8033366: Add configure option to allow RMIConnector IIOP transport be selected compiled in or not

[Erik Joelsson]

Looks good to me. Only a minor comment on indentation in 
jdk/makeGenerateClasses.gmk. We try to keep block indentation at 2 
spaces and continuation indentation at 4 spaces, so line 96 should only 
be 2 spaces.

/Erik

On 2014-02-06 13:06, Alan Bateman wrote:
>
> One of things that we are hoping to do in JDK 9 is to drop support for 
> the IIOP transport from RMI connector in the JMX Remote API [1]. The 
> motive is modules and the first step on this journey was to downgrade 
> the support to optional from required (something that we did in JDK 8 
> as part of the Compact Profiles and Prepare for Modules efforts).
>
> We're not yet at the point yet where we can change the API docs and 
> remove the implementation but we are the point where we need to build 
> without the IIOP transport. Aside from our modularity efforts, it is 
> useful to have it removed early in JDK 9 as this maximizes the time 
> that anyone using it has to move another transport.
>
> This is the first time that I've added a configure option to the build 
> so I may need help getting this right. The patch that I'm currently 
> using is here:
>
>    http://cr.openjdk.java.net/~alanb/8033366/webrev/
>
> This adds the configure option --enable-rmiconnector-iiop to opt-in 
> for including of the RMIConnector IIOP transport in the build. If the 
> option is not specified (ie: the default) then the IIOP transport is 
> left out.
>
> The only part that might need explanation is that SetRMICompilation 
> generates the IIOP tie and stub classes if RUN_IIOP is set to 
> anything. This is the reason why RUN_IIOP isn't just specified as 
> $(RMICONNECTOR_IIOP).
>
> As regards the tests then we changed the JMX tests in JDK 8 so that 
> they gracefully handle the case where the RMIConnector only supports 
> the default transport. I did have to update one langtools test that 
> directly references a generate stub classes. There isn't an 
> replacement for this because there aren't any remaining individual 
> classes excluded from compact3.
>
> The profiles content  file keeps its reference to the _RMI* classes 
> for now as this is needed for cases where someone does build with the 
> IIOP transport. This will be re-visited later in JDK 9 of course.
>
> -Alan.
>
> [1] 
> http://mail.openjdk.java.net/pipermail/jmx-dev/2014-January/000571.html