Hi again!

First of all, I must apologise to IBM for coming down on them on account
of my decade-long frustrations with Informix policy on 4GL! :-)

That being said, I'd like to elaborate on some points.

Fernando Nunes wrote:

>I wouldn't call it an EGL demo... We've saw an .avi of basically a very
>simple 4GL batch progam being converted to EGL. This conversion was done
>by the upcoming 4GL to EGL converter. There was no focus on the IDE or
>even on the EGL language itself.
>
>
True. I'm sure EGL and the IDE are good, complete products, that we
didn't have time to see in this event. Anyway, my comments are from a
point of view of a I4GL developer, for whom the only alternatives of
evolution have been dead-ends or third-party products. And the 4GL-EGL
evolution path seems not to be as easy as it could/should be.

>1- EGL wasn't mean to accept 4GL syntax. EGL is a much bigger picture
>than 4GL (more on this later)
>
>
This is a pity, but understandable in IBM point of view. Not in I4GL
developers point of view, since almost all the work is in place. If you
have a translator you could easily have a compiler...

>2- For people who want an OO language, there are plenty out there...
>Java is the most noticeable one.
>
>
Java is a low-level 3rd-gen language. I'm sick&tired of having to deal
with Java complexity to create the simplest tasks! I4GL is better for
most business-data tasks but lack of object orientation is a big minus.
OO & 4GL are not incompatible and *should* be merged.

>3- EGL environment has Java integration. You can call Java from EGL and
>EGL from Java.
>4- AFAIK EGL will allow yout to have multiple connections, to different DBs
>
>
That's good.

>5- Since you can call Java from EGL, sending mails, parsing XML etc.
>should be easy to do (if you have Java knowledge).
>
>
My point is exactly the contrary! You shouldn't have to go to a lower
level language to do common tasks!

>The "uglier language" is EGL itself... I wouldn't say it's uggly... It
>sure isn't 4GL... But anyone with a reasonable knowledge of 4GL or C
>will find it at least very "familiar". It has the usual procedural
>language constructs...
>
>The complexity of the IDE is a direct consequence of it's potencial...
>As I said earlier, EGL is a much bigger picture than 4GL...
>
>
No doubt! And I believe it's a good product. But I was ranting about why
Informix (and now IBM) doesn't simply create a better I4GL, when it's
already been done partially several times. There are many costumers that
want GUI-I4GL but don't want to go 4J's, Querix, etc. Why has it been so
hard to simply answer to these customers?

>>In summary: 4GL is a simple, powerful, language, in a simple
>>environment. Simplicity is key. Let me simplify your job and tell you
>>what I'd like to see in the next EGL presentation:
>>
>> 1. Start with a 4GL program with a form and a report
>> 2. Run a shell and type "newfglpc program.4gl" - creating program.jar
>> - and then run "newfglrun program" and show a window with the
>> program running in GUI mode
>> 3. Then, run "newfglpc --web program.4gl" - creating a .war, this
>> time - and then drop the warfile in an appserver; then go to a URL
>> like "http://server/4gl/program" and show it run on the browser
>> with the exact same look&feel that in GUI mode.
>> 4. And then, after establishing that the "New Informix 4GL" can
>> really substitute on the old one, only then, very carefully, the
>> demo could go on about JSF, app servers, complex IDE's (editing
>> and debugging 4GL directly), etc., so that nobody is scared. ;-)
>>
>>
>
>Comments:
>
>1- As it was said in the .avi intro, when released it's supposed to
>include a forms converter (or the converter will be able to convert forms)
>
>
Good. Then almost everything is in place... :-)

>2- It would be nice... But as you mentioned, there are products which do
>this.
>
>
But no IBM Informix product! Since this is a long time ambition for many
Informix costumers, would it be so hard to do this? They already have a
parser and a generator, converting 4GL to EGL. Why not wrap it all and
create a 4GL compiler? ;-)

>3- The GUI in a applet was done by 4Js. I haven't touched it for a long
>time, but deploying wasn't easy... You either sent a +300KB applet (not
>nice for network) or you tried to install the applet... but this was not
>very useful over the Internet... JSF makes more sense...?
>
>
4J's, Oracle, etc. do it as applets and they sell. It's much easier to
deploy and very acceptable in a local network, which is the environment
of most I4GL apps. Sure JSF is a good technology for future projects.
But you must remember that the main concern of a I4GL developer is the
millions of lines of code he already has. He knows this code. He loves
it. He doesn't want it to be gone! He, mainly, wants it to appear
prettier to the user and not to be considered an old-fashioned
programmer that creates text apps! ;-)

>4- Once again, EGL is not "the new Informix 4GL" and once again we
>haven't saw an EGL demo... only a film of a very basic (and pre-beta)
>conversion done by a "yet to come product".
>
>
Not "the new Informix 4GL"?!? You better warn the IBM guy who showed us
a slide with this title yesterday! ;-)

>Now... What I mean by a "EGL is a much bigger picture than 4GL"...
>I personaly don't think IBM have done EGL for replacing _I4GL_. EGL
>intends to be a _4GL_ for modern environments. Note the underlines...
>"4GL" and "I4GL". The "target" for EGL is not _only_ the I4GL
>environments out there... It's target are the programmig teams that have
>to focus on business problems and solutions and want a "RAD" to do it.
>
>
Then, I insist: IBM should pick up the pieces already created and make a
"new I4GL", answering the long ambitions of long-time I4GL developers.

>if you want Object Orientation, direct Java and JDBC, J2EE etc. and you
>are productive with, and know all this things, then you probably won't
>need EGL.
>
>
I disagree. Java is low level. OO and 4GL are not incompatible and even
work very well together, as Informix proved with New Era.

>But if you, like me, have to do a program very quickly, and you have no
>practice on OO environments, and you would love to be able to integrate
>your work with modern environments, then, either you're an old timer 4GL
>programmer or not, you may find EGL very useful.
>
>
Certainly. I'm sure this is only the first step for IBM in putting 4GL
in fashion again! :-)

Fernando

--
Fernando Fernandez
http://www.moredata.pt

sending to informix-list