<Swing Dev> [12] JDK-8213516: jck test api/javax_accessibility/AccessibleState/fields.html fails intermittent

Mon Jan 21 18:10:47 UTC 2019

Hi Phil\Sergey, We need to implement the hashCode() and equals() method for the hashtable keys. Here is the link to it for that information: https://docs.oracle.com/javase/10/docs/api/java/util/Hashtable.html

Here is the text from that link: "To successfully store and retrieve objects from a hashtable, the objects used as keys must implement the hashCode method and the equals method."

HashTable class uses these methods for key types used in the hash table, hence they need to be implemented as part of the key types. The JCK test already tests this, so Sergey do you want me to write another open test for this problem?

Thanks and regards,
Shashi

-----Original Message-----
From: Phil Race 
Sent: Wednesday, January 16, 2019 4:35 AM
To: Sergey Bylokhov <sergey.bylokhov at oracle.com>; Shashidhara Veerabhadraiah <shashidhara.veerabhadraiah at oracle.com>
Cc: swing-dev at openjdk.java.net; Dmitry Bessonov <dmitry.bessonov at oracle.com>; Krishna Addepalli <krishna.addepalli at oracle.com>
Subject: Re: <Swing Dev> [12] JDK-8213516: jck test api/javax_accessibility/AccessibleState/fields.html fails intermittent

We definitely don't want a new public class here !
The fix seems very heavyweight for the problem, and I'm sure there must be simple ways of creating a compound key like

resourceBundleName + locale.toString() ... or similar.

-phil.

On 1/14/2019 3:21 PM, Sergey Bylokhov wrote:
> Hi, Shashi.
>
> I am not sure I am understand the purpose of AccessibleBundleKey in 
> your fix, you created a class which can be used as a key, but 
> eventually you do not use it, except the calling of hashCode(). Note 
> that the hashCode which you use as a key for a table may be the same 
> for different objects.
>
> It will be useful to have a test case for the fix, please.
>
>
> On 14/01/2019 02:49, Shashidhara Veerabhadraiah wrote:
>> Hi All, Please find the new Webrev implementing a different key for 
>> the accessibility resource tables.
>>
>> http://cr.openjdk.java.net/~sveerabhadra/8213516/webrev.02/
>>
>> Now with this implementation all the tests passes consistently.
>>
>> Thanks and regards,
>> Shashi
>>
>> -----Original Message-----
>> From: Sergey Bylokhov
>> Sent: Friday, December 21, 2018 12:42 PM
>> To: Shashidhara Veerabhadraiah
>> <shashidhara.veerabhadraiah at oracle.com>; Philip Race 
>> <philip.race at oracle.com>
>> Cc: swing-dev at openjdk.java.net; Dmitry Bessonov 
>> <dmitry.bessonov at oracle.com>; Krishna Addepalli 
>> <krishna.addepalli at oracle.com>
>> Subject: Re: <Swing Dev> [12] JDK-8213516: jck test 
>> api/javax_accessibility/AccessibleState/fields.html fails 
>> intermittent
>>
>> Hi, Shashi.
>> Yes we need a fix for this, see my comments here:
>> http://mail.openjdk.java.net/pipermail/swing-dev/2018-December/009248
>> .html
>>
>>
>> On 20/12/2018 21:57, Shashidhara Veerabhadraiah wrote:
>>> Hi Phil, In that case I have put up the fix which fixes the problem 
>>> as I was unable to reproduce this problem after that fix. So it does 
>>> work for this situation. Can you please review it?
>>>
>>> It adds the same missing element to the list which is being loaded 
>>> to the cache and the same is used. Another solution is to force the 
>>> resource bundle loading but that may be a bigger change and is too 
>>> late to do that. The fix that I proposed is small and should not 
>>> cause any issues at this late moment.
>>>
>>> Thanks and regards,
>>> Shashi
>>>
>>> -----Original Message-----
>>> From: Philip Race
>>> Sent: Friday, December 21, 2018 10:48 AM
>>> To: Shashidhara Veerabhadraiah 
>>> <shashidhara.veerabhadraiah at oracle.com>
>>> Cc: Sergey Bylokhov <sergey.bylokhov at oracle.com>; 
>>> swing-dev at openjdk.java.net; Dmitry Bessonov 
>>> <dmitry.bessonov at oracle.com>; Krishna Addepalli 
>>> <krishna.addepalli at oracle.com>
>>> Subject: Re: <Swing Dev> [12] JDK-8213516: jck test 
>>> api/javax_accessibility/AccessibleState/fields.html fails 
>>> intermittent
>>>
>>> No, we can't remove it in 13 unless we deprecate for removal in 12 
>>> and we are too late for that and in any case 6 months is very short.
>>> This is not important enough to rush such a thing.
>>> Removal before the next LTS should suffiice.
>>>
>>> We can ask the JCK team as discussed but that does not mean we then 
>>> don't need to fix the problem that caused this, even if we can defer 
>>> it to 13.
>>> Also "asking" does not guarantee approval, so a fix still needs to 
>>> be in the works.
>>>
>>> -phil.
>>>
>>> On 12/20/18, 8:49 PM, Shashidhara Veerabhadraiah wrote:
>>>> Hi Phil\Sergey, As I understand from this, we should remove the 
>>>> AccessibleResourceBundle from JDK in 13(By using 
>>>> @Deprecated(forRemoval = true)) and meanwhile ask the JCK team to 
>>>> remove this test from the test suite?
>>>>
>>>> Should the forremoval needs to be done now or later?
>>>>
>>>> Thanks and regards,
>>>> Shashi
>>>>
>>>> -----Original Message-----
>>>> From: Philip Race
>>>> Sent: Friday, December 21, 2018 6:54 AM
>>>> To: Sergey Bylokhov<sergey.bylokhov at oracle.com>
>>>> Cc: Shashidhara
>>>> Veerabhadraiah<shashidhara.veerabhadraiah at oracle.com>;
>>>> swing-dev at openjdk.java.net; Dmitry
>>>> Bessonov<dmitry.bessonov at oracle.com>; Krishna 
>>>> Addepalli<krishna.addepalli at oracle.com>
>>>> Subject: Re:<Swing Dev>  [12] JDK-8213516: jck test 
>>>> api/javax_accessibility/AccessibleState/fields.html fails 
>>>> intermittent
>>>>
>>>>
>>>>
>>>> On 12/20/18, 4:10 PM, Sergey Bylokhov wrote:
>>>>> On 20/12/2018 15:44, Phil Race wrote:
>>>>>> The peers were not part of the SE specification.
>>>>>> This class is, it just became obsolete so has been deprecated 
>>>>>> which on its own has no spec impact. So I would not call it a 
>>>>>> similar situation.
>>>>> No it was not part of the spec(and the deprecation notion is 
>>>>> unrelated).
>>>>> The notion that it should not be used and internal use only, is 
>>>>> there from the moment the class was moved to the 
>>>>> "javax.accessibility"
>>>>> package in 1998.
>>>> It is a public class in a public package and so forth.
>>>> But you can argue it out with JCK, as it is a waste of time to 
>>>> discuss it further here.
>>>>
>>>> -phil.
>>>>>> As I pointed out in what might have been an off-list comment, we 
>>>>>> can consider the deprecation for removal route, but that wouldn't 
>>>>>> solve the problem today.
>>>>>>
>>>>>> But excluding the test is a possible option for 12, so we could 
>>>>>> defer fixing the underlying regression until 13.
>>>>>>
>>>>>> -phil.
>>>>>>
>>>>>>
>>>>>> On 12/20/18 3:33 PM, Sergey Bylokhov wrote:
>>>>>>> On 20/12/2018 15:25, Phil Race wrote:
>>>>>>>> On 12/20/18 2:51 PM, Sergey Bylokhov wrote:
>>>>>>>>> I have checked the test which uses AccessibleResourceBundle 
>>>>>>>>> and I have two comments:
>>>>>>>>>     - This test should not be a part of jck since it is not a 
>>>>>>>>> part of public specification.
>>>>>>>> What isn't ? Do you mean the class ?
>>>>>>> I meant the class and the test which use it. BTW this class is a 
>>>>>>> good candidate for "removal=true"
>>>>>>>
>>>>>>>> If you mean the comment that it is not supposed to be called by 
>>>>>>>> external applications, then yes, as I already pointed out,  but 
>>>>>>>> the class does appear in the spec.
>>>>>>> It is there because we generate the javadoc for all public 
>>>>>>> classes, but the text for this class clearly state that it 
>>>>>>> should not be used. This situation is similar to the API which uses peers.
>>>>>>>
>>>>>
>>
>>
>> --
>> Best regards, Sergey.
>>
>
>