From: Hans Fugal <hans@...>
Date: 2004-08-05T04:46:29+09:00
Subject: Re: YAML.load(ARGF)

I agree it doesn't always make sense, maybe not even often, but 
sometimes it does. There's nothing that says a yaml document has to 
reside in one file, and there's certainly good reason to take it in on 
stdin.

Robert Klemme wrote:
> "Austin Ziegler" <halostatue@gmail.com> schrieb im Newsbeitrag
> news:9e7db91104080410463cea4c22@mail.gmail.com...
> 
>>On Thu, 5 Aug 2004 02:36:29 +0900, Hans Fugal <hans@fugal.net> wrote:
>>
>>>Consider these two short programs:
>>>        ARGF.read
>>>and
>>>        require 'yaml'
>>>        YAML.load(ARGF)
>>>
>>>Then call them like so:
>>>
>>>        $ ruby foo.rb foo
>>>
>>>The first program prints the contents of foo and exits. The second
>>>program waits for ^D from the terminal, and then tries to parse the YAML
>>>and go on with life.
>>>
>>>I want ARGF to behave like it does in the first example, in the second
>>>example. Of course I could do YAML.load(ARGF.read) but that doesn't
>>>satisfy my curiousity. :-)
>>
>>IMO, YAML.load *should* do this with ARGF because the following works:
>>
>>  f = File.open("foo", "rb")
>>  YAML.load(f)
>>  f.close
> 
> 
> Not so fast: ARGF is quite special.  It's especially *no* an IO.  It just
> happens to implement some methods of IO.  IMHO it's not reasonable to invoke
> YAML.load(ARGF) because ARGF might draw its data from any number of files -
> including stdin.  IMHO it's not very practical to distribute YAML data
> across multiple files.  The case is differnt for grep like tools that can
> reasonable operate on a multitude of files.
> 
> Just my 0.02EUR...
> 
> Kind regards
> 
>     robert
>