Unfortunately, it doesn't handle Unicode files correctly (you get a "Failed to parse row" if your data contains characters over 0x7F). RT #32223 suggests this may be due to not passing the binary flag correctly to Text::CSV_XS.
Sadly, this module hasn't been updated since 2006.
2012 update: The module did receive an update in Feb 2012, but the Unicode problem above still hasn't been fixed.
I tried it on a 70,000 line CSV file and it works great. Fast and easy to use. My only complaint was that I still had to load Text::CSV_XS directly to call the "combine" method, so I could spit back out a CSV row in a munged format. Seems more like a "1.0" release than an "0.02".