RFR: 8250646: hdiutil detach fix JDK-8245311 still fails sometimes

[Andy Herrick]

looks good

/Andy

On 8/4/2020 1:05 AM, alexander.matveev at oracle.com wrote:
> Hi Alexey,
>
> http://cr.openjdk.java.net/~almatvee/8250646/webrev.02/
>
> I removed stop method. Also, if we called abort from MacDMGBundler 
> exception will be ignored.
>
> Thanks,
> Alexander
>
> On 8/3/20 10:10 AM, Alexey Semenyuk wrote:
>>
>>
>> On 7/31/2020 7:36 PM, alexander.matveev at oracle.com wrote:
>>> Hi Alexey,
>>>
>>> Without modifying some logic in Executor, yes stop is needed. After 
>>> retryExecutor.stop(); is executed in MacDmgBundler we will still run 
>>> command (same with abort), and this command will fail and exception 
>>> will be thrown. Stop() is basically like abort, but exception is 
>>> ignored.
>> Right, but can't we just have one method that will abort the loop? 
>> Why not test for the value of `aborted` field not only in exception 
>> handler, but also before running executor? I mean instead of
>> ---
>> if (stop) {
>>     break;
>> }
>> ---
>> have
>> ---
>> if (aborted) {
>>     break;
>> }
>> ---
>>
>> This would eliminate the need to introduce stop() method. What do you 
>> think?
>>
>> - Alexey
>>>
>>> Thanks,
>>> Alexander
>>>
>>> On 7/31/20 9:27 AM, Alexey Semenyuk wrote:
>>>> Alexander,
>>>>
>>>> Why introduce RetryExecutor.stop() method? Isn't 
>>>> RetryExecutor.abort() sufficient?
>>>>
>>>> - Alexey
>>>>
>>>> On 7/30/2020 8:10 PM, alexander.matveev at oracle.com wrote:
>>>>> Please review the jpackage fix for bug [1] at [2].
>>>>>
>>>>> - DMG image can get detach even if "hdiutil detach" return 
>>>>> resource busy error. Fixed by stopping trying detach if DMG 
>>>>> already got detach.
>>>>>
>>>>> [1] https://bugs.openjdk.java.net/browse/JDK-8250646
>>>>> [2] http://cr.openjdk.java.net/~almatvee/8250646/webrev.00/
>>>>>
>>>>> Thanks,
>>>>> Alexander
>>>>
>>>
>>
>