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 2020-04-24 00:00:00Z and 2020-04-25 00:00:00Z

Is anyone familiar with the algorythm in gameworld.cpp? I would like to know if its possible to force a player that has the same IP as you to a specific ID, which will never change for this guy for my own map. So he is always visible in my scoreboard no matter how far away he is

02:29

I mean it should be possible, as its also possible to force the ID 15 out, the empty client that is for sending chat messages from the further away clients

02:29

but I couldnt get anything to work :/

you want your dummy to be always visable?

08:17

These kinds of same ip queries get very expensive very quick, you might want to begin with a hashmap of ip -> client-id. After that you can force everyone in the same ip bucket as you to occupy the first couple ids

08:17

the rest you can use the distance algorithm for

08:17

I'm wondering how well the collision stuff is working with 128 players

08:18

I'd be surprised if the server didn't start slipping with that many entities close by

[ddnet-scripts:master] 1 new commit

e2c2b84 locations -> all-locations - def-

[ddnet-scripts:th-auth] 1 new commit

a1f2387 Add mod authentication - Learath2

10:11

[ddnet/ddnet-scripts] Pull request opened: #18 Add mod authentication

I tried to minimise changes

I want whatever the people who came up with http basic authentication is on

11:31

If you send a 401 safari will still save the credentials that turned out to be wrong, chrome will just ignore the response and display a blank page, firefox will also happily save the wrong credentials

isn't that more of a problem of the browsers?

11:32

but yes, the experience sucks

Yeah I guess it's the fault of decades of backwards compatibility trying to be kept

11:33

unlike C/C++ implementation quirks, there is no reference for browser quirks, browsers don't tell you how they implement behaviour undefined by the standard

can someone help me with how the automap data is packed into the datafile?

I guess I wasted the time though, we really should instead let mods download demos, not th files

its from gameitems_ex.h

11:36

its supposed to have something to do with uuids, but i dont really understand how that works

You mean mapitems_ex?

ouch yes

It's a little tough, there is a translation table within the datafile that translates itemids above OFFSET_UUID

hm so its fundamentally different to the other mapitems

11:40

that does sound a bit harsh

11:40

could you tell me how it works?

You look for items of type ITEMTYPE_EX

11:41

These are the table entries

11:42

you extract the UUID from these (look at CItemEx::ToUuid), you match it with the uuid for "mapitemtype-automapper-config@ddnet.tw"

11:43

That items ID is the "Internal type ID" of the automapper config item

11:44

(the item that has matching uuid that is)^^

11:44

After that you can use the internal type id like any other type id

okay, thanks for the leads :)

11:46

ill come back once im stuck again

@heinrich5991 ^^ sounds about right, right?

yea, what you said sounds about right

not sure if i was able to follow those instructions completely what would make sense to me:

each item of the type ITEMTYPE_EX defines a included ex_itemtype
which of the types it is, can be found out by trying to match the uuid of a known type (for example the uuid for "mapitemtype-automapper-config@ddnet.tw") with the item_data of that ex_itemtype item
then the matched itemtypes going are in id order 0xffff - n (n being the number of the itemtype_ex item it was defined in)

I'm not quite sure of the very last statement

12:49

The ID is part of the datafile item, it's i->m_TypeAndID&0xffff

12:50

(which is nothing but the first 2 bytes of m_TypeAndID)

if you want to find out the "actual type" of a type ID typeid

12:52

you look up the datafile item (0, typeid)

12:53

if you find a UUID item there, you'll have your type

(0, typeid)?

you find layer number 5 in (LAYERTYPE, 5)

0 isn't ITEMTYPE_EX, is it?

12:59

I thought it was 65536 for ITEMTYPE_EX

65535 i think ^^

maybe

12:59

if 0 was already taken

typeid as in the internal typeid?

typeid as in typeid in the map on the disk

I think this is only valid for typeid > a certain number though

yes, halfway to max

13:00

from there

13:01

so >= 0x8000, probably

I think you also have to iterate through the ITEMTYPE_EXs and look for the id, I don't see the index getting fixed or these items being ordered in the code

there's a function for it though IIRC

13:03

GetItem(typeid, id) IIRC

GetItem takes an index, FindItem is probably what you are looking for

13:05

@heinrich5991 do you remember why you didn't just subtract the offset to get to an index?

wdym?

13:06

offset into what?

offset into the table of itemtype_exs

13:07

no that didn't sound right

13:07

I mean something like typeid - 0x8000 to get the index of the itemtype_ex item with the uuid

13:07

that'd be better then the linear lookup into the table, no?

hm. I don't know if anything uses the map item indices

13:08

I hope not, because then you don't have to care about how the items are laid out in the file

13:09

the linear lookup is a fault of the implementation, not the map format

oh, the items are not in contiguous blocks?

the items are in contiguous blocks

13:10

in my implementation, I have a hash map from (typeid, id) -> item

13:10

i.e. constant-time lookup

in your implementation as in libtw2?

yes

I guess I don't really know the scope of the id, is id global in the datafile?

(typeid, id) is globally unique

I don't get why this is implemented like this in tw then, are items fixed size?

13:13

arent*

because tw doesn't have hashmaps, I guess

13:13

(that's the same problem that snapshots also have)

@heinrich5991 can index differ from id?

index is unrelated to id

13:18

index is the index of the item in the file

and id?

id is something you specify while saving an item

13:19

you can specify a 16 bit item type and 16 bit item id when saving an item

13:19

they're free form, not interpreted by the map format itself

13:19

they must be unique though

what is it used for? I don't see what they provide. (typeid, index) would already uniquely identify an item in a datafile

index already uniquely identifes an item in a datafile

question for the professionals here. I am having trouble opening demos. I already downloaded tw new and deleted config etc. still i cant open them. I asked other people to open them, they are able to. Do you have any idea? I am using 0.6.4 12.7.3 version

14:14

with one demo, tw crashes, and with 99% of the other demos, i just cant open them. 1% i can open

could you try again with the newest version of ddnet?

yes I will try

I think we had some demo-related fixes recently

can I open demos even if I dont have the maps downlaoded?

i think so

yes, you should be able to

+can confirm that i cant open demos people send me from new ddnet without also using new ddnet

that's expected, I think

14:19

we changed the demo format

funny thing is, most of these demos are my own. and I never used any higher version than 12.7.3

14:19

But I will try with newest client :)

thanks

14:20

if that doesn't work, please send me one of the demos

works clean af, thank you @heinrich5991

nice

14:22

thank @Learath2 for that

you can also thank me for breaking it

14:23

50/50

@heinrich5991 what about the players-cache

15:15

if you wont port it to py3 i will make it work with py2

yea yea, no need to threaten

lmao

wtf @ ddnet scripting

15:23

a grep to extract some values from our own json

i think i got it know apart from one detail

15:27

you match it with the uuid for "mapitemtype-automapper-config@ddnet.tw"

15:28

did you mean by that that i can somehow get the relevant uuid from the string "mapitemtype-automapper-config@ddnet.tw"?

yes

how do i do that?

UUID version 3, base UUID "e05ddaaa-c4e6-4cfb-b642-5d48e80c0029" https://github.com/ddnet/ddnet/blob/974df08539f2ae052725cb238be1f1620063c203/src/engine/shared/uuid_manager.cpp#L9 https://github.com/ddnet/ddnet/blob/974df08539f2ae052725cb238be1f1620063c203/src/engine/shared/uuid_manager.cpp#L27-L47

DDraceNetwork, a cooperative racing mod of Teeworlds - ddnet/ddnet

hm would someone be mad at me if i just hardcode the number? :p

the UUID? no

me

ddnet

1

@Patiga hardcoding the ID however, is going to break in the future

15:36

as we add more UUID map item

15:36

s

15:37

you can generate the uuid with the uuid tool in ddnet. alternatively, you can get a debug dump of uuids known to the server when you start with debug 1

@Patiga besides, whatever highlevel language you are using probably already has a library for uuids

isnt there just one so far?

for map items, yes

eh, for the sake of completeness i guess

hm. I mean we probably are going to add more map items in the future

do i actually need a uuid library? it seemed to me that i will just throw around 128 bit integers and md5

don't need a UUID library if you don't want, you can hardcode the UUID

15:42

just the map item type ID shouldn't be hardcoded

15:42

UUID vs map item type ID

hm how is hardcoding uuids going to break in the future?

it's not

15:42

this was about hardcoding map item type IDs

ah

15:43

then im probably going to skip that hassle

@Patiga if you don't mind me asking: What are you making?

im making a (hopefully sensible) map parser in rust :) (edited)

@deen where does maps/{mapname}.msgpack come from?

@heinrich5991 https://github.com/ddnet/ddnet-django/blob/master/scripts/map_release.py

ddnet/ddnet-django

Contribute to ddnet/ddnet-django development by creating an account on GitHub.

ty

18:11

@Learath2 do you know where I can find that map_properties tool?

18:12

which map_properties on ddnet.tw didn't find it

@heinrich5991

django@ddnet:~$ which map_properties
/home/django/bin/map_properties

ah, it's my libtw2 tool

yep, I just figured it out

it has a nice --help

[ddnet/ddnet-scripts] Pull request closed: #18 Add mod authentication

[ddnet/ddnet-scripts] Pull request opened: #19 Allow non-interactive queries

I ended up moving a lot of things around :/

CENSORSHIP!

it was just flood

19:41

@heinrich5991 can you also take a look at that pr?

php

19:41

yes

(it's really hard to modify this file blind, I don't have access to the server this is hosted on so can't test it)

wdym, blind modification

19:43

ah

19:44

without knowing the file paths

also apparently my editor decided all the indents should be tabs

@Learath2 if($type === "html") { ?> that works??

yep

wtf @ php

why shouldn't it work?

because stuff out of php environment didn't belong to control flow in my mind

(this is a very ugly way to do php btw)

yes

usually presentation and logic is split with a template engine

but if you do it the right way, you might as well use a saner programming

"s/$/ language/"?

language

tbh, I like php

I think it has too many footguns

not sure if it's the nostalgia but I don't really see anything better for the web

19:52

I could see js for rest apis but that's about it

I like too php

19:52

Php is great

what about python? @Learath2

After using it for a while, I don't think I like python much at all, especially the module/package system

19:54

for programming apis I think I could use something like flask or falcon, but for presentation python just doesn't integrate well into the web stack

I also don't think it's the best (but the language is a bit old, it might be fixed eventually, there are some movements to improvements)

19:54

does php fare better in that regard?

Php has very mature templating engines and frameworks

@ modules I meant

19:54

but templating is also fine

19:55

python also has templating engines, did you try some?

19:55

it worked fine when I did a little django work

eh, atleast I never quit php because I couldn't get a module included

19:55

composer usually works great

19:55

you just include vendor/autoload.php and composer sets things up

19:56

I think django is seriously a crime against humanity

19:57

I have never seen such a bloated piece of software with such horrible documentation

19:57

I worked with Jinja2 and flask, it did the job, wasn't too painful

I worked with django for a website and flask for APIs so far

jinja was written by the same guy who wrote twig (the template engine usually used with symfony) so atleast the template language made me feel right at home

20:00

but that's really not the point in which I think python doesn't integrate well, uwsgi itself feels like we are bolting on python to the web

learath will never find anything he likes

20:00

he likes to be the niche type

20:01

hehe im c00l

Not like I'm using some weird language like zig

php is the opposite of niche

If I was commissioned to do web work, I'd use php + symfony

well i dont like server side rendering anyway

20:02

i always go with golang backend and react frontend separated

If I had to use python, I'd go for flask or falcon

20:03

If I'm not doing the backend with the frontend, I'd still go for a symfony backend and a vuejs frontend

https://tailwindcss.com/

Tailwind CSS - A utility-first CSS framework for rapidly building c...

Documentation for the Tailwind CSS framework.

20:03

maybe u will like this

been using pure.css a lot lately

this is more pure

but this also looks very interesting

imo its the best

20:05

im only using this now

20:05

u need to know css tho

20:05

not ur wannabe css framework

if you have something to learn web design, I'm open

20:07

I have no idea about htat

eh, it seems this requires you to be using some packager though @Ryozuki

only if u want to customize it

it'd be much nicer if they used sass like every other customisable css framework

wat

20:08

no

20:08

thats horrendous

20:08

u dont understand how this framework works

20:08

im in middle of dota ranked

20:08

cant talk much

@heinrich5991 turns out there are no good resources out there

20:09

I had to learn each piece I wanted to learn myself

i also learn by myself

20:10

by doing

20:10

like everything

20:10

i know u like ur c books

I rather refer to the documentation for each piece, but to each their own

docs != book

20:13

ofc i look at mdn

mdn felt like a good resource, yes

20:13

but it's only a doc, not for learning

@Ryozuki doesn't this css framework compile the css file at build time? using a node script as far as I can see

u can use a cdn

20:14

thats to customize it

20:14

i said it

and I'm asking why is the customisation done with a custom node script instead of sass variables like every other framework?

20:15

as in what's the advantage to this approach?

its to generate css classes

20:17

its way better than sass at creation extensions

20:17

u may have mt-2

20:17

margin top 2

20:17

u can do hover:mt-4

20:17

etc

20:17

i cant explain rn

@heinrich5991 you should maybe migrate away from try! in libtw2 before it disappears?

it's not going to disappear

(idk the deprecation schedule of rust, maybe it'll stay forever)

but yea, eventually I'll migrate it away from try!

21:33

@Learath2 but rust's deprecation schedule can be roughly summarized as: if it's not a security issue, code will continue working

what is the "use high dpi" check for?

if you have a hidpi display, this determines whether ddnet will run at the native resolution or not

22:55

if you have a resolution smaller than 3840x2160, you likely don't have a hidpi display

22:55

and then the option doesn't matter

alright. so i can just uncheck it

If you don't have a highdpi display, it won't change anything

it doesn't matter if you check it or not

Exported 271 message(s)