Proposed incorporation of specification change for present/directly present/etc. terminology into javax.lang.model

Wed May 22 13:18:09 PDT 2013

On 05/22/2013 12:52 PM, Alex Buckley wrote:
> - For readability the definition of "associated" could use an "and" 
> before "<i>C</i> is a class".

Fixed.

>
> - The definition of "indirectly present" is missing a "the" before 
> "result", and wrongly implies that multiple annotations will always be 
> reported by calling the value() method ("A is one of the annotations 
> included in ...").

Changed to:

* An annotation A is indirectly present on a construct
 * C if both:
 *
 * <ul>
 *
 * <li>AT is a repeatable annotation type with a containing
 * annotation type ATC.
 *
 * <li>An annotation of type ATC is directly present on
 * C and A is an annotation included in the result of
 * calling the {@code value} method of the directly present annotation
 * of type ATC.

In the near future, I will push this version; webrev at

     http://cr.openjdk.java.net/~darcy/8010680.4

Thanks,

-Joe

>
> I updated JDK-8010680 to incorporate the edits I suggested earlier in 
> this thread plus the note you added to "directly present" ("Typically, 
> ..."). It's the text I will add to the PDF for long-term memory about 
> these definitions. If it's not what you want to put in javadoc, please 
> edit the bug.

>
> Alex
>
> On 5/21/2013 9:38 PM, Joe Darcy wrote:
>> Hello,
>>
>> On 05/21/2013 05:20 PM, Alex Buckley wrote:
>>> - "directly present" and "present" are OK.
>>>
>>> - "indirectly present" still doesn't bind A strongly enough ... it's
>>> necessary but not sufficient that an annotation of type ATC is
>>> directly present on C ... the sufficient clause is that the ATC-typed
>>> annotation must have a value element which contains A.
>>>
>>> - "associated" is almost OK except for a type error: the final clause
>>> must be "and <i>A</i> is associated with the superclass of <i>C</i>."
>>> (The domain of "associated" is annotations, not annotation types, so
>>> AT is never associated with anything.)
>>
>> Hopefully final iteration; edits to "indirectly present" and
>> "associated" definitions compared to the current code:
>>
>> + * An annotation <i>A</i> is <em>indirectly present</em> on a construct
>> + * <i>C</i> if both:
>>    *
>>    * <ul>
>> - *  <li> <i>A</i> is <em>directly present</em> on <i>E</i>; or
>>    *
>> - *  <li> <i>A</i> is not <em>directly present</em> on <i>E</i>, and
>> - *  <i>E</i> is an element representing a class, and <i>A</i>'s type
>> - *  is inheritable, and <i>A</i> is <em>present</em> on the element
>> - *  representing the superclass of <i>E</i>.
>> + * <li><i>AT</i> is a repeatable annotation type with a containing
>> + * annotation type <i>ATC</i>.
>> + *
>> + * <li>An annotation of type <i>ATC</i> is directly present on
>> + * <i>C</i> and <i>A</i> is one of the annotations included in result
>> + * of calling the {@code value} method of the directly present
>> + * annotation of type <i>ATC</i>.
>> + *
>> + * </ul>
>> + *
>> + * An annotation <i>A</i> is <em>associated</em> with a construct
>> + * <i>C</i> if either:
>> + *
>> + * <ul>
>> + *
>> + * <li> <i>A</i> is directly or indirectly present on <i>C</i>.
>> + *
>> + * <li> No annotation of type <i>AT</i> is directly or indirectly
>> + * present on <i>C</i>, <i>C</i> is a class, and <i>AT</i> is
>> + * inheritable, and <i>A</i> is associated with the superclass of
>> + * <i>C</i>.
>>    *
>>
>> Full new webrev at
>>
>>           http://cr.openjdk.java.net/~darcy/8010680.3
>>
>> Thanks,
>>
>> -Joe
>>
>>>
>>> Alex
>>>
>>> On 5/20/2013 10:35 PM, Joe Darcy wrote:
>>>> Hello,
>>>>
>>>> On 05/16/2013 11:56 AM, Alex Buckley wrote:
>>>>> 1. "directly present" says the annotation @Foo(x=1) is directly
>>>>> present if any annotation of type Foo is present. (Assuming an int
>>>>> element x in type Foo.) So is @Foo(x=2). And @Foo(x=3). And 
>>>>> @Foo(x=4).
>>>>> And so on. I know we only have type-based lookup for annotations
>>>>> today, but we should not open the door to weirdo answers for "tell me
>>>>> all the annotations of type Foo which are directly present". Try 
>>>>> this:
>>>>>
>>>>> -- 
>>>>> An annotation A is directly present on a construct C if either:
>>>>>
>>>>> - A is explicitly or implicitly declared as applying to the source
>>>>> code representation of C; or
>>>>> - A appears in the executable output corresponding to C, such as in
>>>>> the RuntimeVisibleAnnotations attribute of a class file.
>>>>> -- 
>>>>>
>>>>> (The "declared" in "explicitly or implicitly declared" does not
>>>>> pertain to solely declaration annotations. If you repeat a type
>>>>> annotation, then the type annotations' container annotation will 
>>>>> be an
>>>>> implicitly declared type annotation.)
>>>>>
>>>>> 2. The "present" and "associated" definitions both use the type AT 
>>>>> but
>>>>> it's not bound to anything.
>>>>>
>>>>> 3. The "indirectly present" definition also has AT spring out of
>>>>> nowhere, but also it's important to say "... if both of the following
>>>>> conditions are true:" rather than just "... if:" because it's easy to
>>>>> assume the bullet point are alternatives as in the other three
>>>>> definitions.
>>>>
>>>> Reworked portions of the patch below; full webrev at
>>>>
>>>>      http://cr.openjdk.java.net/~darcy/8010680.2
>>>>
>>>> Thanks,
>>>>
>>>> -Joe
>>>>
>>>> --- a/src/share/classes/javax/lang/model/AnnotatedConstruct.java 
>>>> Fri May
>>>> 17 13:48:41 2013 -0700
>>>> +++ b/src/share/classes/javax/lang/model/AnnotatedConstruct.java 
>>>> Mon May
>>>> 20 22:34:23 2013 -0700
>>>> @@ -39,35 +39,72 @@
>>>>    * are on a <em>declaration</em>, whereas annotations on a type 
>>>> are on
>>>>    * a specific <em>use</em> of a type name.
>>>>    *
>>>> - * The terms <em>directly present</em> and <em>present</em> are used
>>>> + * The terms <em>directly present</em>, <em>present</em>,
>>>> + * <em>indirectly present</em>, and <em>associated </em> are used
>>>>    * throughout this interface to describe precisely which annotations
>>>> - * are returned by methods:
>>>> + * are returned by the methods defined herein.
>>>>    *
>>>> - * <p>An annotation <i>A</i> is <em>directly present</em> on a
>>>> - * construct <i>E</i> if <i>E</i> is annotated, and:
>>>> + * <p>In the definitions below, an annotation <i>A</i> has an
>>>> + * annotation type <i>AT</i>. If <i>AT</i> is a repeatable annotation
>>>> + * type, the type of the containing annotation is <i>ATC</i>.
>>>> + *
>>>> + * <p>Annotation <i>A</i> is <em>directly present</em> on a construct
>>>> + * <i>C</i> if either:
>>>>    *
>>>>    * <ul>
>>>>    *
>>>> - * <li> for an invocation of {@code getAnnotation(Class<T>)} or
>>>> - * {@code getAnnotationMirrors()}, <i>E</i>'s annotations contain
>>>> <i>A</i>.
>>>> + * <li><i>A</i> is explicitly or implicitly declared as applying to
>>>> + * the source code representation of <i>C</i>.
>>>>    *
>>>> - * <li> for an invocation of {@code getAnnotationsByType(Class<T>)},
>>>> - * <i>E</i>'s annotations either contain <i>A</i> or, if the type of
>>>> - * <i>A</i> is repeatable, contain exactly one annotation whose value
>>>> - * element contains <i>A</i> and whose type is the containing
>>>> - * annotation type of <i>A</i>'s type.
>>>> + * <p>Typically, if exactly one annotation of type <i>AT</i> 
>>>> appears in
>>>> + * the source code of representation of <i>C</i>, then <i>A</i> is
>>>> + * explicitly declared as applying to <i>C</i>.
>>>> + *
>>>> + * If there are multiple annotations of type <i>AT</i> present on
>>>> + * <i>C</i>, then if <i>AT</i> is repeatable annotation type, an
>>>> + * annotation of type <i>ATC</i> is implicitly declared on <i>C</i>.
>>>> + *
>>>> + * <li> A representation of <i>A</i> appears in the executable output
>>>> + * for <i>C</i>, such as the {@code RuntimeVisibleAnnotations} or
>>>> + * {@code RuntimeVisibleParameterAnnotations} attributes of a class
>>>> + * file.
>>>>    *
>>>>    * </ul>
>>>>    *
>>>> - * <p>An annotation A is <em>present</em> on a construct E if either:
>>>> + * <p>An annotation <i>A</i> is <em>present</em> on a
>>>> + * construct <i>C</i> if either:
>>>> + * <ul>
>>>> + *
>>>> + * <li><i>A</i> is directly present on <i>C</i>.
>>>> + *
>>>> + * <li>No annotation of type <i>AT</i> is directly present on
>>>> + * <i>C</i>, and <i>C</i> is a class and <i>AT</i> is inheritable
>>>> + * and <i>A</i> is present on the superclass of <i>C</i>.
>>>> + *
>>>> + * </ul>
>>>> + *
>>>> + * An annotation <i>A</i> is <em>indirectly present</em> on a 
>>>> construct
>>>> + * <i>C</i> if both:
>>>>    *
>>>>    * <ul>
>>>> - *  <li> <i>A</i> is <em>directly present</em> on <i>E</i>; or
>>>>    *
>>>> - *  <li> <i>A</i> is not <em>directly present</em> on <i>E</i>, and
>>>> - *  <i>E</i> is an element representing a class, and <i>A</i>'s type
>>>> - *  is inheritable, and <i>A</i> is <em>present</em> on the element
>>>> - *  representing the superclass of <i>E</i>.
>>>> + * <li><i>AT</i> is a repeatable annotation type with a containing
>>>> annotation type <i>ATC</i>.
>>>> + *
>>>> + * <li>An annotation of type <i>ATC</i> is directly present on
>>>> <i>C</i>.
>>>> + *
>>>> + * </ul>
>>>> + *
>>>> + * An annotation <i>A</i> is <em>associated</em> with a construct
>>>> + * <i>C</i> if either:
>>>> + *
>>>> + * <ul>
>>>> + *
>>>> + * <li> <i>A</i> is directly or indirectly present on <i>C</i>.
>>>> + *
>>>> + * <li> No annotation of type <i>AT</i> is directly or indirectly
>>>> + * present on <i>C</i>, <i>C</i> is a class, and <i>AT</i> is
>>>> + * inheritable, and <i>AT</i> is associated with the superclass of
>>>> + * <i>C</i>.
>>>>    *
>>>>    * </ul>
>>>>    *
>>>>
>>>>
>>>>>
>>>>> Alex
>>>>>
>>>>> On 5/15/2013 11:22 PM, Joe Darcy wrote:
>>>>>> Hello,
>>>>>>
>>>>>> I've been working to update the javadoc specification in
>>>>>> javax.lang.model.* with the new present / directly present / etc.
>>>>>> terminology from the latest 8misc.pdf file.
>>>>>>
>>>>>> Proposed patch is below.
>>>>>>
>>>>>> Full webrev of the change available from
>>>>>>
>>>>>> http://cr.openjdk.java.net/~darcy/8010680.1/
>>>>>>
>>>>>> Thanks,
>>>>>>
>>>>>> -Joe
>>>>
>>