From: Mike Cunningham
EGL sounds like a great tool for web development but then so did net.data
when it first came out. While net.data still works I don't know of anyone
doing anything new in that tool. How long do we give EGL to see if it will
survive before committing production applications that may be around for
10-15 years?
That's an interesting point, Mike. And while you may be right that EGL is
destined for the OS/2 rubbish heap of abandoned IBM software, I'll try to
point out a few reasons why I think it's not.
First, it's a lot older than Net.data. ND came out in, what, 1996? EGL is
the latest step in the evolution of CSP, the Cross Systems Product that has
been around since the 1981. And while it may sound sort of contrary to
picture a 1980s technology as the next great advancement, as we all know
there's little new under the sun, and what's important is how well something
is done. With EGL, I think IBM got it right. It's simple, elegant, with an
easy-to-use procedural syntax on top of a first-class OO architecture.
Second, a huge part of IBM is behind EGL. Net.data, while a great product,
was always one of IBM's best kept little secrets. Heck, most AS/400
developers didn't know about it, much less people outside the midrange
market. EGL is known throughout the IBM hierarchy, especially since it's
targeted at all levels of business, from mainframe on down.
Third, Net.data was a scripting language. A good one, to be sure, but it
was still a scripting language and all scripting languages have a
fundamental problem when it comes to analysis and debugging: they're
basically just interpreted text files. While that's great for initial
programming and proof of concept, it's very hard to develop an industrial
strength business application when your primary analysis tool is a string
search. With the Rational tooling, EGL has a complete IDE built around it
that is aware of the language from the ground up.
Fourth, EGL will be a standard. After my last statement I did a little more
poking around and found out that EGL is already being proposed to OMG (the
Object Management Group) as a PIM, a Platform Independent Model. Long and
short of it is that EGL is destined for great things, and rather than EGL
relying on i5/OS as Net.data relied on OS/400, instead it will be the other
way around, and i5/OS will actually be buoyed up by EGL's presence in the
standards world.
I don't know, I could be wrong. There are issues with EGL, such as the fact
that it will never generate RPG. But please go back and take a look at my
track record. The primary technologies I have pushed for over the last
couple of decades years are RPG, Java, the browser, JSP Model 2, Eclipse and
WDSC. I'm pretty much five for five on those, I think (even if you count
Eclipse and WDSC as one).
Oh, and what did I vote against? EJBs and Struts. And thick clients.
That's eight for eight. I only wish I was that good on my football pool...
The closest I've come to a miss is native I/O, which I love and which more
people than I care to admit seem to think should be completely thrown away
in deference to SQL. (But the jury is still out on that one, and I think
time will tell that I'm right there as well <grin>.)
But I haven't gone this far out on a limb for a technology since WDSC, and I
think it's a very similar situation: a new technology that actually makes us
old green screen dinosaurs more productive. In fact, I would probably be
willing to say that if you don't like WDSC, you won't like EGL and in fact
you're probably better off with RPG-CGI or PHP, but if you do like WDSC,
you're going to love EGL.
Joe
midrange.com
