[erlang-questions] log_roller log parsing / binary match speeds

Antonio SJ Musumeci <>
Wed Aug 15 16:29:47 CEST 2012

I've been using log_roller for a project and want to be able to dump
the binary files as erlang terms in plain text.

Looking through the log_roller code I found the below which works but
seems rather slow. My question is: is this expected performance? Is
there anything I can do to speed this up?

loop(<<16#FF:8, 16#FF:8, 16#FF:8, 16#FF:8, LogSize:16/integer,
           Log:LogSize/binary, 16#EE:8, 16#EE:8, 16#EE:8, 16#EE:8,
Tail/binary>>) ->
  % process log
loop(<<>>) ->

On a 2.3GHz Xeon this is taking 25 seconds for a 100MB file. The
binary is sourced by calling file:read_file/1. It's being run as an


More information about the erlang-questions mailing list