Patrick Hurley
12/21/2005 3:46:00 PM
On 12/21/05, Explosiv0SX <explosiv0SX@datafront.com> wrote:
> Hi -- long time lurker, first time poster.
>
> While searching for Ruby app configuration best practices, I came
> across the config.rb file for WEBrick. In this file, a few hashes are
> created as follows:
>
> General = {
> :ServerName => Utils::getservername,
> :BindAddress => nil, # "0.0.0.0" or "::" or nil
> :Port => nil, # users MUST specifiy this!!
> :MaxClients => 100, # maximum number of the concurrent
> connections
> :ServerType => nil, # default: WEBrick::SimpleServer
> :Logger => nil, # default: WEBrick::Log.new
> :ServerSoftware => "WEBrick/#{WEBrick::VERSION} " +
> "(Ruby/#{RUBY_VERSION}/#{RUBY_RELEASE_DATE})",
> :TempDir => ENV['TMPDIR']||ENV['TMP']||ENV['TEMP']||'/
> tmp',
> :DoNotListen => false,
> :StartCallback => nil,
> :StopCallback => nil,
> :AcceptCallback => nil,
> }
>
> I'm curious as to why the author used symbols here as the hash keys.
> Is it for performance reasons? My first inclination was to use
> strings in the same way, but after seeing this, I started to wonder
> how these symbols are represented internally compared to strings.
>
> Can anyone shed any light on why symbols are used here?
>
> Rick
>
>
Largely a matter of style - plus you have to type one less letter :-)
Symbols will be slightly more efficent and use a little less memory,
as every reference to an existing symbol will become a reference to
the existing symbol in the symbol table. Of course the flip side is
there is no garbage collection of symbols, so they are there to stay,
where string objects can be collected and discarded when they are no
longer needed. This is only a real issue in long running processes
that might create many dynamic symbols.
pth