From: Dave Thomas <Dave@...>
Date: 2002-05-07T09:14:35+09:00
Subject: Re: require and case

matz@ruby-lang.org (Yukihiro Matsumoto) writes:

> Hi,
> 
> In message "Re: require and case"
>     on 02/05/07, Dave Thomas <Dave@PragmaticProgrammer.com> writes:
> 
> |I'm becoming convinced by this thread that rb_require should normalize
> |all paths before executing, and should store those normalized paths in
> |$". Thus on a Windows box the names should be converted to lower case
> |and expand_path called. On a Unix both, the case should be left
> |unchanged but expand_path is called. On other boxes, other rules will
> |apply.
> 
> I feel other way around.  Requiring "feature names" should appear same
> on all platforms, i.e. (basically) lowercase words, joined with
> slashes.  This convention also enhances script portability.

Absolutely - I say

   require 'dbi'

But under the covers, Ruby expands this to the canonical name, and
stores that canonical name in $". That way, we don't get duplicates
for

    require 'dbi'
    require 'dbi.rb'

etc

I agree - the feature name stays the same. But when rb_require uses
the relative path to remember what files have been loaded, it's really
a bug, as there are an infinite number of relative paths that all
name the same file. If the contract for 'require' is that it only
loads a given feature once, then not using a canonical name in $"
breaks that contract.


Cheers



Dave