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 2017-09-13 00:00:00Z and 2017-09-14 00:00:00Z

<eeeee> so what's the plan around shaders PR? we have to check that it really improves perf (especially on intel and other weak gfx), and doesn't break on older systems

02:05

<eeeee> could also try to condense some copypasta to avoid adding 3.7k new lines

02:07

<eeeee> should we release some kind of beta client (out of band of regular autoupdate) and ask people to try it?

<deen> eeeee: sure, I can do that when the reviewers think it's ready to be tried

from a quick glance at the source yesterday, it still keeps the old method of rendering

10:12

@HMH AFAIK Vulkan ist the opposite of bloated

10:12

OpenGL is the one that is bloated beyond repair, has a stateful API and so on

[ddnet/ddnet] Pull request opened: #862 removed crypto lib from bam.lua (macOS compile improvement)

Im having some problems with compiling ddnet on some macOS machines. I figured out that the crypto library causes this problem. And it works fine without on all different macOS machines i tested it. I guess this lib isn't needed anyways so i suggest to remove it to make compiling a bit easier c:

@heinrich5991 what you recommend learning first? vulkan or opengl

I said that I don't know either ^^

11:19

I think I read that you usually want to learn OpenGL

i'd learn ogl

11:19

it's been around forever, guess it'll stay for quite some time, and even if it becomes obsolete the knowledge doesn;t

"Vulkan offers programming models that are much closer to how contemporary GPUs work, so if you learn Vulkan, you'll have a better understanding of how the technology really works"

11:20

https://computergraphics.stackexchange.com/questions/3575/should-new-graphics-programmers-be-learning-vulkan-instead-of-opengl

Should new graphics programmers be learning Vulkan instead of OpenGL?

From the wiki: "the Vulkan API was initially referred to as the 'next generation OpenGL initiative' by Khrono", and that it is "a grounds-up redesign effort to unify OpenGL and OpenGL ES into one c...

11:20

well i already started opengl so ill continue with it

[ddnet/ddnet] Pull request closed: #862 removed crypto lib from bam.lua (macOS compile improvement)

11:39

[ddnet:master] 2 new commits

939ea3c removed crypto lib from bam.lua (macOS compile ... - ChillerDragon e689925 Merge pull request #862 from ChillerDragon/master - def-

https://www.twitch.tv/sentdex

@Ryozuki you forgot all the answers that tell you otherwise

AI playus GTA 5

11:45

learning of live

@heinrich5991 what i do?? D:

11:46

i found a really good tutorial for opengl

11:46

but vulkan i only could find the oficial one

top answer: "Should new graphics programmers be learning Vulkan instead of OpenGL?" "Hardly!"

11:46

(i.e. no)

i know

11:46

i read more than that

11:47

they also say its easier to go from vulkan to opengl when u know the first

@Savander GTA5 might be a tad too complex to learn for a bot

11:54

it seems to have understood that if(vehicle == TANK && seepolice) shoot(); which is pretty cool

he's training him

11:55

since may

11:55

i guess

11:55

https://www.youtube.com/watch?v=edWI4ZnWUGg

Stream, FPV, and more data - Python plays GTA p.15

11:57

Soon, cars in real life

11:57

@ChillerDragon is there a reason why you don't compile with cmake?

i have alot more expririence with bam. And it works fine for me.

12:20

Never Change a Running System

kk

12:21

unlike bam, cmake is used by other projects as well btw

<deen> but there are thousands of cmake versions everywhere

@ChillerDragon sooner or later we prolly will nuke bam out of the orbit, so you might want to give cmake a try before then

<deen> anyone want to make sure the cmake build can be used to release?

you're missing cross-compilation for that, mainly?

12:23

or instructions on how to cross-compile, that is

<deen> well, you have my scripts for building

12:23

<deen> if you can change them to work with cmake that would be the best :D

kk

<deen> https://github.com/ddnet/ddnet-scripts/blob/master/ddnet-release.sh

ddnet/ddnet-scripts

ddnet-scripts - The magic scripts that make DDNet run

12:24

<deen> right now it's basically "bam release && scripts/make_release.py"

12:24

<heinrich5991> I was tinkering with the CI automatically uploading the releases, so you don't even have to run them locally anymore (+getting nightlies)

12:24

<deen> yeah, that would be nice too

12:24

<deen> so we could just grab a package from there and release that

12:25

<heinrich5991> although we should prboably retain the ability to do releases without them

12:25

<deen> (even though that would require reproducible builds)

12:25

<heinrich5991> (see the previous mac outage on travis)

12:25

<heinrich5991> do we have reproducible builds?

12:25

<deen> i wouldn't trust them not to have malware either

12:25

<deen> probably not

12:25

<deen> git version might land in

12:26

<deen> ah wait

12:26

<deen> that's ok

12:26

<heinrich5991> that's not non-reproducibele

12:26

<deen> anyway, could check this: https://wiki.debian.org/ReproducibleBuilds/Howto

12:26

<heinrich5991> (in fact, I think that our build doens't work without git right now)

12:26

<heinrich5991> (which should probably be fixed)

12:26

<deen> that's bad as well...

12:26

<deen> I don't have git on mac

12:26

<heinrich5991> I didn't change anything I just saw the script

12:26

<deen> ah, then it works

12:27

<deen> just throws a warning

12:27

<heinrich5991> print('#define GIT_SHORTREV_HASH "%s"' % os.popen('git rev-parse --short HEAD').readline().strip())

12:27

<heinrich5991> that thing should prboably be fixed to allow for git not to be present

12:29

<heinrich5991> deen: btw, reproducible builds across OSs is probably going to be hard

12:29

<deen> ok

12:29

<deen> then I will prefer to keep building the binaries myself

12:29

<heinrich5991> i.e. with cross-compiling and normal compiling both trying to produce the same executable

12:29

<heinrich5991> ok

[ddnet-web:master] 1 new commit

084568d Add some changelog - Learath2

12:53

[ddnet-web:master] 1 new commit

9555fa9 One too many <li> - Learath2

languages in English are written with capital letters

<Learath2> I added some changelog to the past couple of builds

12:54

<Learath2> oh missed that one

[ddnet-web:master] 1 new commit

e5eee3e Fix capitalization - Learath2

<Learath2> well the S for simplified could have been lowercase

13:00

<heinrich5991> nah, don't htink so

13:00

<heinrich5991> lemme check

13:02

<heinrich5991> ok, wikipedia writes it in lowercase

[ddnet-web:master] 1 new commit

199216f Add the changelog for 10.8.6 - Learath2

<Learath2> feel free to fill in 10.6.7 and 10.6.6, couldn't find anything to put there

13:07

<deen> (you can also just remove them from the download page)

13:07

<deen> they were just quick bug fix releases

13:08

<deen> not sure which ones to keep listed in the long term btw

13:08

<deen> site is kind of long already

pagination?

<deen> I don't believe in pagination, as is obvious when you look at ddnet.tw

why not

<deen> pages are used to increase ad revenue

And to improve perforemence for clients

13:11

he shouldn't load tons of information at once (edited)

13:11

if not needed

<deen> maybe browsers shouldn't load the entire site immediately

Maybe, but I still prefer pages instead of almost infinite scroll :P.

[ddnet-web:master] 1 new commit

505fe93 Remove 10.6.{6,7} - Learath2

<deen> i guess it's a matter of how much. news articles shouldn't paginate

13:16

<deen> ddnet.tw probably should

13:16

<fstd> def needs endless scroll

13:16

<fstd> all the hip websites have it

Articles shouldn't be paginated? You mean, every post should be displayed on mainpage

<fstd> also a hamburger button. is there even a hamburger button?

or certain post shouldnt be paginated

13:17

There were concept

13:17

you really don't like modern websites, right?

<fstd> i hate it when websites have html in their javascript

13:19

<fstd> like html is so 1990s

modern website don't means, full of javascript everywhere

13:20

because WE can

13:20

ofc it does

you can do a lot with pure css

13:20

as i do

try loading a modern reactjs angularjs w/e web app

13:20

without js

anway, that what you are describing is called one page website

13:20

withr eact, angular

13:20

react*

<fstd> i tend to call it app-grade

13:21

<fstd> because it's obviously designed to look like a smartphone app

13:21

<fstd> which is modern and hip by definition

yup, because webdevs forces

13:21

"mobile first"

13:21

look

"<deen> pages are used to increase ad revenue", but we don't have ads (?)

13:33

definitly the downloads page needs infinite scroll or pagination

13:33

also the skin database

13:34

and ddnet.tw/releases/all/

I would like to see in dwonload section

13:34

maybe list of versions

13:34

isntead of download links, and descriptions

13:35

then if you want to see what were changed, then you click and you have. but it would be a long journy to change it

13:35

i guess

13:35

journey

<deen> write a script to do it

[ddnet/ddnet] Pull request opened: #863 Tidy up circle.yml, test artifact collection

Let's see if artifact collection works

<Learath2> god knows where that failed :/

<Learath2> heinrich5991: any idea? https://circleci.com/gh/ddnet/ddnet/1677#tests/containers/0

17:29

<Learath2> eh think i figured it out

17:29

<heinrich5991> the fancy log doesn't show up

17:33

<Learath2> well it's not xz-utils :/

17:33

<heinrich5991> can you paste the log somewhere?

17:34

<heinrich5991> circleci doesn't work for me

17:34

<Learath2> http://paste.pr0.tips/fL

17:35

<Learath2> but artifact generation does work \o/

17:36

<heinrich5991> yea, I know where that error comes from

17:36

<Learath2> oh cmake might be too old to have TXZ

17:36

<heinrich5991> no, the error is something different

17:37

<Learath2> well the escape sequence thing doesn't happen on my computer, and the generated TGZ archive looks fine on circle

17:37

<heinrich5991> it does? hmmm

17:37

<Learath2> https://1678-22821674-gh.circle-artifacts.com/0/tmp/circle-artifacts.e4dAqaE/DDNet-10.8.6-linux_x86_64-portable.tar.gz

17:38

<Learath2> oh nvm

17:38

<Learath2> the TGZ is empty

17:38

<Learath2> heinrich5991: sry :P so what's the actual problem?

17:42

<heinrich5991> sec

17:45

<heinrich5991> CPACK_VERBATIM_VARIABLES

17:46

<heinrich5991> https://cmake.org/cmake/help/v3.9/module/CPack.html

17:46

<heinrich5991> I guess cmake on circleci is reasonably

17:47

<heinrich5991> old

17:49

<Learath2> well we could build on a newer ubuntu container but not sure if the builds will be any use then

17:50

<heinrich5991> we could also install a newer cmake version

17:55

<heinrich5991> what did you do for josh back then, fstd

17:56

<heinrich5991> w/c

<fstd> discreetely disposed of the ever increasing amount of dead hookers he produced

18:11

<fstd> w/c

18:16

<heinrich5991> :P

<heinrich5991> string(REGEX REPLACE "\\" "\\\\" ESCAPED "${STRING}")

20:36

<fstd> looks like production-quality code

20:36

<heinrich5991> not that this replaces \ with \

[ddnet/ddnet] Pull request opened: #864 Add compatibility with older CMake for `package`

<heinrich5991> Learath2: fixed it

20:42

<heinrich5991> merge it and try your stuff again

20:44

<heinrich5991> deen: this might fix your problem as well btw, not sure

Exported 184 message(s)