On 19/Dec/2009 18:57, Nathan Beyer wrote:
> RE: https://issues.apache.org/jira/browse/HARMONY-6404
>
> I posted up two proposed patches for String, please comment on which
> is preferable. One is the change previously mentioned, the other is a
> slightly bigger reorganization.
Neither. There is nothing wrong with the original code.
Any JVM or compiler that would actually move a load up above a
conditional store to the same variable would be seriously broken all round.
Regards,
Tim
> On Sat, Dec 19, 2009 at 4:04 AM, Egor Pasko <egor.pa...@gmail.com> wrote:
>> On the 0x68A day of Apache Harmony Tim Ellison wrote:
>>> On 12/Dec/2009 10:54, Egor Pasko wrote:
>>>> On the 0x685 day of Apache Harmony Nathan Beyer wrote:
>>> <snip>
>>>>> In any case, it does seem a pinch more efficient to only do one read
>>>>> of hashCode ... switch up the code to be something like this.
>>>>>
>>>>> public int hashCode() {
>>>>> final int hash = hashCode;
>>>>> if (hash == 0) {
>>>>> if (count == 0) {
>>>>> return 0;
>>>>> }
>>>>> for (int i = offset; i < count + offset; i++) {
>>>>> hash = value[i] + ((hash << 5) - hash);
>>>>> }
>>>>> hashCode = hash;
>>>> one more 'return hash' here, please :)
>>> Why?
>> argh, what was I thinking about? the original Nathan's code is great.
>>
>> --
>> Egor Pasko
>>
>>
>
