katzarsky/websocket

WebSocket codec in C++. Implements RFC6455. Robust one-class parser implementation.

websocket's Issues

opcode == 0 && fin != 0 should be a ERROR frame

if(msg_opcode == 0x0) return (msg_fin)?TEXT_FRAME:INCOMPLETE_TEXT_FRAME; // continuation frame ?

payload_length in getFrame

In my project, payload_length went wrong. I modified codes like this ans it work:

if(length_field <= 125) {
	payload_length = length_field;
}
else if(length_field == 126) { //msglen is 16bit!
	//payload_length = in_buffer[2] + (in_buffer[3]<<8);
	payload_length = ((in_buffer[2]) << 8) | in_buffer[3];
	pos += 2;
}
else if(length_field == 127) { //msglen is 64bit!
	//payload_length = in_buffer[2] + (in_buffer[3]<<8);
	payload_length = ((in_buffer[2]) << 56) | ((in_buffer[3]) << 48) | ((in_buffer[4]) << 40) | ((in_buffer[5]) << 32) | ((in_buffer[6]) << 24) | ((in_buffer[7]) << 16) | ((in_buffer[8]) << 8) | in_buffer[9]; 
	pos += 8;
}

'const char *' or 'string' instead of 'unsigned char *' for input params

This is more of a question than an issue.

I'd like to know what is the reasoning behind taking 'unsigned char *' for input buffer. Usually library user will hold a string, and it is more intuitive to get 'const char *' from string. From code it seems the input is constructed into a string. Then why not take string itself as input.

Thanks.

implement makeFrame totally when msg_length > 65535

https://github.com/katzarsky/WebSocket/blob/master/WebSocket%2FWebSocket.cpp#L178.
Why not implement ? Because it happens barely?