14:04 < bridge> <jxsl13> I got a question in regard to the huffman implementation in the Teeworlds code:
14:04 < bridge> <jxsl13> https://github.com/teeworlds/teeworlds/blob/26d24ec061d44e6084b2d77a9b8a0a48e354eba6/src/engine/shared/huffman.cpp#LL75C1-L76C1
14:04 < bridge> <jxsl13> HUFFMAN_MAX_SYMBOLS being 257
14:04 < bridge> <jxsl13> and m_Symbol being of type unsigned char 
14:04 < bridge> <jxsl13> The loop reaches the value 256 which is out of bounds of what an unsigned char can represent. Is that behavior correct and intended?
14:07 < bridge> <jxsl13> it basically overflows.
15:37 < bridge> <chillerdragon> Are you building a client in go from scratch?
15:38 < bridge> <jxsl13> somewhat yes
15:39 < bridge> <chillerdragon> Cool
15:39 < bridge> <chillerdragon> Never roll your own Huffman it’s such a time killer
15:40 < bridge> <chillerdragon> For my ruby client I just took the C++ code from teeworlds and added some ruby bindings
15:40 < bridge> <jxsl13> I don't see any huffman library in the Go space that I'm capable of using D:
15:40 < bridge> <chillerdragon> Just wrap the C++ code or tell Heinrich to publish his rust crate for go
15:40 < bridge> <jxsl13> C bindings bad
15:41 < bridge> <jxsl13> really bad
15:41 < bridge> <chillerdragon> Why
15:42 < bridge> <jxsl13> in the Go space using C bindings comes most of the time with lots of problems so a pure Go implementation is preferred imo.
15:42 < ChillerDragon> i see
15:42 < bridge> <jxsl13> and Go - C data transfer has an overhead
15:43 < ChillerDragon> sad
15:44 < ChillerDragon> @jxsl13 then this might be a good read https://github.com/heinrich5991/libtw2/blob/master/doc/huffman.md
15:44 < bridge> <jxsl13> maybe I should watch your linked video 😄 about the algorithm
15:45 < bridge> <jxsl13> but porting the code seems pretty straight forward
15:45 < ChillerDragon> the video just helps with the bigger picture
15:45 < bridge> <jxsl13> except for not having pointer arithmetic
15:45 < ChillerDragon> but i never did dive into the technical details because i was too lazy
15:45 < bridge> <jxsl13> same
15:46 < ChillerDragon> python also has no pointers
15:46 < bridge> <jxsl13> I do have pointers
15:46 < ChillerDragon> https://github.com/edg-l/TeeAI/blob/master/engine/huffman.py
15:46 < ChillerDragon> this has known bugs tho ._.
15:47 < bridge> <jxsl13> and if I used the unsafe standard library, I'd also have a way to convert pointers to integers but meh.
15:51 < bridge> <jxsl13> I can only imagine the pain of implementing that in a dynamically typed language D: