Re: namespace instance methods?

John Doe wrote in post #1072359:
> hostname = WebProperty.new 'www.yahoo.com'
> hostname.DNS.lookup
> hostname.HTTP.lookup

HTTP doesn't have a "lookup" operation, so do you mean something like

hostname.HTTP.get('/index.html') ?

This is the typical OO conundrum. For a method which takes two object 
arguments, which class does the logic belong to?

If there is no polymorphism involved then it doesn't really matter, and 
in this case arguably it doesn't belong to either:

HTTP.get(hostname, '/index.html')

Or maybe the logic belongs in a URI object.

One advantage of your approach is that hostname.HTTP could keep track of 
a persistent HTTP connection, so that a series of hostname.HTTP.get() 
operations would re-use the same connection.

My natural inclination would be instead to have a "HTTP fetcher" class: 
this is not too dissimilar, but it makes it more obvious how to set 
options which are specific to the HTTP connection, rather than to the 
thing you're connecting to. (Timeouts, for example).

http1 = HTTP.new :hostname=>'www.google.com', 
:http_proxy=>"gw1.example.com"
http2 = HTTP.new :hostname=>'www.yahoo.com', 
:http_proxy=>"gw2.example.com"
yahoo = http1.get '/index.html'
google = http2.get '/index.html'

This approach is also arguably more pluggable.

But this is entirely subjective: hostname.HTTP.get(path) will obviously 
work, and if it makes sense for your particular application, then by all 
means use it.

-- 
Posted via http://www.ruby-forum.com/.