Just pushed jDBI 1.2.3. Much thanks to Patrick and Robert for prodding me make a couple big changes, and for prodding me to not make those changes until I found the right way, respectively (for 1.2.2 (transparently handling different transactional contexts) and 1.2.3).
The biggest change is making externalized SQL pluggable. It was something that had been itching for a while (previously it could only pulled named statements from the classpath, though pretty smartly), and Patrick's need for sticking the SQL in the database prodded me to finally support that explicitely. He's not using it, I don't think, but you'd better bet I will before too long =) Maybe stick them in a JNDI or an LDAP instead, or whatever. It's all good.
Speaking of the classpath approach, what I did on the last new thing I used jdbi for was to take advantage of Java's nice resource loading and bundle all the sql into its own jar, under an sql/ directory, so you'd have:
sql/
find-foo-by-id.sql
find-foo-by-name.sql
...
Map foo = handle.first("sql/find-foo-by-id", Args.with("id", new Long(fooId)));
Map same_foo = handle.first("sql/find-foo-by-name", foo); // gets "name" from map
The hack here is that the named statement is fetched correctly, and if you unzip the sql.jar to tweak the sql, it's in its own dir, not the working dir. This is minor, but I have things that untar (zip|jar) into the working directory =)