NetNews Usenet Archive 1992 #31/NN_1992_31.iso/spool/comp/database/informix/2770

Internet Message Format | 1992-12-22 | 5.4 KB

Path: sparky!uunet!cs.utexas.edu!swrinde!emory!emory!not-for-mail
From: [e-mail] (Will Hartung - Master Rallyeist)
Newsgroups: comp.databases.informix
Subject: Re: An insider's view
Date: 21 Dec 1992 15:08:04 -0500
Organization: Mailing List Gateway
Lines: 101
Sender: [e-mail]
Distribution: world
Message-ID: <[e-mail]>
Reply-To: [e-mail] (Will Hartung - Master Rallyeist)
NNTP-Posting-Host: emory.mathcs.emory.edu
X-Informix-List-ID: <list.1730>

> I recieved this message today from someone inside Informix. I've been
> asked not to identify the individual but am posting the message with
> permission.
>
> -----------------------------
>
> Hello,
>
>
> I hear every day that we have the best 4GL on the market ! It seems
> that our 4GL has a very big share of the Unix market. I don't hear
> very much from the marketing department that there should be so many
> missing features like popup menus, timer functions, improved input
> and display array statements, better cursor handling ( by the way
> ESQL/C has the same problem with cursor definitions. The only work-
> around would be to use the level below. On the C-level you see that
> cursors are structures. And there you can play with the address of
> these structures. ) and many others as we saw in the emails.

Gosh, I don't really recall the Marketing Department asking recently
either. Maybe Marketing doesn't hear about it because it's almost
impossible to point out these "little nits" in a simple demo. Early in
our sales we end up having to talk away from some of the "flash" that
clients like to see (namely lots of colors). And then later, they
start asking several questions very similiar to the suggestions we've
been hearing. Generally, we reply "No", or "Yes, but it's alot of
work." Which basically means "no".

Thankfully, the FourGen Tools do a lot of things for us. They've put
a lot the extra work into giving us some really nice functionality.
But as I'm sure has been noted, there's a lot of code there --
generated or not.

> And if these feature requests should be really justified - what is
> the problem ?  As you already mentioned - is there something better
> on the market ?  No - and people who come from Cobol or Assembler
> are happy with the product. Forget the few ones who want to make
> software with a screen handling which is state of the art. I would

Oh Heavens no, we never want anything state of the art...we know we're
asking too much there.

> say our 4GL is really a fourth generation language. And if there is
> nothing which is much better - why should we spend money for
> implementing all these features if we can sell it as it is ?

Yup, I can see your point. Resting on your laurels is always an
attractive alternative. Stagnation is cheap, easy to implement, easy
to support. Don't have to rewrite any sales literature, don't have to
retrain the sales and support staff. "We're number 1, let's shut down
R & D and head to Bermuda for 6 months..."

We poor sods that are banging all this code out for a buck certainly
don't need any improvements...we've already written work arounds for
everything, right? How does it work in the development meetings up
there:
    "We've got a request for dynamic scrolling arrays? Any takers?"
    "Oh, I saw some guy post 14 pages of 4GL and C to do that, we
     don't need to do that."
    "Okay, how about...."

My only hope is that Informix is putting all of its "Big Guns" into
the 4GL++. That 4GL development is basically bug fixes with the intent
that 4GL++ will knock its socks off and still be somewhat backward
compatible. Done properly, an Object Oriented 4GL can give us real
high level commands like "INPUT ARRAY", down to real detailed things
(like changing form attributes on the fly...). Heck, it might even
have consistent string expressions and maybe a few extra string
functions. It can do all this by letting the programmers get as
detailed as they need to, in the places where they need to, whenever
they want to.

Clearly, nobody wants to go deep down into bit twiddling to get
everthing they need done. The fact that we've seen request for
everything from UPDATE BY NAME to insanely flexible DISPLAY ARRAY
statements tells me that the base functionality is there, but sometime
the programmer has to go deeper.

I certainly can't imagine that the attitude of this "Informix Insider"
is prevalent throughout Informix. I sure hope not. "They can have any
color as long as its black" died out a long time ago. What is clear to
me is that the folks developing the 4GL don't use it. When the
developer of a product USE the product, in the real world, they
inevitably find something they can improve, if for no other reason
than to make their own life easier.

Dave Snyder is a great example of this. Every month or two his code
generator goes up a version point. I'm sure that Dave's is the "Number
1 PD 4GL Code Generator". Yet, gee, it gets better everytime it comes
out. Everytime he uses his generator he probably hits some wall that
he can't stand, so he fixes it, cleans it up, and makes it better.
Heck, I imagine he even listens to the folks that are using it in the
field as well.

FourGen does the same thing with their product. So, maybe Informix
should use their product in the field, or let their Consultants send
mail back to the development team. Of course, that assumes that the
Dev. Team is listening.

Will
(uunet!la4gen!villy - [e-mail])