From: Rick Bradley <rick@...>
Date: 2002-05-07T14:09:20+09:00
Subject: Re: require and case

* Rich Kilmer (rich@infoether.com) [020507 00:04]:
> I think that since the SHA-1 digest extension is included with ruby,
> transitioning this code into the core to distinguish loaded files is
> appropriate.  It would internally store a hash of every loaded file at
> 20 bytes apiece.  Like Jim said, this would prevent the same file from
> being loaded from two different directories.  I don't think this would
> add significant overhead (but I have not benchmarked it ;-)

What is the problem being solved again?


Is solving that problem worth turning an fstat()+fopen()+read()+fclose()
into twice that plus SHA computation over every included file, plus a
hash comparison and the overhead of maintaining an extra data structure
(minus the ability to optimize with a cache)?

Couple that with a potential race condition and I don't see what
is gained.

Rick
-- 
 http://www.rickbradley.com    MUPRN: 828    (79 F)
                       |  the mythical wall
   random email haiku  |  of hoodoo). Ah. I guess the
                       |  hoodoo will come down.