Thanks for the responses!  Obviously, removing the Apple installation of 
ruby and replacing it with simlinks works.

I ended up installing mongrel and using that to solve the problem though 
if only because I don't like blowing away files that Apple might simply 
replace in future updates.

I did want to point out that the other recommendation I received does 
not work.  It was actually what I did when I originally installed ruby 
manually, so perhaps it is useful to post a sample session that shows 
the problem more clearly:

/* my local install ruby is first in my path */
bash $ which ruby
/usr/local/bin/ruby

/* note the versions of my manual install and of the default */
bash $ /usr/local/bin/ruby -v
ruby 1.8.6 (2007-03-13 patchlevel 0) [powerpc-darwin8.10.0]
bash $ /usr/bin/ruby -v
ruby 1.8.2 (2004-12-25) [powerpc-darwin8.0]

/* see which one WEBrick is picking? */
bash $ ./script/server
=> Booting WEBrick...
=> Rails application started on http://0.0.0.0:3000
=> Ctrl-C to shutdown server; call with --help for options
[2007-09-20 19:37:03] INFO  WEBrick 1.3.1
[2007-09-20 19:37:03] INFO  ruby 1.8.2 (2004-12-25) [powerpc-darwin8.0]
[2007-09-20 19:37:03] INFO  WEBrick::HTTPServer#start: pid=795 port=3000


For some reason, even though I have /usr/local/bin listed first in my 
path (as Hivelogic suggests in their manual install instructions) the 
WEBrick server still uses the Apple default ruby installation.  It could 
be a weirdness with the way Apple sets up their default $PATH variable 
or it could be that WEBrick simply ignores local environment variables.

I can't determine which is the case because I really don't know much 
about WEBrick and I can't find where OS X sets up its default 
environment variables.  (Though, I would like to know if anyone knows 
off the top of their head...)

If it is a problem with WEBrick, I would be willing to file a bug or 
whatever to help the developers.  Obviously, I would like to confirm it 
before wasting their time.  I guess it's kind of a simple bug, but this 
is the sort of thing that looks like it would have high visibility with 
new rails developers on Mac OS X.  Maybe it only shows up as a problem 
until Apple releases their next OS, but there could be folks stuck on OS 
X 10.4.x for a while afterwards that might run into this.  It could also 
manifest later in some form.

It is perhaps useful to point out that because I've also installed gem 
in the following location:


bash $ which gem
/usr/local/bin/gem


As a result, anything I install via gem is ignored by WEBrick.  For 
example, trying to start up a rails app with an sqlite3 database (which 
uses the sqlite3 plugin I've installed via gem) provides an error 
message:


bash $ gem list --local sqlite3

*** LOCAL GEMS ***

sqlite3-ruby (1.2.1)
    SQLite3/Ruby is a module to allow Ruby scripts to interface with a
    SQLite3 database.
bash $ ./script/server
=> Booting WEBrick...
=> Rails application started on http://0.0.0.0:3000
=> Ctrl-C to shutdown server; call with --help for options
[2007-09-20 19:44:13] INFO  WEBrick 1.3.1
[2007-09-20 19:44:13] INFO  ruby 1.8.2 (2004-12-25) [powerpc-darwin8.0]
[2007-09-20 19:44:13] INFO  WEBrick::HTTPServer#start: pid=799 port=3000

===+ WEB BROWSER ERROR MESSAGE +===
No such file to load -- sqlite3
/usr/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:27:in 
`gem_original_require'
/usr/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:27:in `require'
===+ TRUNCATED FOR CLARITY +===


Anyhow, because mongrel is installed via a gem, it seems to recognize 
the other gems installed on the system, so I'm going to use that for 
now.  However, it does seem that the only fully functional recourse in 
some cases appears to be actually blowing away the Apple install of 
ruby.

Thanks again for the help!
-- 
Posted via http://www.ruby-forum.com/.