From: Robert Klemme <bob.news@...>
Date: 2004-03-11T01:39:37+09:00
Subject: Re: require is excluding too much


"Simon Strandgaard" <neoneye@adslhome.dk> schrieb im Newsbeitrag
news:pan.2004.03.10.14.13.43.358057@adslhome.dk...
> On Wed, 10 Mar 2004 15:23:08 +0100, shasckaw wrote:
> > Simon Strandgaard wrote:
> >> Is this a bug?
> >> Shouldn't 'require' load both modules in this case?
> >> If not, what is the motivation for doing so?
> >>
> >> Just curious :-)
> >>
> >> --
> >> Simon Strandgaard
> > [snip]
> >> require 'mod'
> >> $:.unshift('dir')
> >> require 'mod'
> >> server> ruby a.rb
> >> hello1
> >> server> ls
> >> a.rb    dir/    mod.rb
> >> server>
> >>
> >> I would have expected it to output both 'hello1' and 'hello2'.
> >
> > Hello Simon!
> >
> > When requiring a module twice, it is probable that ruby just use its
> > name to check if it has to reload it or not, forgetting to also check
if

I don't think that "forget" is the appropriate term.  One of the files
named 'mod.rb' wins and gets loaded.  If location would matter, all
mod.rb's would have to be loaded, don't they?

> > it is in a different directory or not.
>
> It seems so.

Moreover IMHO this is the only approach to go.  Doing "require 'foo'"
announces that a "foo.rb" (or shared lib) has to be loaded from any
directory in $:.  I guess the first occurrence wins.  If there is another
foo.rb that is silently ignored.  If at a later point in time either $:
changes or there is a new foo.rb in an earlier location in $:

Additionally it's not usual that $: changes or that there are more files
with the same relative path in an installation's lib directories.  *If*
you create a situation like you did, it is IMHO more appropriate to use
"load" since you know where to get the file from anyway.

In short, I don't regard this an error.  Sorry if I haven't been clear
enough. (Out of words error)

Kind regards

    robert