river-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dennis Reedy <dennis.re...@gmail.com>
Subject Re: [Discuss - Remove JRMP and IIOP support]
Date Sat, 14 Nov 2015 03:13:32 GMT
+1 for keeping IIOP

You never know what you’re going to need to integrate with. Right now for me it’s fortran
(not that IIOP helps here, just an example), go figure



> On Nov 13, 2015, at 918PM, Peter <jini@zeus.net.au> wrote:
> Rivers IIOP implementation is very small and yet provides cross language capability.
 i think the minor maintenance burden is outweighed by its benefits.  Corba is also an actively
maintained standard.
> I think it should stay part of the platfom.
> JRMP should be deprecated, it provides less functionality than JERI and has security
> Just my thoughts,
> Peter.
> Sent from my Samsung device.
>   Include original message
> ---- Original message ----
> From: Greg Trasuk <trasukg@stratuscom.com>
> Sent: 14/11/2015 11:31:32 am
> To: dev@river.apache.org
> Cc: user@river.apache.org <user@river.apache.org>
> Subject: Re: [Discuss - Remove JRMP and IIOP support]
>>  On Nov 13, 2015, at 6:23 PM, Peter <jini@zeus.net.au> wrote: 
>>  i have done some investigation into implementing dynamic iiop stub generation. 
glassfish does this. 
>>  I'd like to retain iiop. 
> I guess my question is why keep IIOP?  Is there a use case that IIOP/CORBA covers that
is not adequately addressed by JERI?  If I’m not mistaken, the Sun crew put it in there
originally in hopes of being able to interop with EJBs.  I don’t think that’s a reasonable
use case. 
>>  We could certainly remove JRMP, but it's worth remembering that MarshalledObject
still ties us to RMI, hence the need to set a property that allows downloaded code  
>>  java.rmi.server.useCodebaseOnly 
>>  In my local copy of River, all instances of MarshalledObject are marshalled and
unmarshalled using MarshalledInstance .  The next step would be to use interface default methods
to replace MarshalledObject parameters with MarshelledInstance and deprecate. serial form
would retain MarshalledObject where it existed for compatibility. 
>>   Phoenix and the test suite still use JRMP.  In my local copy of River I've converted
the test suite to JERI, I may have committed it for River 3.0, I don't recall and would need
to check.  My local copy of Phoenix uses JRMP , but only to establish a connection with rmi
>>  i have also removed rmi stubs from Phoenix except for one required for activation
compatibility, that Isn't downloaded.  Instead phoenix  now uses reflection's proxy (again
my local copy). 
>>  rmi stubs have been removed from all other services in my local copy of River. 
>>  so there's quitez a lot of work required, something for River 3.1? 
>>  Regards 
>>  Peter. 
>>  Sent  from my Samsung device. 
>>  ---- Original message ---- 
>>  From: Greg Trasuk <trasukg@stratuscom.com> 
>>  Sent: 14/11/2015 05:02:42 am 
>>  To: dev@river.apache.org 
>>  Cc: user@river.apache.org 
>>  Subject: [Discuss - Remove JRMP and IIOP support] 
>>  Hello all:  
>>  I’d like to suggest removing JRMP support (i.e. pre-compiled proxy classes), and
IIOP support (i.e. CORBA).  JRMP is nicely replaced by JERI, which offers security and doesn’t
require you to create compiled proxy classes by running rmic.  JRMP support was originally
included in the Jini 2.0 release waaay back in 2001 or so, in order to allow backwards compatibility
with Jini 1.2 installations.  The IIOP support could in theory provide compatibility with
native CORBA services (does anyone still do that?) or EJBs, but to my knowledge, it wasn’t
widely used.  
>>  The main reason for pruning theses capabilities is that unused code still requires
maintenance and increases the chance of bugs.  Also I think that as we go forward with refactoring,
renaming, restructuring the build and so on, it seems wasteful to do that work on code that
isn’t actually in use.  
>>  Obviously, the code remains in Subversion and in the 2.2.2 release, so if someone
wants to get it back, we (or they) could package it into a different deliverable.  But I wouldn’t
plan on doing that unless there’s actual demand for it.  
>>  My thought is to put this out there for discussion - If there is consensus after
a few days I’ll call a lazy-consensus vote.   I’ll be happy to do the work in the 2.2
>>  So, I propose to drop support for the following:  
>>  netjini.jrmp.**  
>>  net.jini.iiop.**  
>>  QA Harness classes that test any of the above.  
>>  Cheers,  
>>  Greg Trasuk 

View raw message