The Diff File
Despite the pains taken to port Ruby as completely as possible, there are a few important differences with JRuby: database access, native-extension support and anemic command line performance. The database access differences seemed like more pluses than minuses. Although JRuby does not yet support every database as does Ruby, the pure Ruby MySQL and postgreSQL drivers run fine. Other databases are at various stages of support. Oracle support should be in Continuous Integration shortly after the 1.0 launch.
Alternatively, JRuby developers can use JDBC for access. The ActiveRecord-JDBC gem is available as a Ruby extra. Charles noted that "support for databases is a little better than the C implementation at this point, because it's very easy for us to bring up new databases with a common API like the JDBC." Another plus is having JNDI for connection pooling, which lets you avoid dedicating the resources needed for one connection per process. On the down side, JDBC doesn't have any schema management API, similar to Rails migrations.
The second caveat was the lack of support for C-based extensions. A few have been ported to Java, such as Mongrel and Hpricot. Charles and Thomas gave a hat tip to fellow JRuby project member Ona Bini for his porting efforts, which include OpenSSL. They offered four workarounds for native extensions: Don't use them; use a Java equivalent; port the library yourself; or the easiest route--just write Ruby code to wrap a Java library.
The third difference was command line performance, which they cautioned is not stellar. Thomas showed some Fibonacci benchmarks that revealed how Java is particularly slow at startup. But once it optimizes performance, it matches C-based Ruby, and in many cases, JRuby is exceeding Ruby. Thomas mentioned another project in the works to pre-prime a server or JRuby daemon in the background to mitigate the startup lag.