Java Magazine - May/June 2012

//java architect /

the Unicode code points. Although this is fine in many cases, it does not necessarily result in the same ordering as you would expect in a phone book or dictionary. For example, an uppercase letter is not sorted close to its corresponding lowercase variant, and punctuation characters are considered to be significant.

In JDK 7, Java DB can also order character strings by using the collation rules defined for specific locales, so-called language-based ordering. These rules typically resemble more closely what you would find in a phone book.

Language-based ordering has to be enabled using connection attributes when the database is created. To create a database that obeys the collation rules of the en_US locale, you can use the following URL: language-based ordering (on the right). Notice that all the names that start with capital letters are sorted before the names that start with lowercase letters in the default ordering, whereas they are interleaved in between the other words, in alphabetical order, with the language-based ordering.

COMMUNITY

jdbc:derby:memory:MyDB;create=true More or less, everything you can do with an ordinary database can be done with an in-memory database, including taking a backup of it and restoring it. This allows you to dump the database to disk before you shut down your application, reload it into memory the next time you start the application, or even turn it into an ordinary on-disk database.

JAVA IN ACTION

GIVE BACK!

ADOPT A JSR

Find your JSR here

jdbc:derby:db;create=true; territory=en_US; collation=TERRITORY_BASED

ABOUT US

Table 1 shows how a set of strings would be ordered by Java DB with default ordering (on the left) and with

Table 1

DEFAULT ORDERING

VANDERBILT WOOLF DA VINCI VAN BEE THOVEN VAN GOGH VON GOE THE

LANGUAGE-BASED ORDERING

DA VINCI VAN BEE THOVEN VANDERBILT VAN GOGH VON GOETHE WOOLF

Performance Enhancements Many performance enhancements have gone into Java DB between JDK 6 and JDK 7. The code paths of many common operations have been optimized so that fewer CPU cycles are spent when performing queries. Also, some performance-critical modules have been rewritten to use collection classes from the java.util.concurrent package,