00:41 < bridge> [teeworlds] std::regex 00:42 < bridge> [teeworlds] yeah regex for tokenization 😮 00:42 < bridge> [teeworlds] https://stackoverflow.com/questions/16749069/c-split-string-by-regex/16752826 00:43 < bridge> [teeworlds] well, if one does not care about performance at all 00:43 < bridge> [teeworlds] use python 😄 00:43 < bridge> [teeworlds] with regex 00:44 < bridge> [teeworlds] no in python I use the tokenizer of sklearn or spacy 00:44 < bridge> [teeworlds] to split string 00:44 < bridge> [teeworlds] bette rthan a whole state machine ._. 00:45 < bridge> [teeworlds] finite state machine 00:45 < bridge> [teeworlds] there are dynamic and static tokenizers 00:45 < bridge> [teeworlds] i can give you literature if you want 11:01 < bridge> [teeworlds] why is there a check whether client is ingame and not quitting in the loop BUT NOT for a single send? this could remove all the checks from where we call SendPackMsg and would be nicer in general 11:01 < bridge> [teeworlds] https://github.com/teeworlds/teeworlds/blob/master/src/engine/server/server.cpp#L519-L527 12:01 < bridge> [teeworlds] you can tell me the isbn 12:02 < bridge> [teeworlds] ISBN 12:03 < bridge> [teeworlds] opening an issue leads to better results, as discussions here don't seem to lead to any change of such small, but imo important refactoring changes. 12:14 < bridge> [teeworlds] Probably because packet.m_clientid has a sane value if SendMsg is called 12:15 < bridge> [teeworlds] Additional checks wouldn't hurt ig 12:18 < bridge> [teeworlds] it doesnt have to have a sane value 12:19 < bridge> [teeworlds] can pass any client id you want 14:36 < bridge> [teeworlds] but it assumes it is sane 14:42 < bridge> [teeworlds] yes