[omniORB] (no subject)

[Duncan Grisby]

On Monday 28 October, "Donahue, Michael J." wrote:

> Can anyone tell me if the issues that came up in OmniORB version
> 3.x.x with the Name Service and Java versions prior to 1.4 were
> resolved in OmniORB 4.x.x?

The issue was not resolved since the problem is with the JDK ORB, not
with omniORB.

>    Of if there are any suggested work
> arounds.  I need to upgrade my Name Service from 2.8.0 to the latest
> version, but need to know if it can properly work with older
> versions of Sun's and IBM's Java ORBs.

The problem only occurs if omniNames is running on the same machine as
the Java client. It's only in that situation that it makes the
unwarranted assumption that if it's the same host, it must be using
the same ORB (!) and therefore can assume things about the length of
object keys.

However, there is hope, in the sense that if you get a CVS checkout,
the object keys of all naming contexts except the root one will be
longer than the 12 bytes that the JDK ORB is expecting (due to an
unrelated issue). Then you only have to deal with the root naming
context which is required by the interoperable naming service to have
the key "NameService". If you change line 377 of log.cc to use a
longer string, it will no longer be compliant, but the JDK ORB will be
happy.

Cheers,

Duncan.

-- 
 -- Duncan Grisby         --
  -- duncan@grisby.org     --
   -- http://www.grisby.org --