Subject: | |
From: | |
Reply To: | |
Date: | Fri, 6 Sep 1996 14:39:45 -0700 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
Jeff Vance wrote:
<< SNIP SNIP most of the good stuff >>
> John wrote:
> > Specific reasons for what you are asking might be in an environment where
> > separate applications are spread over several accounts and you may not want
> the
> > users in one application looking at the "global" variables in another
> > application on the same system.<< SNIP SNIP all of the reply - my software
says I can't send
a smaller reply than the original message >>
Just a quick confirmation of this point.
We use those old HP3000 boxes(System 70).
We've ordered the new ones.
Don't know much about the IX.
We often have a job run in one account then have another job
complete a task. For example: Job1 does a database change.
Job2 moves new object code into place. Job3 sends a
verification note.
Job1 and Job2 don't need to communicate togeather; however,
Job3 must know the results of both of the other jobs.
A simple global "hunk of data" that is shared would solve the problem.
I, for one, would like to see the existing methods left unchanged
(add new functions with a different "name"). Keep it easy
to TEACH to others please.
Secuity in global data is self-contradictive. What it sounds like
is - this thread is moving the concept toward a OOVAR. A process
would obtain all the restrictions of a "hunk of data" based
on the methods which are a part of it. An Object VAR would
know what the rules are for its use when it is created since
it is created by a specific process which enforces the rules.
I would like to see the methods used in our system to be similar to
methods used on other systems. POSIX would be fine. This way I
learn one concept (OO or POSIX or distributed document jibberish).
|
|
|