From: Dave Thomas <dave@...>
Date: 2007-11-20T07:05:35+09:00
Subject: Re: Change in system() behaviour


On Nov 19, 2007, at 3:59 PM, Wolfgang N�dasi-Donner wrote:

> Dave Thomas schrieb:
>> Shame...  It makes a simple interface a lot uglier...
>
> You wrote about long ago: PickAxe/Kernel#system/Last sentence:
>
> Raises SystemCallError if the command couldn't execute
> (typically Errno::ENOENT).
>
> I never recognized this sentence, because it never happend. Is
> it possible, that this was an earlier behavior of Ruby that was
> changed and is now re-implemented?
>
> On Windows "system" returns "true" only if a normal exit occurs
> ("exit(0)"). If the command exits with another exit value "system"
> returns "false".

Wow That's strange. I have other code in the book that relies on  
false coming back (I know, because that code is now failig with  
ENOENT exceptions)

>
> Why not use a return value "nil" ("command not in list of command")
> if the command cannot be executed, instead of raising an Exception?


I'd definitely second that.  Matz?