In a recent note, McKown, John said:
> Date: Mon, 5 Mar 2007 12:19:54 -0600
>
> > >
> > > For example, I have the following Java jdk's installed:
> > >
> > > /usr/lpp/jdk140/J1.4
> > > /usr/lpp/jdk141/J1.4
> > > /usr/lpp/jdk142/J1.4
> > > /usr/lpp/jdk150/J5.0_64
> > >
> > > you can do:
> > >
> > > cd /usr/lpp
> > > ln -sf jdk150/J5.0_64 java
>
...
> If it were me, I might look at a member called JAVA with one statement
> in it:
>
> // SET JAVAHOME='/usr/lpp/jdk150/J5.0_64'
>
> Then in other JCL do:
>
> // INCLUDE MEMBER=JAVAHOME
> ...
> //SMPJHOME DD PATH='&JAVAHOME',
> ...
>
Almost as good, but it fails to cover applications that are not started
from JCL.
> But, then, how does this differ from, say, CICS? We have the SDFHLOAD in
> CICSTS23.TS23.SDFHLOAD. I've seen others say that they have it in
> SYS1.TS23.SDFHLOAD. The problem with any product which is not integrated
> with z/OS is that the location may vary. In your example, the binder is
> always called IEWL and is always in SYS1.LINKLIB. Well, unless, as I
> have done, you put a "back level" binder somewhere else due to some
> compatability problems (this was long ago).
>
Ah, but I see on our system:
Menu Functions Confirm Utilities Help
------------------------------------------------------------------------------
BROWSE SYS1.LINKLIB Row 13635 of 18253
Name Prompt Alias-of Size TTR AC AM RM
_________ IEWL IEWBLINK 000008D0 01622D 00 ANY 24
_________ IEWLOAD IEWBLINK 000008D0 01622D 00 ANY 24
_________ LOADER IEWBLINK 000008D0 01622D 00 ANY 24
So, for these either the IBM designers or our system administrator did us the
courtesy of providing us the useful aliases. I believe it has been so through
many releases. Something similar could be done with a symlink for Java, but
that onus should be not on individual customers, nor ISVs, nor on SMP/E, but
on the Java development group. At least they could tender a strong suggestion
to customers to create the Java symlink in a portable way.
Prior to APAR IO03469, SMP/E ran a login shell which found the JAVA path from
/etc/profile. I favored the change in IO03469; I saw good reason for it, and
IBM saw even more cogent cause than I. But I would have appreciated the
convenience of a symlink offsetting the lost function.
-- gil
--
StorageTek
INFORMATION made POWERFUL
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
