net.digest – July
2004
Off-topic
political rants unfortunately returned big time, prompted in part by US Vice
President Dick Chaney’s intemperate comments on the floor of the US Senate. In
a lighter moment, Roy Brown posted,
Q:
How many MPE users does it take to change a light bulb?
A:
Four. One to change the bulb (two seconds, hot-swappable) and three to dissuade
management from changing the whole bl**dy light fixture.
Fortunately,
there was plenty of good technical comment, some of which we report here, to
balance the off-topic muck.
I
always like to hear from readers of net.digest and Hidden Value.
Even negative comments are welcome. If you think I’m full of it or goofed, or a
horse's behind, let me know. If something from these columns helped you, let me
know. If you’ve got an idea for something you think I missed, let me know. If
you spot something on HP3000-L and would like someone to elaborate on what was
discussed, let me know. Are you seeing a pattern here? You can reach me at
john@burke-consulting.com.
What if they held a meeting and nobody came?
I
was scheduled to speak at the 13th Annual INRUG Conference on June 7th.
During the afternoon of Friday, May 28th (my birthday and only 10
days before the conference was to start), I received an unwanted present; an
e-mail informing me the conference was cancelled for lack of interest, i.e. for
lack of paying registrants. So much for non-refundable airline tickets and my
belief the number 13 is not unlucky.
This
is just the latest incident showcasing how the once vibrant HP 3000 user
community is rapidly falling apart. The Solutions Symposiums, as well as
HPWorld, are suffering from sparse attendance by HP 3000 community members.
Vendors and consultants often outnumber end-user sites. Ironically, we know there
are still many HP 3000 shops that depend upon their systems, but they seem to
have developed a bunker mentality, perhaps hoping they will wake up one day to
find this is all a bad dream. I’ve got bad news for you: you are awake. It is
not just a bad dream. It is painful reality.
This bunker
mentality has had some unfortunate side effects. Even though HP is still
providing patches and even enhancements, many are not getting out of beta into
general release because too few people are willing to try them. I’ve spoken
about this with regard to network patches before. The latest problem was
reported by Mark Bixby in June: “While the patches for WebWise A.04.00 based on
Apache 1.3.29 were still in beta status, apache.org rolled out Apache 1.3.31. So I took the opportunity to rebuild WebWise
on 1.3.31 and issue new beta patches.
The current beta patch IDs are:
WBWHD95A - MPE/iX 6.5
WBWHD96A - MPE/iX 7.0
WBWHD97A - MPE/iX 7.5
“Few
customers have shown any interest in the old beta patches. If new versions of
WebWise/Apache are important to you between now and 12/31/06, please help the
above patches move towards GR status by contacting the HPRC and asking them to
send you a BT version.”
It
is time for HP to reconsider the beta process and the way patches become GR. If
they were downloadable, like GR patches, that would help. As it is, you have to
have an HP support contract to get a beta patch. It would also help if the
number of beta users required for a patch to achieve GR status were reduced. HP
is currently working on several enhancements from recent SIBs. If these never
go GR, we are wasting everyone’s time.
For example,
Jeff
Vance reported, “Based on the tactical SIB '04 results, vCSY has agreed to add
to the CI three new functions: SPOOLINFO, DEVINFO, and VOLINFO. The ballot
description for these three items explicitly states that the new functions are
to be integrated into the CI in a manner similar to FINFO and JINFO. We are
willing to do that; but integrated into the CI means a patch, and though the patch
will certainly target 7.5 and probably (maybe?) 7.0, it probably will not
target 6.5. The absence on 6.5 is to conserve R&D and MPE Support
resources.
“Assuming most
of the tactical SIB items are not back ported to 6.5, do you still wish to see
these functions integrated into the CI, or would it be more useful to do it
another way? We could, as a group, define the many CI variable names and types
(string, int, bool) which apply to each function, and then vCSY could write a
program to return the desired piece of information via the predefined variable.
This approach does not require a patch and would work on 6.5 as well as the
newer MPE releases. But this approach implies a process create each time the
function is invoked, thus, could not be used in break mode, and it is less
'natural' to invoke.”
While most of
the support dollars are probably coming from 7.0 and 7.5 systems, many believe
that MPE/iX 6.5 is running on the largest number of systems, and possibly even
a majority of all systems. vCSY is tacitly acknowledging both this and the
difficulty in getting users to test beta patches by proposing the above
implementation of the three “CI” functions. It must have been discouraging to
Jeff that his question received only two responses, and one of the two was from
me. The other response was a vote for integrating the functions into the CI,
admittedly the most aesthetically pleasing approach. However, this means that
few shops will ever actually get to use them.
Another
enhancement from the SIB process that HP is working on is CI User Functions. In
fact, this enhancement is close to being ready for alpha testing. It too is
being developed for 7.5, with back ports to 7.0 probable, but to 6.5 unlikely.
My suggestion is to take the engineering time required to integrate the three
specific “CI functions” into the CI and devote that to back porting CI User
Functions to 6.5. Once that was done, the new CI user function interface could
be used, with appropriate command file wrapper, to implement SPOOLINFO, DEVINFO
and VOLINFO "in line".
CI User
Functions were originally proposed as a way to make the CI extensible, allowing
anyone to create CI functions such as SPOOLINFO, DEVINFO and VOLINFO. The idea
was let HP do what only HP can do and let the rest of us create utility
functions. All three of these proposed CI functions can be implemented with
relatively straightforward AIF programming. I am not too concerned about break
mode. The process overhead is potentially worrisome, but so be it if this is
the best we can do to get this combined functionality to the widest possible
audience.
Both net.digest
and Hidden Value get their material primarily from HP3000-L. Many of us
have come to take it for granted – it has always been there and always will be.
An incident in June made it apparent again how lucky we all are that Jeff Kell
has donated so much of his time and effort to making HP3000-L what it is.
Unfortunately, few people realize this since Jeff has never been one to toot
his own horn.
The primary
reason we see very little spam on HP3000-L is that Jeff monitors everything not
sent by a subscriber to the list. If you always post to HP3000-L from your
subscriber account, you are not aware this is going on. However, if you post
from a non-subscriber account, it appears initially your post may have been
rejected and the post does not appear until Jeff has had a chance to identify
it as legitimate. Obviously, this can sometimes take days. This caught one of
the true gurus of the HP 3000 community in June. He unsubscribed in a huff
because he thought his posts were being moderated and that a particular post
was rejected because of its off-topic content. Fortunately cool heads prevailed
and he re-subscribed, understanding now the need to post from his subscriber account.
The whole
incident had the effect of focusing some of us on the tremendous asset Jeff has
been to the HP3000 community, a community that during the last decade
practically owes its very existence to HP3000-L. Thanks, Jeff.
Don’t
touch that dial. No need to check the cover. You are still reading the 3000
Newswire. Recent laws such as HIPPA and Sarbanes-Oxley have increased
tremendously the need for IT managers to be able to track who changed what and
when and who has access to specific information. Someone asked, "As a
requirement of Sarbanes-Oxley we are in need of a way to automatically log
changes to IMAGE data sets, KSAM and MPE flat files with a user ID and timestamp."
George Willis replied, "we have enabled Transaction Logging for our
TurboIMAGE databases coupled with the reporting tool, DBAUDIT, from Bradmark.
For your other files, consider enabling a System Level logging #105 and #160.
The LISTLOG utility that comes with the system can extract these records and
provide you with detail or summary level reporting."
Art
Bahrs provided more detail and a caution; “Transaction Logging will meet the
requirements for Sarbanes-Oxley and HIPAA for tracking the ‘touching’ of data.
However, you must also have a corporate policy relating to this tracking and
either a SOP or a formal procedure for reviewing the logs. The SOP and/or
procedure need to address what constitutes normal and abnormal activity with
regards to reviewing the logs and what action to take when abnormal activity is
noted. The fines for not being able to show who did what and who has access to
what can be very, very eye opening. Of course, these comments only apply to the
U.S. and businesses linked into the U.S. An interesting side note for all those
who are offshoring or considering offshoring Privacy or Personal HealthCare
Information (PHI), there are bills in both the Senate and the House to restrict
(read make illegal) the sending of Privacy and/or PHI Offshore.“
Finally,
we are seeing some real experiences that haven’t been washed and waxed by a
vendor before being presented. It all started with a rather broad request,
"I would like to hear from people who have migrated software from the
HP3000, or failed in the attempt. I'm interested in services, automated
translation tools, emulation, Eloquence, SQL databases, platforms, anything. We
have a lot of Powerhouse code--Quiz and Quick, with a tiny bit of QTP. Have
people had good experiences with Powerhouse on other platforms? We also have
about a million lines of Protos code. Is there an automated solution for this?
Do people take the generated COBOL and translate that?"
From
someone with obvious painful experience, "No matter who it is, don't let
anyone sell you on the idea of 'Best of Breed’ software. This concept is just
that, a concept. It doesn't work. The idea that you can buy Joe's financial
package, Mary's warehouse management system, Tom's order management system, and
then just plug 'em all together and they run like magic is pure myth. Every
purveyor of this concept/software will tell you how they have API (application
program interfaces) for all of their programs and, since everyone is using
JAVA/C++ and some flavor of UNIX/LINUX, you just use these API's to pass data
back and forth (don't even get me started on this one). In reality, and all too
often, you'll run into situations where Mary's warehouse management software's
API outputs a three-pronged-grounded-plug, but Tom's order management system
only accepts a 2-prong-non-grounded-plug. Then, my friend, you begin to
hemorrhage money to both Mary's software company and Tom's software company, so
they can fit together. And then there is the issue of ensuring synchronization
of version levels, etc.”
Another
caution, "The issue with Protos is the runtime library, which the
generated COBOL calls, and which you will need migrated or replicated on the
new platform. It's not impossible, but you have to reverse-engineer what the
library does, and provide that to satisfy the calls your COBOL code
makes."
On a
hopeful note, we have, "We have been converting from
HP/Cobol/Powerhouse/IMAGE to W2K/C#.NET/MS SQL over the last 18 months. After a
tricky beginning (painful re-learning process) I have to say it's been a
positive experience. The OO (object orientated) thing took a long while to get
my head around, but re-designing the systems using this methodology has been
beneficial. I have to say the new environment is much more powerful and
flexible than when we worked with Cobol/Powerhouse/IMAGE. The only thing I
would say against this whole experience is dealing with new staff that thinks
the computer world came into existence about 3 years ago. According to them,
XML is the only way to do everything and anyone who doesn't agree is
stupid."
Also on a
hopeful note, “Our migration experience to date has gone better than we
originally anticipated. As an HP3000/MPE/IMAGE shop for 30 years, our technical
expertise was naturally mostly in that environment, particularly with COBOL and
Speedware. After much study, we decided to migrate our HP-resident
administrative applications (Payroll, Accounting, Purchasing, Inventory, Human
Resources) to an Oracle-based PC package from a third-party designed for the
Quebec college market. Following a lot of hard work by our small MIS team and
on the part of our user community, the migration of all these applications was
completed in a matter of several months. Other than as a repository for
archival data, we no longer use the HP 3000 at all for these systems. Our
academic systems (Student Records, Scheduling, etc.) were all developed
in-house and remain on the HP 3000 for now but, following a great deal of
discussion within the college and beyond, we've decided on a collaborative
approach with several other local colleges to jointly redevelop these
applications. They'll be migrated over time to a new, networked Windows
environment. A strong point of this approach will be that the new applications
can continue to update our IMAGE database structure on a parallel basis so that
we'll only need to finally pull the plug on the HP 3000 when we're good and
ready.”