From: Robert Klemme <shortcutter@...>
Date: 2007-03-05T07:30:09+09:00
Subject: Re: How to download file from web site

On 04.03.2007 20:50, ara.t.howard@noaa.gov wrote:
> On Mon, 5 Mar 2007, Joel VanderWerf wrote:
> 
>> Bil Kleb wrote:
>>> ara.t.howard@noaa.gov wrote:
>>>>
>>>> require 'open-uri'
>>>> uri = 
>>>> 'http://rubyforge.org/frs/download.php/17670/RMagick-1.15.3.tar.gz'
>>>>
>>>> open(uri) do |fin|
>>>>   open(File.basename(uri), 'w') do |fout|
>>>>     while(buf = fin.read(8192))
>>>>       fout.write buf
>>>>     end
>>>>   end
>>>> end
>>>
>>> He asks naively...
>>>
>>>  Why is this so complicated, e.g., the 8192 buffered write, etc.?
>>
>> It does seem too low-level, doesn't it? Especially in the same snippet 
>> with open(uri), which abstracts so much.
>>
>> Didn't the rio library[1] have a way to abstract this operation?
>>
>> This is from the docs:
>>
>>> Copy a web-page to a file  rio('http://rubydoc.org/') > rio('afile')
>>
>> I suppose it's using the same buffering technique underneath that 
>> abstraction.
>>
>> [1] http://rio.rubyforge.org/  (never used it myself)
> 
> reminds me of an RCR i've had in mind:
> 
>   io.relay another_io, bufsize
> 
> seems like a nice one huh?

Yes, although I'm not sure whether "relay" is immediately clear to 
everyone - especially non native speakers.  I briefly thought of 
"copy_to" but that is not correct in every case, i.e. not always it is 
actually a copy.  What about "forward_to"?  Hm...

Kind regards

	robert