From: "Iñaki Baz Castillo" <ibc@...>
Date: 2009-09-03T05:51:11+09:00
Subject: Re: [ANN] clogger 0.0.4 - configurable request logging for Rack

El Miércoles, 2 de Septiembre de 2009, Eric Wong escribió:
> Iñaki Baz Castillo <ibc@aliax.net> wrote:
> > El Miércoles, 2 de Septiembre de 2009, Iñaki Baz Castillo escribió:
> > > El Miércoles, 2 de Septiembre de 2009, Eric Wong escribió:
> > > > > > Clogger is Rack middleware for logging HTTP requests.  The log
> > > > > > format is customizable so you can specify exactly which fields to
> > > > > > log.
> > > > >
> > > > > Is it possible to make Clogger using Logging class?
> > > >
> > > > If you mean Logger, then yes,
> > >
> > > No, I mean Logging:
> > >
> > >    http://logging.rubyforge.org/
> > >
> > > "Logging is a flexible logging library for use in Ruby programs based
> > > on the design of Java‘s log4j library. It features a hierarchical
> > > logging system, custom level names, multiple output destinations per
> > > log event, custom formatting, and more."
> > >
> > > > you can specify anything that responds to
> > > > "<<" as your :logger parameter:
> > > >
> > > >   use Clogger, :logger => Logger.new("/path/to/log")
> > >
> > > How to set the Logger debug level there?
> >
> > For example, Logging has "<<" method, but it's a limited one:
> >
> > ---------------------------------------------
> > log << "message"
> >
> > Log the given message without any formatting and without performing any
> > level checks. The message is logged to all appenders. The message is
> > passed up the logger tree if this logger‘s additivity is true.
> > ---------------------------------------------
> 
> For request logging, the point of using "<<" is to avoid formatting done
> by the Logger.  Both Rack::CommonLogger and Clogger already do their own
> timestamps and other formatting features (Clogger being highly
> configurable)).
> 
> So you really don't get much benefit from using any Logging/Logger
> object over a File object unless you need builtin log rotation[1]
> or you want userspace buffering[2].
> 
> 
> [1] - which doesn't work if you're using multiple processes anyways,
>       logrotate is a better solution.
> 
> [2] - which means you lose data if your application dies unexpectedly
> 

Thanks for the explanation.

-- 
Iñaki Baz Castillo <ibc@aliax.net>