RFR: [15]JDK-8247955,doclint: don't complain about summary/caption when role=presentation

[Pavel Rappo]

Looks good to me.

> On 23 Jun 2020, at 01:09, Jonathan Gibbons <jonathan.gibbons at oracle.com> wrote:
> 
> Hi Pavel,
> 
> Thanks for the comments/questions.
> 
> On 6/22/20 6:17 AM, Pavel Rappo wrote:
>> Jon,
>> 
>> 1. Should we refine this wording to reflect the change?
>> 
>>     dc.no.summary.or.caption.for.table=no summary or caption for table
> This is substandard legacy use of HTML tables. I prefer not to do more work than is necessary to provide minimal support.  I also don't think we should advertise or encourage the use of role="presentation".  For the one use case in the JDK docs, it would arguably be better to use a list and some CSS, but since what is there is technically legal, I thought I would let it pass.
>> 
>> 2. Does it make sense to test all the possibilities? Currently our tests cover 50% of that space.
>> 
>> tested|caption|summary|role
>> ----------------------------
>>           +       +     +
>>           +       +     -
>>           +       -     +
>>    +      +       -     -
>>           -       +     +
>>    +      -       +     -
>>    +      -       -     +
>>    +      -       -     -
> 
> See answer to #1.
> 
> 
>> 
>> ***
>> 
>> I was surprised by
>> 
>>     Objects.equals(v, "presentation")
>> 
>> I guess, I was expecting a more familiar
>> 
>>     "presentation".equals(v).
>> 
>> But there's nothing wrong with using Objects.equals.
> 
> I have never, ever liked the backwards form of <object>.equals(<subject>)  just as a lazy way of avoiding a null check. It's up there with C-array syntax (int foo[]) as one of those reasons why I prefer to code in Java, for clarity, not in C, for obscurity.
> 
> You're lucky that I used the modern new-fangled API Objects.equals, as compared to the Good Old Days,  if (a != null && a.equals(b)) ...
> 
> -- Jon
> 
> 
>> 
>> -Pavel
>> 
>>> On 20 Jun 2020, at 03:40, Jonathan Gibbons <jonathan.gibbons at oracle.com> wrote:
>>> 
>>> Please review a small adjustment for doclint: to suppress the check for
>>> a table having a caption when the table has `role=presentation`.
>>> 
>>> There's an existing AccessibilityTest.java, but that is still focussed on
>>> HTML 4, which does not support role=presentation. So, a new test
>>> is added (for now) that is focussed on this test case, in HTML 5.
>>> When we drop support for HTML 4, we should merge the two files.
>>> 
>>> -- Jon
>>> 
>>> JBS: https://bugs.openjdk.java.net/browse/JDK-8247955
>>> Webrev: http://cr.openjdk.java.net/~jjg/8247955/webrev.00/
>>>