<AWT Dev> <Swing Dev> [13] JDK-8216008: -Djavax.accessibility.assistive_technologies empty list leads to crash

[Philip Race]

FWIW I think this could be closed out as "not a bug".
An empty value string is an error and the spec. says AWTError is
how errors are reported. In fact I'll argue that 8 was wrong not
to have thrown an exception in such a case.
It was an accident of the implementation that it did not.
But if we want to be behaviourally compatible then ...
Note that since you now document this you need an approved CSR BEFORE 
pushing it.

And if you go the CSR route you may need to be more precise about what
you mean by "empty".

I think it needs to say

"If the list of assistive technology providers is null, or contains only 
white
space characters then the method returns immediately.
All other errors are handled by throwing {@code AWTError}"

@throws AWTError if there is any error in parsing or loading the ATs.

-phil.

On 1/21/19, 9:12 AM, Shashidhara Veerabhadraiah wrote:
> Hi Sergey, Here is the new Webrev for your comments:
>
> http://cr.openjdk.java.net/~sveerabhadra/8216008/webrev.02/
>
> Thanks and regards,
> Shashi
>
> -----Original Message-----
> From: Sergey Bylokhov
> Sent: Saturday, January 19, 2019 3:47 AM
> To: Shashidhara Veerabhadraiah<shashidhara.veerabhadraiah at oracle.com>; awt-dev at openjdk.java.net; swing-dev at openjdk.java.net
> Subject: Re:<Swing Dev>  [13] JDK-8216008: -Djavax.accessibility.assistive_technologies empty list leads to crash
>
> On 16/01/2019 23:03, shashidhara.veerabhadraiah at oracle.com wrote:
>> Another point is that the whitespace trimming should not trigger additional input processing  for the custom class that will be used for assistive technology. For eg., if custom class "FooProvider" is implemented and if we pass "  FooProvider  " by mistake, a bug may be created to trim the whitespace in this case as well!!
> Yes, my previous email suggested to always trim the content of the "input", it will cover all cases, the "atNames" is empty, the "atNames" contains only the whitespace, or the name of the class has some spaces at the start/end.
>
>> Thanks and regards,
>>
>> Shashi
>>
>>
>> On 17/01/19 10:28 AM, Shashidhara Veerabhadraiah wrote:
>>> We need one way to tell the system not to load any assistive technologies and which is being provided.
>>> If we add trailing white spaces removal then we may need to add another functionality to see if that class exists or not and based on that not to load anything(without throwing the error because we are parsing the input content). I think that would take a different direction. I feel it is fair enough to provide one way to tell not to load any assistive technologies and additional parsing would only lead to other ways/expectations.
>>>
>>> Thanks and regards,
>>> Shashi
>>>
>>> -----Original Message-----
>>> From: Sergey Bylokhov
>>> Sent: Thursday, January 17, 2019 1:33 AM
>>> To: Shashidhara Veerabhadraiah<shashidhara.veerabhadraiah at oracle.com>; awt-dev at openjdk.java.net; swing-dev at openjdk.java.net
>>> Subject: Re:<Swing Dev>  [13] JDK-8216008: -Djavax.accessibility.assistive_technologies empty list leads to crash
>>>
>>> I guess you meant whitespaces, then it is unclear why we should not consider "empty" parameter as a "mistake"?
>>>
>>> On 15/01/2019 20:37, Shashidhara Veerabhadraiah wrote:
>>>> Hi Sergey, I felt that a string with an empty space is intentional and should be considering it as a 'mistake' if done by the developers. Hence I feel it is not required(as there is a way to tell nothing to load). Please let me know if you think otherwise.
>>>
>>>> Thanks and regards,
>>>> Shashi
>>>>
>>>> -----Original Message-----
>>>> From: Sergey Bylokhov
>>>> Sent: Tuesday, January 15, 2019 7:32 AM
>>>> To: Shashidhara Veerabhadraiah<shashidhara.veerabhadraiah at oracle.com>; awt-dev at openjdk.java.net; swing-dev at openjdk.java.net
>>>> Subject: Re:<Swing Dev>  [13] JDK-8216008: -Djavax.accessibility.assistive_technologies empty list leads to crash
>>>>
>>>> Hi, Shashi.
>>>>
>>>> Should we trim all whitespaces from the "atNames"?
>>>> Otherwise the "atNames", which contains only one whitespace, will trigger the same error.
>>>>
>>>> On 09/01/2019 00:20, Shashidhara Veerabhadraiah wrote:
>>>>> Hi Sergey, Thanks for pointing that out. Here is the new Webrev:
>>>>>
>>>>> http://cr.openjdk.java.net/~sveerabhadra/8216008/webrev.01/
>>>>>
>>>>> Thanks and regards,
>>>>> Shashi
>>>>>
>>>>> -----Original Message-----
>>>>> From: Sergey Bylokhov
>>>>> Sent: Friday, January 4, 2019 3:48 AM
>>>>> To: Shashidhara Veerabhadraiah<shashidhara.veerabhadraiah at oracle.com>; awt-dev at openjdk.java.net; swing-dev at openjdk.java.net
>>>>> Subject: Re:<Swing Dev>  [13] JDK-8216008: -Djavax.accessibility.assistive_technologies empty list leads to crash
>>>>>
>>>>> Hi, Shashi.
>>>>> I think you can update an existing test:
>>>>> open/test/jdk/javax/accessibility/AccessibilityProvider/basic.sh
>>>>>
>>>>> On 03/01/2019 01:36, Shashidhara Veerabhadraiah wrote:
>>>>>> Hi All, Please review a fix for the below bug.
>>>>>>
>>>>>> Bug: https://bugs.openjdk.java.net/browse/JDK-8216008
>>>>>>
>>>>>> Webrev: http://cr.openjdk.java.net/~sveerabhadra/8216008/webrev.00/
>>>>>>
>>>>>> Since the mentioned list of assistive technologies for the system property(javax.accessibility.assistive_technologies) is loaded by jvm am unable to write a test case for this bug. The test file attached with the bug can be used for testing this fix.
>>>>>>
>>>>>> This fix removes the error(class not found error) by passing an empty list to system variable javax.accessibility.assistive_technologies and does not load any assistive technologies(as the list is empty). Currently empty list produce a crash which is not required as there was no class name mentioned in the list. Please read the bug comments for more information.
>>>>>>
>>>>>> Thanks and regards,
>>>>>>
>>>>>> Shashi
>>>>>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.java.net/pipermail/awt-dev/attachments/20190208/2c1b822c/attachment.html>