Prior to this change, the ShallowEtagHeaderFilter would use a
ResizableByteArrayOutputStream to internally write data and calculate
the ETag. While that implementation is faster than the regular
ByteArrayOutputStream (since it has a better strategy for growing the
internal buffer), a lot of buffer copying/writing still happens.
This change adds a new FastByteArrayOutputStream implementation that
internally uses a LinkedList<Byte[]> to store the content. So when
writing bytes to that OutputStream implementation, new byte[] are
added to the list when the previous ones are full. This saves most
of the instantiating/copying operations.
Note that new methods were added in DigestUtils to allow usage of
Streams instead of byte[], which is more efficient in our case.
Fixes#653
Issue: SPR-12081
SPR-12483 introduced automatic type conversion support for EnumSet and
EnumMap. However, the corresponding changes in CollectionFactory
contradict the existing contract for the "create approximate" methods
by creating a copy of the supplied set or map, thereby potentially
including elements in the returned collection when the returned
collection should in fact be empty.
This commit addresses this issue by ensuring that the collections
returned by createApproximateCollection() and createApproximateMap()
are always empty.
Furthermore, this commit improves the Javadoc throughout the
CollectionFactory class.
Issue: SPR-12533
This commit introduces superfluous casts in CollectionFactory to
address a bug in Eclipse 4.4.1 that prevents the code from compiling
within Eclipse. Specifically, without these casts Eclipse displays the
following error for use of the EnumSet.copyOf() and EnumSet.noneOf()
methods:
Type mismatch: cannot convert from EnumSet<Enum<Enum<E>>> to
Collection<E>
Prior to this commit, JOptCommandLinePropertySource prevented the
possibility of non-String option arguments. This effectively prevents
the use of JOpt's #ofType support (which allows specifying custom
argument types).
Now, non-String arguments are detected and converted to strings as
necessary. JOpt's #ofType now works as expected. A test has been added
to cover this case.
Prior to this commit, the ResolvableType static cache was holding a lot
of duplicates for simple types. We are using too much metadata to compute
the key when the class has no generic information. so setFoo(String foo)
and setBar(String bar) would result in two entries in the cache because
the TypeProvider is different. On a very simple application 65% of the
entries in the cache were duplicate.
When the type is a Class with no generic information, the ResolvableType
instance is a simple wrapper around it so we might just as well not cache
it at all as the cost of finding it back from the cache is higher than
creating that simple wrapper.
This commit adds an explicit check; if the type is a simple Class we just
return a "resolved" ResolvableType instance for it. On a few test cases,
this reduces the size of the cache by 85%
Issue: SPR-12275
Juergen Hoeller
Sam Brannen
Rossen Stoyanchev
Juergen Hoeller
Craig Andrews
John
Chris Beams
Stephane Nicoll