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 2020-02-07 00:00:00Z and 2020-02-08 00:00:00Z

._. anyone wanna give some feedback on this? https://github.com/jxsl13/twapi

00:55

rand or heinrich might be the goto people to ask :)

looked into a random file, https://github.com/jxsl13/twapi/blob/master/compression/varint.go

an attempt to port the tw api tools to Golang. Contribute to jxsl13/twapi development by creating an account on GitHub.

15:49

size of variable-sized integers do not depend on the host machine, teeworlds assumes int is 32 bit

15:50

for that, the variable-sized int can be between 1 and 5 bytes

15:51

if value < -3.6028797e16 || 3.6028797e16 < value { this is a very weird range, should be value < -2147483648 || 2147483647 < value

15:52

intSize := unsafe.Sizeof(value) this shouldn't happen in a network protocol. don't let network datatypes depend on the host's datatype width

15:52

I just looked it up, int in go varies between 32 and 64 bits, but teeworlds integers are always 32 bit

hm k

15:54

I was kinda curious, whether go's varint is the same as teeworld's varint. it seems to be a seemingly often used format, otherwise it would not be part of the standard library

probably not

15:57

    a := [size]byte{}
    b := a[:]

15:57

what's this dance?

15:57

(don't know go very well)

stack allocation

15:57

of fixed size array

15:57

andcreating a slice from that array

15:58

might be not needed, as the slice "escapes" the stack anyway

15:58

into the heap

looks good, overall

16:09

an additional possible feature would be retrying server info/list packets

current abstraction is only on the packet level

16:10

slowly going upwards

16:10

to connections and whatnot

I must say, Assa's hint about jupyter lab supporting C++ is rather nice

^.^

16:43

https://github.com/jupyter/jupyter/wiki/Jupyter-kernels

16:44

jupyter is great for quick testing, it can even support conda

I installed it with conda

16:57

c++ build process is simply said: rip

Do we have some padding bytes somewhere in the demo header I can utilise to upgrade demos without breaking backwards compatibility?

18:39

Actually could append new chunks after the map

oof

clever approach

I'm going to add sha256 to the demos, would you be interested in a more extensible approach where we can extend the demo format without breaking compatibility in the feature? Or just a version bump and the sha256 after the map?

learath2: aren't there some bytes for the demo version?

there are

Well using the bytes for the demo version would "break" backwards compatibility

added a retrying fetch, maybe there are better approaches to this.

Exported 38 message(s)