From: Steven Spencer
Our system runs a food importer impeccably.
You've been raising concerns about a lack of PC integration since last July at least, and previously admitted to various gaps in your platform. But that's not my main point. I'd like you to start thinking in terms of running many food importers impeccably, instead of just one. Because if you don't, somebody else will, and the applications that you're supporting in the S/36 environment will probably be retired within a few years by another vendor that understands the cloud computing business model and technology associated with it.
I recently became acquainted with a Canadian company that is signing up dental offices throughout Utah. They are running cloud services based who knows where? But "where" doesn't matter. The point is that they understand the economies of scale of the cloud services business model, and that dental offices are eager to get rid of their in-house systems because they are not in that line of business. Dentists want to use technology, not manage it. Managing technology is not their core competence.
If you were to buy a model 720 today and connect 20-40 concurrent users to it you'd probably run your CPU pretty much all day long at 1-2% utilization. We need to start thinking in terms of hosting hundreds or even thousands of users on 720's.
Regarding PC integration, there has been an on-going discussion on Linkedin about various ways to generate Excel spreadsheets under IBM i. The number of alternative solutions actually surprised me. We use RPG to generate Excel spreadsheets, Word documents, email, PDF documents, HTML reports. For our daily backups, we generate encrypted .zip files and FTP them to an offsite linux server, instead of backing up to tape.
You may not be thinking in terms of mobile computing now, but get and iPad or an iPhone and your outlook will change. ERP vendors are adding handheld applications to their legacy suites. More and more organizations are demanding them.
Before you can use tooling to generate mobile applications, your first step must be to externally define your physical files, and learn how to create SQL views.
I have to admit that IBM is investing a lot more in cross-platform languages and interfaces and related tooling, but it is not too hard to learn how to create HTML templates, and scan and replace field markers, using RPG. You'd probably be surprised just how well suited RPG is for generating formatted output, whether that be HTML, XML, JSON, or whatever.
I noticed this week that our daily backups include over 3,000 HTML templates which we use in more than 400 web applications.
-Nathan
<stevenspencer@xxxxxxxxxx>
To: RPG programming on the IBM i / System i <rpg400-l@xxxxxxxxxxxx>
Cc:
Sent: Tuesday, March 13, 2012 9:31 AM
Subject: Re: the RPG problem - adrift in the PC-web wind
Hi Folks,
Steven
I see no purpose in working with RPG and the jambalaya of EGL,
or CGI and PHP, or Java and this and that. Or Open Access
and the three vendors.
Steven
Let me add that the context of this was new, fresh projects. In
terms of enhancement and maintenance and support of existing systems,
the RPG environment is fine. However, I am not a big fan of the
modernization route of RPG-ILE followed by ... whatever. It all
seems so unnecessary, it has multiple steps in many cases, and is a
ton of work for excellent functioning, "legacy" systems that happen
to run a company superbly.
If you can achieve the same results of modernization with simply
using a 4GL directly on iSeries data, why not (whether in DB2 or
QS36F) ? And IBM never supplied the web and PC interface to RPG,
so you ultimately have to do .. something. The alternatives are 4GL,
or only web-PC interface languages added to your RPG, or semi-kludge
(not meant to be offensive) fixes like Open Access.
The 4GLs tend to incorporate the web-PC interface into their basic
engine, so I do believe they have a conceptual advantage over the
attempts to help along the RPG with the last few years of
jambalaya. In other words, you end up with a single-unit programming
environment for new projects.
Incidentally, Joe could indicate whether, when he develops a new app
with EGL, he uses RPG at all, and if so, precisely where he moves
back to RPG. Apologies if that has been asked a few times before,
but it would help in evaluating the relationship. Maybe you end up
using RPG as a .... "report program generator" :) .
Nathan
It might help if readers knew more about where you're coming from in
order to clarify your meaning. If I recall correctly you developed
an RPG II ERP system with internally described screens and files and
have managed to earn a living maintaining it for something like the
past 30 years for small customer base. Now it appears that your
product is increasingly being marginalized and apparently adrift
while facing a PC-Web wind. Is that a reasonable conclusion on my part?
Steven
See above. Unless you felt you had a superb private toolkit of
progam generators or templates, or were copying over an existing app
as a starting point, why would you develop a new app in RPG instead
of a 4GL ? This is not meant combatively, but sincerely. Is there a reason ?
And if you are not going to develop new apps in RPG, and have a base
of RPG running in ancient mode, why would you spend a year's worth of
man-time to "modernize" the RPG. (e.g. All the way up to RPG-ILE
with external screens.) What is the big gain.
Joe Pluta makes a solid pitch for that 4GL being EGL (and yes, that
has longevity in the same manner like Lansa, Magic, and WinDev, if
you know of others that really have that full implementation without
too much language complication, share away) . And we might see other
ports of long-range 4GLs to be comfortable running on iSeries. We do
see many attempts in the SQL inquiry-type products, but they can get
bogged down in things like the Java interface internals of
communication, so that while they work, they do not place a
moderate-tech end-user in a very comfortable position. That is, they
were not designed for direct support of iSeries in the way that
iSeries utilities are (or program systems like MRC and NGS). There
is a vast difference in installation and long-term support
concerns. EGL does not have that negativity, I gather, being now
more or less a native iSeries install. My point is that the only
products of real interest are those that have many years of native
iSeries support.
John E. makes some points about EGL being late to the party, and I
think that is appropriate, and he mentions the architectural concerns
of the 4Gl being in a sense data driven. EGL came to the party after
a number of IBM iSeries hesitation, and the Visual RPG failure, and
it brought together an unusual marriage of environments. Joe
supports it nicely, others tend to be a bit skeptical, but I agree
that it sounds like it needs a fair shake as a 4GL alternative and
may give it a look-see. Incidentally, Joe, please indicate whether
you know if the EGL data dictionary really cares about the data being
in DB2 or flat file QS36F.
Nathan
If so, then you should be given some credit for a legacy product
that has withstood the waves of change during the past 30 years. But
you should also give some credit to IBM and the RPG compiler team
for the longevity of your product.
Steven
Yep, I agree. The RPG compiler is rock solid. I can debug anything
even with the ancient print debugger (which often seems more
effective than interactive). The language is well matched with the
control language and the OS. That is why it was such a shame that
IBM never did the direct and proper attempt at enhancing RPG direct
to the net and PC.
Nathan
You shouldn't complain that IBM has not provided a completely
seamless modernization path for it, or complain about the cost of
3rd party alternatives. It appears to me that you've already gained
remarkable value from your relatively small investment.
Steven
In terms of current analysis, I think this is a bit of a non
sequitur. I still tell the execs that the "new system" alternatives
from the golf course friends are pie in the sky, and would like cost
many $$ for nothing. (And they understand.) Our system runs a food
importer impeccably. However that does not mean I have to avoid
the plain truth about the lacks in RPG. The point is ... I am not
even going to go to RPG-ILE because I still end up largely
dead-ended, even after a ton of work, even with all sorts of
tools. What IBM did was incomplete. Joe argues for EGL, an
IBM-friendly 4GL, filling the gap, I say that 4GLs in general fill
the gap. Some techies love to show what they can do in their
personal implementation of CGI, PHP, Java or other tools, but I think
that is only sensible for an enterprise set-up where they really like
having the programming group with a wide range of overlapping PC-web
skills. e.g. If they are doing a bunch of website development
independent of the iSeries.
Nathan
And you should show more respect for how the IBM i platform and the
RPG language have evolved over the past 30 years while you and your
customer base chose to continue operating within the confines of the
S/36 environment.
Steven
Well, if I came off as disrespectful, I will apologize. The fact
that I can run the code from 25 years back perfectly today is truly
amazing, and it gives a system that is up 24/7 and runs a medium size
business with a relatively small crew. However, again, I do believe
that IBM missed the boat on moving forward.
Nathan
It would really surprise me if your customers didn't chose some
cloud offering providing support for tablets and mobile phones,
while retiring your RPG II code base sometime within the next 5
years. IBM i and RPG are well positioned to play a role in what is
quickly becoming a post-PC, cloud services era. But legacy systems are not.
Steven
Tablets and mobile phones are not our issue, however, and I do not
think that has a lot to do with the overall discussion, which is more
simply .. modernization, web-friendly, PC-integration-friendly, and
new development easy. If IBM had made RPG a full-orbed language,
then RPG templates and code-generators and 4GL shells would be in the
marketplace. IBM did not, thus all the discussion about EGL, open
access, the web languages, and robust 4GLs running on iSeries data.
Feel free to tell me what I am missing.
Steven Spencer
Queens, NY
midrange.com
