RFR 8081027: Create a common test to check adequacy of initial size of static HashMap/ArrayList fields

Martin Buchholz martinrb at google.com
Wed May 27 14:30:36 UTC 2015


Your methods like ofArrayList declare that they
throw ReflectiveOperationException, but also have a try/catch to prevent
that from ever happening.

The assertions e.g.
OptimalCapacity.ofArrayList(XClass.class, "theList", N)
could look more like assertions, e.g.
OptimalCapacity.assertOptimallySized(Class<?> clazz, String fieldName,
            int initialCapacity)

getStorage returns the actual internal array, but you only ever need the
length.  Maybe I would have a single method

static int internalArraySize(Object x) { ... }

"enteties" is misspelled.

For ArrayLists, I would have been happy enough just testing that we have
100% utilization, i.e. size of array is the same as size of the List,
without checking the initial capacity.

On Mon, May 25, 2015 at 4:07 PM, Ivan Gerasimov <ivan.gerasimov at oracle.com>

> Hello!
> This is in some way continuation of JDK-8080535 (Expected size of
> Character.UnicodeBlock.map is not optimal).
> A new utility class OptimalCapacity was added to jdk.testlibrary package.
> The test NonOptimalMapSize.java that had been added with JDK-8080535, was
> rewritten with use of this new class.
> A couple more tests were added, which utilize methods of OptimalCapacity
> for checking sizes of ArrayList and IdentityHashMap static variables.
> Optimization of initial sizes of two more variables saves us one
> reallocation during java start-time and a few more bytes of memory.
> Would you please help review this fix?
> BUGURL: https://bugs.openjdk.java.net/browse/JDK-8081027
> WEBREV: http://cr.openjdk.java.net/~igerasim/8081027/00/webrev/
> Sincerely yours,
> Ivan

More information about the core-libs-dev mailing list