I'm trying to use file:read from erlang combined with compiled scanf pattern or some regex. There are two ideas:<br>1) combine two binary:split will give double memory walking. I think it is a good idea to parse memory once<br>
2) intermediate creation of binaries for all those floats (12 millions of them in this example) is also evil. Perhaps line parsing should be combined with float extraction in the same manner as decode_packet does.