Well as of EAP 6.x (Java EE 6) JBoss seems to have them separately also at
runtime;-)
I don't have a more recent version here, so maybe they'll change it with
Wildfly...?
Kind Regards,
Werner
On Thu, Aug 13, 2015 at 4:00 PM, manfred riem <manfred.riem(a)oracle.com>
wrote:
Hi Werner,
Arjan is correct.
The runtime classpath will have the impl JSF 2.3 JAR and it will contain
the javax.faces and com.sun.faces packages.
The API jar we are referring to will not be part of the runtime classpath
and as such should never be on the server runtime classpath, but could be
used to compile against (and it will contain only the javax.faces packages)
Thanks!
Kind regards,
Manfred Riem
On 8/13/15, 8:55 AM, Werner Keil wrote:
OK, thanks for the clarification.
The way Arjan phrased it sounded like 2.3 would generally keep them in one
place.
Hence, if separate JARs are something CDI 2 has to consider,
getBeanClass() and other methods must work for separate API and impl JARs
;-)
Werner
On Thu, Aug 13, 2015 at 3:44 PM, Edward Burns <edward.burns(a)oracle.com>
wrote:
> >>>>> On Thu, 13 Aug 2015 15:24:00 +0200, Werner Keil <
> werner.keil(a)gmail.com> said:
>
> >> In this particular case it concerns the javax.faces.jar for JSF 2.3.
> >> Some vendors split up this archive in two, an API jar and an impl, jar
> >> (although for JSF 2.3 we don't really encourage this).
>
> WK> Could you explain, why the JSF 2.3 API and RI are molded together?
> WK> If the API JAR is no longer separate from the RI (as it was till
> WK> javax.faces-api-2.2.jar
> WK> <
>
http://search.maven.org/remotecontent?filepath=javax/faces/javax.faces-ap...
> >)
> WK> it seems, every vendor has to build that API JAR from scratch, or
> carry a
> WK> 3-4 MB JAR where the API in 2.2 was just a little under 700kB ?;-|
>
> We will be producing a separate API artifact for all the JCP
> milestones.
>
> It is very important to note that the for all Java EE API jars,
> including JSF, the only supported use is to satisfy compile time
> dependencies. Including a Java EE API jar in a runtime classpath is not
> supported and may cause unexpected behavior.
>
> If there is demand we can investigate producing a JSF API jar for
> -SNAPSHOT releases as well.
>
> Ed
>
> --
> | edward.burns(a)oracle.com | office: +1 407 458 0017
> | 58 Business days til JavaOne 2015
> | 73 Business days til DOAG 2015
>