This type of report is a good idea though because it's always going to e more thorough than the TCK signature tests will be.

The hard part though is eliminating all of the false alarms so that it's useful.

Cases that can trip it up:

- Public classes which are not directly exposed by the API but simply used to by the API (Util classes, factory impls etc)

- Java SE classes which likely trip up the report because the Java EE version is more recent and therefore requires endorsed or special class loading

- Vendor Implementation Hooks. These are allowed to exist and in some cases are even required. This is why one should always prefer the API shipped by the container they are using.

- Bug Fixes / MR releases might report slight changes to method signatures or additional methods.

Per Tomaz's point, a focus on EE7 is better but if someone wants to volunteer time on discovering EE6 compliance issues we would certainly address any issues discovered. Of course patches would be even better :)

-Jason

Sent from my iPhone

On May 21, 2014, at 10:29 AM, Tomaž Cerar <tomaz.cerar@gmail.com> wrote:

Agreed, that is why I am saying that javaee6 jar is not really representative, what was done with javaee7 one is completely different story.


On Wed, May 21, 2014 at 5:16 PM, Shelly McGowan <smcgowan@redhat.com> wrote:


Tomaz,

Those are provided by reference implementation.  We started the Java EE Spec API project a while ago for the reasons listed here:
https://github.com/jboss/jboss-javaee-specs/blob/master/README.md

These APIs were later promoted with the WELD archetypes[1] as the recommended approach due to a few issues with the javaee-api JAR from GlassFish repository; specifically,

- It's a shaded JAR, so you have no idea what it actually bundles
- The code is stripped out of all the method bodies, so you get an Absent Code exception if you try to touch any of the classes


Shelly


[1]https://community.jboss.org/thread/179735


----- Original Message -----
From: "Tomaž Cerar" <tomaz.cerar@gmail.com>
To: "Andrey Ponomarenko" <aponomarenko@rosalab.ru>
Cc: wildfly-dev@lists.jboss.org
Sent: Wednesday, May 21, 2014 11:01:14 AM
Subject: Re: [wildfly-dev] API differences javax EE6 vs JBoss EE6 specs


On Wed, May 21, 2014 at 4:27 PM, Andrey Ponomarenko < aponomarenko@rosalab.ru > wrote:



http://download.java.net/ maven/2/javax/j


I know where it is from, but question is, is it official javaee artifact?, AFAIK only with JEE7 did this became mandatory as part of JSR
for example just compare

http://repo1.maven.org/maven2/javax/javaee-api/7.0/javaee-api-7.0.pom
vs
http://repo1.maven.org/maven2/javax/javaee-api/6.0/javaee-api-6.0.pom

to see what I mean,
tomaz

_______________________________________________
wildfly-dev mailing list
wildfly-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/wildfly-dev

_______________________________________________
wildfly-dev mailing list
wildfly-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/wildfly-dev