Teeworlds - bridge

Teeworlds

IRC / bridge

One-way IRC channel bridge. If you want to be able to send messages to IRC, contact @Dune or @heinrich5991. https://www.teeworlds.com/?page=docs&wiki=rules/irc_rules

Between 2020-03-25 00:00:00Z and 2020-03-26 00:00:00Z

Ryozuki: go for curve25519

what

ecdsa uses curves crafted by NIST, ed25519 uses curve25519 which is not crafted by people with ties to the NSA and is performant

im using ecdsa with NIST P-521 aka secp521r1

10:16

is it bad?

well it doesn't have any glaring security issues that people found, but there were concerns raised about constants used by NIST to craft the curve

well its not to encrypt anything ilegal

10:19

tbh im just working on the https masterserver in go

Ah, yeah isn't really important, any elliptic curve is fine really

does tw have a json lib already in?

10:22

or curl?

10:22

i guess not

no curl, yes json, no improved engine jobs

10:27

I think heinrich5991 has curl linked and working with the old engine jobs

Learath2 if a server registered using a ipv6 should it be allowed to register using a ipv4 afterwards?

yeah, I think my PoC server allowed servers to register with both in any order

10:55

iirc I used a token to identify the server, and made the server contact the master alternatingly on ipv4 and ipv6

wdym by identify

10:56

ah

e.g. if a server first registers on ipv4, how do you make sure it's the same server on ipv6?

i see

I'm not sure if I ever completed that part though, you register, you get a token, you use it in all further requests

but the server may fake the token

10:57

and get listed 2 times?

we hand out the tokens

10:58

hmm 1 ipv4 1 registration is what i enforced, but there usually being thousands of ipv6s in one allocation makes that a bad criteria

11:02

Maybe one /64 one registration? But even that has issues, some ISPs are particularly retarded and don't understand the point of ipv6 so they give you only a /67

11:04

some ISPs give out entire /56s, that's 256 /64s you'd be able to register 256 times :D

11:04

heinrich5991: I bet you investigated this problem, did you come up with a scheme?

11:09

(they really should be assigning at the very least a /56 and IETF recommends an entire /48)

there are things implemented in teeworlds for parallel jobs

11:17

but I think it's barely used

IEngine jobs handle threaded tasks, vanilla doesn't use them much indeed, iirc only for host lookups and loading sounds

Assa: just like vanilla

I used it for my map generator

12:59

works great

tests in https://github.com/teeworlds/teeworlds/pull/2513 failed due to a invalid gpg signature when preparing linux...

Rename some structs and classes to conform with the conventions by ...

13:53

azure.archive.ubuntu.com is down i think

Ryozuki: MSs shitty mirror occasionally breaks because their admins are inept

Learath2 r u in quarantine?

15:38

so much free time but even then im lazy af

yeah

15:44

I'm so unmotivated study that I'm using my time for tw so atleast I don't feel completely useless

15:51

I seem to have trouble making things fit in the boxes given :P

Don't sweat the UI stuff too much, as long as the backend is there :P

15:53

so unmotivated study that I'm using my time for tw so atleast I don't feel completely useless

I can relate

i think my master server will have a admin shell via ssh

15:57

sounds cool

don't we already have econ?

Learath2: we don't have any enforced registration limits right now on the masters

heinrich5991: well worst case you can register a server on every port of your ipv4 address, with ipv6 a single /64 contains sextillions of addresses * 65536 ports

even the first would DoS the masterserver

Either way, it'd be nice if we could fix that attack vector while upgrading to http masters

100 servers per IP

i think ddnet is not even 30 on GER

100 server per ipv4/32 and ipv6/48?

who would need more? if there is a need, what I doubt, just request the master to increase that limit.

16:17

to 110.

16:17

etc.

it was a question whether the /x seems reasonable

Well a /48 is the suggested allocation per end-site as per IETF, and they also suggest that if they allocate a /56 they should still reserve the /48 if the end-site would require more

16:31

However, as I said there are ISPs that ignore the reccomendation and allocate a /64 instead and I'm not sure those people would be responsible enough to reserve atleast a /56 along with it

hm. 100 still feels like a conservative limit that could be applied

16:32

/64 in the server world is probably rare and I'd guess we don't even have 100 servers that are hosted from home

Well hetzner allocates a /64

16:33

We can go with 100 and users can argue with their degenerate isps about it

or we just give them an exception

16:35

hetzner. really.

16:35

I thought they generally reasonable

They are a budget company, I'm guessing their CEO probably doesn't know anything about networking and told the engineers a /64 would be more then enough for anyone

winapi is so fun to work with

even though this message does not contain indicators; given your history I'd say this line was sarcastic

I love how they reinvented everything from scratch AND split the api up between ascii and wide-chars

and their wide chars ("UTF-16") is a fuckup, too

16:55

but you should exclusively use the W versions in modern software, IMO

The best part is that they use UCS2 wide characters everywhere but the entire api is supposed to be utf8 compatible

wdym? utf8 compatible, yes

16:56

you need to convert it to utf-16

why couldn't they stick to utf8 like everyone else? :(

there are three huge users of sorta utf16: java, javascript, windows(.net)

16:57

thank god we have emojis nowadays so that these languages can no longer pretend that utf-16 is a fixed-width encoding

The only reason to like emojis :)

16:59

I still don't get the appeal of utf16 though, utf8 is perfectly capable of representing any character without requiring an extra byte for the most common character set and breaking legacy software

ugh winapi

in the beginning windows etc. used ucs-2

17:00

when unicode said that this was the unicode encoding for characters in common use

17:01

later they realized that 65536 characters weren't going to cut it (edited)

17:01

preaching to the choir: https://utf8everywhere.org/

17:02

(and then they "backwards-compatibly" fixed it by using utf16)

I don't see the appeal of utf16 today either, it's only slightly easier to decode at the cost of breaking compatibility

17:03

I'd probably feel the same way about ucs-2 back in the day

argh, apparently even qt uses utf16

eh, some research reveals that utf-8 wasn't available in 1990 when ucs-2 was defined, so I'll only forgive the windowsfolk if they jumped on the ucs2 ship before 1993 when utf8 was released

17:08

I guess there was already utf1 but it was apparently too inefficient

Learath2 are u fiddling with sdl?

17:08

thats why u are with winapi?

nah, was messing with winapi while getting the modified/created times of files

i only used winapi to capture keyboard input secretly cuz i was rly bored

17:09

and that was the last time

17:09

i hate it

https://cdn.discordapp.com/attachments/620064767788318720/692197146975535114/unknown.png https://cdn.discordapp.com/attachments/620064767788318720/692197380518576138/unknown.png Hi, the one image shows the UI I took from ddnet client and put into my client, the other one is my reworked one in 0.7 style. Is there anything I can do to make the sliders start from the same x value? Its the 0.7 slider option function, but as it seems the beginning is based on the text length, in this case Hue, Lgt and Sat.

17:11

The length of Hue is a bit longer than the others, so you see it starts more right than the others

I don't know anything about the UI code

i dont like tw UI code

17:12

as of now i only copy pasted stuff when adding new things

Its quite simple once you understood it

17:12

but still, its not nice yes

@fokkonaut what do you use for the sliders?

DoScrollbarOption

18:16

i mean technically they work correct, but i want to have all 3 sliders starting at the same point

18:16

which isnt possible with using this function

18:16

It doesnt really matter, would just be the extra point to make it even nicer

18:17

https://github.com/fokkonaut/F-Client/commits/F-Client

That won't work, there is another HSL slider in the code you can steal from

? wjere

18:17

where*

18:17

well it works as i said

18:18

I want it to be seperated in 3 sliders tho

Tee customization has a hsl slider

18:20

You can also add a MinLabelWidth argument to DoScrollbarOption and pass the width of the longest one

hm, discord sucks, I don't know about ipv6, as I have not come across is enough to know whether servers are even given a whole subrange of ipv6 ips, currently hav enot seen this, so don't know ._.

servers are usually given at least a /64 (edited)

Image attachment

20:42

Decoration tab has Alpha, this is bugged

invisible skin pog

@fokkonaut resolution?

1920x1080

20:55

sorry, its marking tab, not decoration

heinrich5991: I think allocations smaller then /64 isn't even considered by the IETF

Exported 129 message(s)