DDraceNetwork - developer

<heinrich5991> deen: do you know which server sent the "Cannot write: No space left on device"?

Does hookcoll is lost while you spectating someone?

10:51

I mean, you can see that red line

10:51

but it not indicate if the guy can hook someone, or can hook wall

10:51

it's always red, and hookthrough

10:52

hmm weird, i restarted client and it works now

10:52

curious..

About hook collision, maybe add player collision too, that would show the exact center of tee and its collision box.

11:29

So like red box with cross on the middle

11:31

would be useful, since all skins aren't centered perfectly, which makes tee towers hard

And it would help hooking and shooting ppl

11:39

And landing edges

11:41

And would generaly just make sense

<Henningstone> sounds cheaty to me

11:44

<Henningstone> wait, I'll do a quick mockup to see what it'd be like

Well now that I think about it, hook collision lines are in a way, cheaty too, since they help you aim, and so on...

<Henningstone> in a way you're right

I know

<Henningstone> ok done so far...

12:15

@Mokumoku something like that?

Yea about right...

green circle is hook-collidable, red circle is tee-collidable and the square is for world-collision

Yea

12:18

Howabout center point/cross

ah sorry forgot that x)

No problem!

Here you go:

12:22

12:27

the collision boxes looking wrong on this screenshot is caused by the bumping-around on the tee I'm sitting on

I can see that being usefull

woah

12:32

it loooks.. interesting

12:32

but, i wouldn't give it to normal use

How so

Maybe only as a option

12:33

for auth users

12:33

to test maps etc

12:33

like in csgo, you can enable model wireframe

12:33

but only in cheat mode

right!

cough cough hook collision cough

tbh, collision boxes gives you much more

12:35

than simple usefull hook coll

12:35

maybe instead of those boxes

12:35

we can make option, to make all players looks

12:35

"default"

12:35

?

Hmm would help pro tee towerers like myself

actually, you can do it

12:36

by remove all skins

12:36

but you know, its stupid

Yea

I'm curious why all skins are loaded on startup. Isn't it a waste of memory?

12:40

Why why didn't load it while joining to game

12:40

or something

That would mean frozen tee as well.

Why ?

Show all tees as default, should show frozen tee as default too

You would do it in another thread. He would have default skin, then he's own after load

12:41

ah

12:42

yea, that's right

I see the main benefit of this in a faster client startup for people who have got many skins. The memory consumptions is neglecable imo.

14:08

although tbh I don't even feel a real difference in startup time with the 112 skins I've got...

yes, but we should think about weak computers

computers that don't have 5 MB of ram to spare are probably too weak to run any game made in the past 20 years tho

ask deen then

5 MB?

18:31

100 skins?

approximately... I've looked how much the process's memory usage in the taskmanager would increase when I load the skins

Skin DB has 700 skins, takes 350 MB of RAM for me

18:33

not sure about video ram, same there maybe?

18:33

I think it's reasonable to expect DDNet to keep running on 10-15 year old computers

definitely

ok maybe I did something wrong, I probably I didn't load all skins. I've did the estimation again and observed an increase in ram usage by 15,7 MB, and the 'gfxmem' on the debug screen increased by 63148k

def-

[ddnet/ddnet] Issue opened: #934 Compilation fails

```nim /ddnet-master/src/game/client/components/ghost.cpp: In member function ‘void CGhost::StopRecord(int)’: /ddnet-master/src/game/client/components/ghost.cpp:411:48: error: use of deleted function ‘CGhost::CGhostItem& CGhost::CGhostItem::operator=(const CGhost::CGhostItem&)’ In file included from /ddnet-master/src/game/client/components/ghost.cpp:14:0: /ddnet-master/src/game/client/components/ghost.h:81:8: note: ‘CGhost::CGhostItem& CGhost::CGhostItem::operator=(const CGhost::CGhostIte...

<deen> disable texture compression?

18:46

<Henningstone> yes it was disabled

<Learath2> redix: #934

<eeeee> i have this old patch for on-demand skin loading https://github.com/ddnet/ddnet/compare/master...eeeee:ondemand_skins2

ddnet/ddnet

ddnet - DDraceNetwork, a mod of Teeworlds

19:43

<eeeee> don't remember why i didn't pullreq though

bind k "kill; (wait 5 sec); say /weapons"

20:06

is something like this possible?

no

i thougth of a sleep in rcon and bind sys too

20:11

would be great

@ezy it is, even with ddnet client, you need to parse the client output and (ab)use cl_input_fifo

can u explain what it is?

20:13

never used it

20:13

i heard of rcon fifo

oh, right

20:14

fifo

20:14

b ut not working on windows, if I remember

20:14

yes only works for unixsystems

how do i use it?

20:15

clientside

you can send commands to fifo

20:15

use whatever language you like, and send commands to fifo file

??

20:16

how

it is pretty much writing to a file, except it blocks until someone starts reading

what do i do in detail?

https://github.com/ddnet/ddnet-scripts/blob/c837208e75d8cd633da1138f9c5384d58d264725/servers/scripts/snake.py

ddnet/ddnet-scripts

ddnet-scripts - The magic scripts that make DDNet run

20:17

snake used fifo

20:17

right?

20:18

https://github.com/ddnet/ddnet-scripts/blob/c837208e75d8cd633da1138f9c5384d58d264725/servers/scripts/snake.sh

ddnet/ddnet-scripts

ddnet-scripts - The magic scripts that make DDNet run

mkfifo ddnet.fifo, then set cl_input_fifo ddnet.fifo then you can write to it like this for example: echo some_f1_command > ddnet.fifo

20:19

your paths might be different tho

ah understand (edited)

20:19

live?

fifo redirect messages from file to server/client

yes the commands are instantly executed (at least nearly)

thats crazy

I bet, it's how ddnet send messages

20:20

to world

20:20

at once

that is true, global broadcasts work like this

u could use the other thing for that

20:21

how was it called

econ

econ?

20:21

ye

but, it's simpler to send message to all files

20:21

than connect to every server and send message

20:21

no ?

use whatever language you like, and send commands to fifo file

20:21

languages?

C C++ Python...

i thought i send tw commands?

Yes, but you can use those language to make expressions, statements

20:22

etc

20:22

waiting events

can u give me a sample?

i already gave you a basic shell example

ah

20:24

i understand

20:24

but how do i wait events ?

20:24

like tw events?

you need to parse the clientlog then

can i get tw input

20:24

ah

20:25

ye makes sense

20:25

thats cool

I remember, that once, someone made a bot

20:25

with quiz feature

then i could add a bind bind x echo testcmd0 and if my script read testcmd0 in log it sends all the delayed commands (edited)

maybe it was made with fifo ?

thats nice

yes, except that deen removed the echo command / made it the same as say afair

But then you have to make a program, which will wait for those echo's

ye

20:27

nice

Learath2: considering #934... do you know what version of gcc deen is using?

echo as a say?

20:29

You mean, that echo send message to cchat?

20:29

but if i see, it's indicated as a "-2" player

@redix not really

g++6 and g++7 as well as clang-5 seem to work for me

the compiler tries to use the copy constructor although the move constructor exists and I'm using std::move =\

<Learath2> well he is probably compiling on Debian 6 so maybe gcc5 or gcc 4.8?

Even Debian 7 uses gcc 4.7 which has no full c++11 support :/

<Learath2> redix: think deen was looking to upgrade to the first version with c++11 support

23:01

<Learath2> that's why i said check 4.8

Ah okay. I'll try it tomorrow

bors[bot]

[ddnet/ddnet] New branch created: staging.tmp

23:30

bors[bot]

[ddnet:staging.tmp] 1 new commit

3dab672 [ci skip] - bors[bot]

23:30

bors[bot]

[ddnet:staging.tmp] 2 new commits

0614c03 Update German translation (27 Oct 2017 UTC) - Wuzzy 53e553d [ci skip] -bors-staging-tmp-932 - bors[bot]

23:30

bors[bot]

[ddnet:staging] 2 new commits

0614c03 Update German translation (27 Oct 2017 UTC) - Wuzzy 8227a72 Merge #932 - bors[bot]