Vamsi,
I don't feel strongly about the case
that you describe.
If a using <component/> cares
about the value of a property, then it will set the value in the configuration
of that property. The component
can always get the value it wants.
I certainly don't think that componentType
side files are a good solution to this question - a huge hammer
to crack a very small nut.
The idea of a defaultValue attribute
to the @Property annotation is an elegant solution to the problem of
not being able to introspect an initializer.
However, I don't think the function implied is significant enough
to be worth bothering with.
Yours, Mike.
Strategist - Emerging Technologies, SCA & SDO.
Co Chair OASIS SCA Assembly TC.
IBM Hursley Park, Mail Point 146, Winchester, SO21 2JN, Great Britain.
Phone & FAX: +44-1962-818014 Mobile: +44-7802-467431
Email:
From:
C Vamsi <>
To:
Mike Edwards/UK/IBM@IBMGB
Cc:
Date:
25/02/2009 10:45
Subject:
Re: [sca-j] Why is @Property.required
attribute MUST be true for constructor parameter?
Mike,
The one case that will not be supported by doing away with componentType
side files for POJOs is the default value for a property since the field
initializer value can not be obtained through introspection (at least in
Java 5). Does this call for introducing a new attribute (say,
@defaultValue) in the @Property annotation?
Another case, but not of immediate concern, would the componentType file
support for implementation types defined in Java EE spec.
++Vamsi
Apache Tuscany Committer http://tuscany.apache.org
Apache Geronimo Committer and Member of PMC http://geronimo.apache.org
Mike Edwards
<mike_edwards@uk.
ibm.com>
To
25/02/2009 15:47
cc
Subject
Re: [sca-j]
Why is
@Property.required
attribute MUST
be true
for constructor parameter?
Vamsi,
You make a good point in reminding me that the Java TC has not voted
through a resolution to remove the use
of componentType side files for Java POJOs. I have now raised a new
issue
to do this - which will remove
Section 10 entirely.
I note that the Liaison committee posted a recommendation that
implementation types either use pure introspection
OR purely use a side file and never try to combine the information from
two
sources. Java POJOs can have all
their component type metadata derived via introspection - as a result side
files are not useful and add complexity.
Yours, Mike.
Strategist - Emerging Technologies, SCA & SDO.
Co Chair OASIS SCA Assembly TC.
IBM Hursley Park, Mail Point 146, Winchester, SO21 2JN, Great Britain.
Phone & FAX: +44-1962-818014 Mobile: +44-7802-467431
Email:
From: C Vamsi <>
To: Mike Edwards/UK/IBM@IBMGB
Cc:
Date: 24/02/2009 10:15
Subject: Re: [sca-j] Why is @Property.required attribute MUST be
true
for constructor parameter?
Comments inline in <vamsi> tags.
++Vamsi
Apache Tuscany Committer http://tuscany.apache.org
Apache Geronimo Committer and Member of PMC http://geronimo.apache.org
Mike Edwards
<mike_edwards@uk.
ibm.com>
To
24/02/2009 15:22
cc
Subject
Re: [sca-j] Why
is
@Property.required
attribute MUST
be true for constructor
parameter?
Folks,
When using constructor parameters for injection, there are a couple of
considerations to bear in mind:
1) Each constructor parameter MUST be annotated with either @Reference
or
with @Property, depending
on the type of the parameter. This is required in order that each
parameter has a NAME which then appears
in the componentType - without this, it is impossible to get a name for
the
constructor parameter (none is
available via introspection)
2) Clearly, each constructor parameter MUST supply a value, even if that
value is NULL. There is no
avoiding this. The runtime MUST pass in some value when it invokes
the
constructor. This is the reason
behind the requirement for required=true on the @Property annotation. The
using component can set
any value it likes for the property - potentially including "nil"
- but it
must set something.
On the point that Raymond makes that the property value "can be inherited
from the componentType" - this
is not correct. The componentType is entirely derived from the
implementation itself, not from some other
metadata. In fact, there is no separate componentType file any longer
for
Java implementations. So the
<vamsi>
Have we done away with componentType side files for Java implementations?
Is there an issue resolution pending to be applied? In
sca-javaci-1.1-spec-wd02.pdf, section 10, lines 509 thru 512, I still see
a
description of componentType side file for Java implementations. What
am I
missing?
</vamsi>
only place for a property value to come from for a constructor parameter
is
from the component configuration
that uses the Java implementation.
Yours, Mike.
Strategist - Emerging Technologies, SCA & SDO.
Co Chair OASIS SCA Assembly TC.
IBM Hursley Park, Mail Point 146, Winchester, SO21 2JN, Great Britain.
Phone & FAX: +44-1962-818014 Mobile: +44-7802-467431
Email:
From: Raymond Feng <>
To:
Date: 23/02/2009 21:09
Subject: Re: [sca-j] Why is @Property.required attribute MUST be
true
for constructor parameter?
My understanding is that @required=true means the property MUST be set
with
a value which can be either configured at the component level or inherited
from the componentType.
For example, if we have a constructor as follows:
public MyServiceImpl(@Property(name="myProp") String myProp)
{
...
}
The injection for "myProp" is required no matter what value it
is set.
Thanks,
Raymond
From: C Vamsi <>
To:
Date: 02/23/2009 10:41 AM
Subject [sca-j] Why is @Property.required attribute MUST be true for
: constructor parameter?
sca-javacaa-1.1-spec-cd02-rev2.doc:
Sec 9.15 @Property
Lines 1622-1623:
required (optional) – specifies whether injection is required, defaults
to
true. For a constructor parameter annotation, this attribute MUST have
the
value true.
Q: Why is, "For a constructor parameter annotation, this attribute
MUST
have the value true." required?
The componentType can define a default value for the property. In this
case, even if a property is not specified in the component definition,
the
runtime will still have a value to pass as an argument to the constructor.
++Vamsi
Apache Tuscany Committer http://tuscany.apache.org
Apache Geronimo Committer and Member of PMC http://geronimo.apache.org
Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number
741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6
3AU
Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number
741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6
3AU
Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number
741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6
3AU
