void)

[quakenet] <Dune> (32 bit)

[quakenet] <day> is the teeworlds jungle background screen time dependent? i.e in the night it's darker than during the day?

15:23

[quakenet] <day> screen/theme

15:27

[quakenet] <day> is this a skin render bug? The skin looks like a mix of two different skins https://i.imgur.com/ULpkpeb.png

15:28

[quakenet] <day> or rather three. (blue)kitty, redboob, cammo

15:29

[quakenet] <day> nvm. i just found the custom skin feature :^)

[quakenet] <Dune> themes have day/night time detection yes

15:40

[quakenet] <Dune> you should read the patch notes :D

[quakenet] <Oy> Dune: seems that was patched in gcc 5 years ago https://gcc-patches.gcc.gnu.narkive.com/Ypx83P9u/patch-rfa-move-x86-mm-pause-out-of-pragma-target-sse-scope

16:00

[quakenet] <Dune> oh, interesting find, Oy

16:03

[quakenet] <Oy> could remove the flag on default and set a config option for that, which you can turn on for old distros

16:04

[quakenet] <Dune> hmm. how comes I don't have it in my gcc :/

16:04

[quakenet] <Dune> hold on

16:04

[quakenet] <Dune> 5 years sounds long

16:05

[quakenet] <Oy> seems u are using 4.9 and the guy from the link has 4.10

16:07

[quakenet] <Dune> yeah I used an old distro for 32 bit :|

16:07

[quakenet] <Dune> > could remove the flag on default and set a config option for that maybe we should propose that fix on the tracker?

16:12

[quakenet] <heinrich5991> we can just manually set the flag when we build it on an old distro I guess

16:15

[quakenet] <Dune> something needs to be done for packaging though

16:16

[quakenet] <Dune> e.g require gcc >= 4.10

16:17

[quakenet] <heinrich5991> this is the things that distributions figure out by themselves, normally (which versions of gcc they need to compile packages)

16:19

[quakenet] <Dune> that would be hard to figure, it should be somewhere in the notes

16:19

[quakenet] <Dune> wouldn4t it?

16:24

[quakenet] <Dune> heinrich5991, there is a stray commit in your PR (?)

16:24

[quakenet] <heinrich5991> oops

16:25

[quakenet] <heinrich5991> fixed

[quakenet] <heinrich5991> ok, so reordering substitutions is now possible

17:23

[quakenet] <heinrich5991> Dune: do you know where this could be documented so translators know of this possibility?

17:23

[quakenet] <Dune> hmm

17:23

[quakenet] <Dune> it sounds difficult to do that on Transifex

17:24

[quakenet] <Dune> There is always https://github.com/teeworlds/teeworlds-translation/wiki

teeworlds/teeworlds-translation

Language files for teeworlds - we use transifex, please read the wiki - teeworlds/teeworlds-translation

17:24

[quakenet] <Dune> and then use the announcements at https://www.transifex.com/teeworlds/teeworlds/announcements/

The Teeworlds translation project on Transifex

Translate Teeworlds into your own language.

17:25

[quakenet] <Oy> Dune: could u test on ur 32bit system using this bam.lua https://pastebin.com/yGdaXRk2

CheckVersion("0.5") Import("configure.lua") Import("other/sdl/...

17:26

[quakenet] <Dune> okay

https://pastebin.com/YbLVd14d

[Bash] bam: setup done [1/2] [1] c++ src/engine/client/client.cpp...

[quakenet] <Dune> I had to copy the patch so I might have messed up something but it doesn't seem to work, Oy

17:38

[quakenet] <Oy> did u do bam config?

17:38

[quakenet] <Dune> yes

17:38

[quakenet] <Oy> did the compiler option turned to no?

AUTO buildwithoutsseflag no

[quakenet] <heinrich5991> you could try replacing the == 0 with a !

17:40

[quakenet] <heinrich5991> !config.buildwithoutsseflag.value

17:41

[quakenet] <heinrich5991> (I don't know how lua handles 0 == false or similar)

17:42

[quakenet] <Dune> it doesn't seem to enter the if

17:44

[quakenet] <Dune> ok so == false fixes it

17:44

[quakenet] <Oy> yeah false

17:44

[quakenet] <Oy> good

17:49

[quakenet] <Dune> that's a smart fix, packagers will have to use bam though

17:50

[quakenet] <Oy> could be done for cmake too i guess?

17:51

[quakenet] <Dune> I don't know cmake at all

17:51

[quakenet] <Oy> seems to have a try_compile - so could do the auto-detect that way

17:52

[quakenet] <Dune> cool

17:52

[quakenet] <Dune> we should notify downstream

17:52

[quakenet] <Oy> should also fix the stuff from the guy last time - some window manager thing

17:52

[quakenet] <Dune> was it day?

17:52

[quakenet] <Dune> ah nvm

17:53

[quakenet] <heinrich5991> I can do the cmake one

17:53

[quakenet] <Dune> cool, thx

17:53

[quakenet] <Oy> yeah, they should probably use the patch we added instead of removing the msse2 flag in general

17:53

[quakenet] <Oy> good

17:53

[quakenet] <Dune> is there any way to fix that sdl legacy mode?

17:54

[quakenet] <Oy> yeah, that was the thing

17:54

[quakenet] <Oy> yeah

17:54

[quakenet] <Oy> we could enable the newer window manager for x11 per default

17:55

[quakenet] <Oy> and set a console config, to turn it off in case sth goes wrong

17:55

[quakenet] <Oy> cause sdl reports there might be issues

17:56

[quakenet] <Dune> this would be only in linux builds or?

17:56

[quakenet] <Oy> yeah

17:56

[quakenet] <Oy> mac uses x11 too i guess

17:56

[quakenet] <Dune> ah, right

[quakenet] <Oy> too bad i can't remember the guys name, so that he can test again

18:17

[quakenet] <Dune> maybe that should be added to the FAQ if someone has gfx issues on unix

18:17

[quakenet] <Dune> gfx_use_x11xrandr_wm 0

18:17

[quakenet] <Dune> the guy was verdre

18:23

[quakenet] <Oy> yeah, would be good to add a faq entry for it, in case someone has an issue with the window or window manager

18:23

[quakenet] <Oy> right, but doesn't seem to be around

18:25

[quakenet] <Dune> he's not on the gnome irc and you can't pm on gitlabn

18:31

[quakenet] <Oy> another thing: 208 commits since 0.7.2. we should do some 0.7.3 in the near future

18:36

[quakenet] <Dune> that is a good point

18:38

[quakenet] <Dune> I haven't heard of zatline for a month unfortunately

18:38

[quakenet] <Oy> :(

18:39

[quakenet] <Oy> we really need a bot flag, skin

18:39

[quakenet] <Dune> so gfx issues may get stuck, yeah :/

18:39

[quakenet] <Dune> https://github.com/teeworlds/teeworlds/issues/1911 would be nice to fix too

tileset bugs in 0.7 · Issue #1911 · teeworlds/teeworlds

I just looked into the mapres in 0.7 and found some bugs, some are minor, some not. I made a gallery on imgur: https://imgur.com/a/dQRWfrd

18:40

[quakenet] <Oy> yeah, just wanted to link it here :)

[quakenet] <rand> about lua, anything different than 'false' or 'nil' (undefined) is coerced to 'true'

19:09

[quakenet] <rand> (so '0' is true)

19:10

[quakenet] <Dune> interesting, thx

19:11

[quakenet] <Dune> so it was a if(<bool> = <int>) and lua converted the int to bool instead of the bool to int

[quakenet] <day> isn't that the normal procedure? e.g. C. every int != 0 => True

22:04

[quakenet] <Dune> I mean, it could do 0 == false => 0 == 0 => true

22:05

[quakenet] <day> and what would 0 == true => 0 == ? => ? become?

[quakenet] <Dune> so what you mean is there is no bool -> int conversion in lua?

22:15

[quakenet] <day> me? I don't know, never writen lua. i just find the auto casting from int -> bool intuitive, as bool is a subset of int.

[quakenet] <Dune> C/C++ has autocasting from bool -> int and it's bitten me a few times

22:23

[quakenet] <day> so false becomes 0. and true becomes 1?

22:24

[quakenet] <Dune> yes

22:24

[quakenet] <day> but if that was true then '2 == True' would be false?

22:24

[quakenet] <day> and afaik everything !0 is True in C

22:25

[quakenet] <day> thus my understanding C autocasts Int -> Bool

22:25

[quakenet] <Dune> yeah 2 == True should be false iirc

22:25

[quakenet] <Dune> not sure

22:25

[quakenet] <day> never. lemme check

22:26

[quakenet] <day> well C doesnt really have bool

22:27

[quakenet] <day> but would you agree that if(2 == True) is synonymous with if(2)?

22:28

[quakenet] <Dune> try with C++

22:28

[quakenet] <Dune> huh

22:28

[quakenet] <Dune> never seen it like that

22:28

[quakenet] <Dune> I guess you're right then

[quakenet] <heinrich5991> 2 == true doesn't evaluate to true in C/C++ or python

22:39

[quakenet] <day> heinrich5991: what is 'true' in c?

22:39

[quakenet] <day> it's not a keyword that exists afaik

22:39

[quakenet] <day> but 'int 2' does evaluate to true

22:40

[quakenet] <day> or rather very thing not 0 is evaluated as true

22:42

[quakenet] <heinrich5991> day: I think it exists from c99 onwards

22:42

[quakenet] <day> regarding python "if 2:" is interpreted as 'true' but "if 2 == True" is interpreted as 'false'

22:43

[quakenet] <heinrich5991> I think our disagreement stems from the fact that you consider everything that evaluates the "then" branch of an "if" as "true" and everything else as "false"

22:45

[quakenet] <heinrich5991> in both python and c/c++ false and 0 behave pretty similarly and true and 1 do as well

22:45

[quakenet] <heinrich5991> try e.g. True + True in python

22:45

[quakenet] <day> you are right about bool existing. its in 'stdbool.h'. and "2 == true" is false

22:48

[quakenet] <day> so this means that 'if' uses its own 'true/false' logic?

22:49

[quakenet] <day> i find it rather odd that 'if(2)' is interpreted as 'true', but if(2==true) is not

22:49

[quakenet] <heinrich5991> the logic for if is (without looking into the standard) that everything nonzero will cause the "then" branch to be executed, and 0 will cause the "else" branch to be executed

22:50

[quakenet] <day> which is equivalent in my mind to "everything is seen as true that is non zero."

22:51

[quakenet] <heinrich5991> but true is just one possible value that will cause this behavior

22:51

[quakenet] <heinrich5991> there are many more :)

22:52

[quakenet] <heinrich5991> in python, if calls the __bool__ function on the object passed to the if

22:53

[quakenet] <day> i see

22:54

[quakenet] <day> so true/false in c literally act like integer '1' and '0'. the implementation is so shady that 'int a = true + true' nets you a == 2 :"D

22:56

[quakenet] <heinrich5991> yes. also works in python

22:56

[quakenet] <day> so in reality bool doesnt really exist imho. if(1 == false) is equivalent to if(1 == 0)

22:56

[quakenet] <heinrich5991> if you want to call it that, yes

22:56

[quakenet] <day> ah. thats why you mentioned true + true in python. didnt expect it would happen there o0

22:57

[quakenet] <day> i mean true/false are concepts, whereas 1 and 0 are numbers