RFR: [6904367]: (coll) IdentityHashMap is resized before exceeding the expected maximum size (original) (raw)

Martin Buchholz martinrb at google.com
Tue Jul 8 07:27:09 UTC 2014

Ah, yes, now I see.

My latest attempt at this method looks like:

private static int capacity(int expectedMaxSize) {
    // assert expectedMaxSize >= 0;
    return
        (expectedMaxSize > MAXIMUM_CAPACITY / 3) ? MAXIMUM_CAPACITY :
        (expectedMaxSize <= 2 * MINIMUM_CAPACITY / 3) ?

MINIMUM_CAPACITY : Integer.highestOneBit(expectedMaxSize + (expectedMaxSize << 1)); }

On Mon, Jul 7, 2014 at 6:12 PM, Ivan Gerasimov <ivan.gerasimov at oracle.com> wrote:

On 08.07.2014 4:47, Martin Buchholz wrote:

I think this code has an off-by-factor-of-2 bug.

+ if (expectedMaxSize > MAXIMUMCAPACITY / 3) + return MAXIMUMCAPACITY;

No, even though it looks like a bug. (MAXIMUMCAPACITY / 3) * (3 / 2) == MAXIMUMCAPACITY / 2. if expected size > MAXIMUMCAPACITY / 3, then the minimum capacity must be > MAXIMUMCAPACITY / 2 then the minimum capacity == MAXIMUMCAPACITY.

More information about the core-libs-dev mailing list