OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

sca-j message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]


Subject: Re: [sca-j] NEW ISSUE: JCI90003 -- Java SE 5 requirement should be on the runtime not on the class


Hi Anish,

#1 makes sense. However, I don't understand why someone would want to disallow a runtime supporting Java SE > 5.0 on portability grounds. Is this a view someone is advocating?

Given that Java SE 5 has been end-of-lifed, I would expect many runtimes would not want to waste QA resources validating against it and instead use many of the new features in Java SE 6: 

http://www.oracle.com/technetwork/java/eol-135779.html

I don't see how supporting Java SE >5 materially affects portability given the existing extension point mechanisms in SCA. If an application requires 100% portability, it should use the minimum specified Java SE and no vendor extensions points. Taking this to an extreme, if application portability is the overriding requirement for SCA, then extension points should be eliminated as well (a bad idea). IMO, 100% portability is impossible for application development in SCA given all of the things it does not address (data access, UI, clustering, etc.).      

On a practical level, I believe the test cases assume Java SE 6 as they rely on the JAX-WS API as a JDK dependency. This could be changed (I believe it is possible to run the JAX-WS RI on JDK 5), but maybe the minimum should be Java SE 6?

Jim



On May 14, 2011, at 2:22 AM, Anish Karmarkar wrote:

> Thanks for bringing this up. It was discussed on the last call.
> 
> There are two related things here:
> 1) what is the minimum version of a complaint SCA runtime? Obviously that has to be 5.0, since we use annotation. Any runtime < 5.0 should be considered non-complaint.
> 2) can runtimes support Java SE > 5.0?
> 
> Point 2 is interesting. If we allow complaint runtimes to use features that are present in SE>5, then we can't guarantee portability of contributions but we provide flexibility and future-proof the spec (no need to rev the spec everytime there is a new Java version released -- Java 7 is slated to be released this summer).
> 
> My view is that, on balance allowing SE>5 makes more sense. Anyone who deals with deploying Java classes always checks to ensure that the underlying system's Java version is compatible. So, the potential loss of portability isn't that much of an issue.
> 
> -Anish
> --
> 
> On 5/9/2011 2:18 AM, Jim Marino wrote:
>> Hi Anish,
>> 
>> I'm not sure the intent of this is to disallow classes that use a feature in Java 6 or greater, but the issue reads that way. For example, a component implementation class could make use of a JDK 6 API. I think it should be valid for a runtime to support use of those APIs.
>> 
>> Jim
>> 
>> 
>> 
>> On May 9, 2011, at 8:22 AM, Anish Karmarkar wrote:
>> 
>>> Title: JCI90003 -- Java SE 5 requirement should be on the runtime not on the class
>>> 
>>> Specification: Java POJO CI
>>> 
>>> Description:
>>> 
>>> JCI90003 say --
>>> The Java class referenced by the @class attribute of<implementation.java/>  MUST conform to Java SE version 5.0. [JCI90003]
>>> 
>>> This seems unnecessary. We should allow classes compiled using earlier version of Java SE to be used with SCA. We should also allow classes complied with Java SE version>  5.0 as long as the classes don't use any features that are not present in 5.0.
>>> 
>>> If we need a requirement for Java SE version, that requirement should be on the Runtime not the class.
>>> 
>>> Proposal:
>>> 
>>> Change the target of the requirement to be the Runtime not the class.
>>> 
>>> ---------------------------------------------------------------------
>>> To unsubscribe from this mail list, you must leave the OASIS TC that
>>> generates this mail.  Follow this link to all your TCs in OASIS at:
>>> https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php
>> 
>> 
>> ---------------------------------------------------------------------
>> To unsubscribe from this mail list, you must leave the OASIS TC that
>> generates this mail.  Follow this link to all your TCs in OASIS at:
>> https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php
>> 
> 
> ---------------------------------------------------------------------
> To unsubscribe from this mail list, you must leave the OASIS TC that
> generates this mail.  Follow this link to all your TCs in OASIS at:
> https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php 



[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]