system classes (not 100% done, if you use a ByteEnum in a key right now it
doesn't do the right thing). Also moved SimpleCacheKey into impl because it's
an implementation detail, not meant to be used by external parties.
us so that if we use multiple PersistenceContexts in the same VM they don't all
point to the same cache buckets which causes badness. At some point we'll
probably have to make it easier to customize these configurations, but this
should work for now.
indicative to the caller that they're not just getting another array list back
but rather a magical lazily converted list but is more useful than Iterable.
memory so that if we turn around and recreate a PersistenceContext with a brand
new classloader and start loading things from the cache, we don't have
freakoutery because there's a record in memory that was resolved with the old
classloader. Plus it allows those old classloaders to get collected.
all of our fromByte() methods already do which is iterate over the Enum and
find the one with the right byte. If we care about performance (which we don't
because this is trivial since all of these enums are small), we could even
cache a mapping from byte to enum in the field marshaller. No need to foist
this reponsibility on the ByteEnum implementor.
that return a collection return an XList which extends List and adds the
method:
public <R> Iterable<R> map (Function<T, R> mapper);
which allows you to do things the following:
return findAll(CategoryRecord.class,
new Where(CategoryRecord.PARENT_ID.eq(parentId))).
map(CategoryRecord.TO_CATEGORY);
which is sufficiently awesome that I think it merits taking these liberties.
Plus it opens the door to adding other transformations that one might want to
do with a collection just loaded from the database.
java.sql.Date and java.util.Date and we support custom getters and setters to
do fancier aggregate conversion (multiple persistent fields into a single
runtime field and/or creation of partially initialized runtime objects from the
available persistent data).
foolish to complicate a bunch of client code. Either we'll do some magical
optimization on the Depot side and not even send queries to the server that we
know can't match anything, or we'll just suck it up and occasionally send a
query to the database that we know will match no rows. C'est la vie.
creating common operators and expressions that don't fit nicely into the fluent
style. For example: Ops.and(expr1, expr2) is preferable to expr1.and(expr2)
and Ops.not(expr) is definitely nicer than expr.not() (which we don't even
provide).
methods for expressions. Now you can do FooRecord.BAR.eq(5) instead of new
Equals(FootRecord.BAR, 5), which also saves you the Equals import. Added
methods for operators, some are probably missing and some don't make sense.
I'll fill this out more as I discover where else this style is useful.
tidier, it's a shame that ColumnExp is named as it is because really it's a
FieldExp as it represents the name of the Java object field, not the actual
name of the column which may be something different (if you use
@Column(name="foo")). Maybe next time.
Michael Bayne