Whatever Happened To 4GL Tools?

ghoti asks: "A while back, I read about 4GL tools, and what could be done with them. Now I wonder why nobody is using them. The idea sounded so logical: Design a program at an abstract level, then add more information until the tool can create code or at least a framework. Now this might not be the best way to go about writing Quake IV, but for many applications, this would be very nice. Is anybody using such tools? Why not?"

Comments

Several Problems with 4GLs

[Christopher B. Brown]

• They tend to be extremely proprietary.

  You can't just buy into the 4GL; you have to buy into the database on the back end, and whatever "TUI" or "GUI" is on the front end.

  In the case of tightly integrated options, which buy you decent performance, such as Oracle's tools, this means "marrying" Oracle. (And no "With this ring I thee wed," and no "You may kiss the bride." There does, however, tend to be a rather one-way form of "wedding night"...)

  Alternatively, SAP R/3 provides a DB-independent and OS-independent option, but is a rather heavyweight system.

• When the Development Model Du Jour changes, the 4GL becomes obsolete.

  Text-based systems were obsoleted by GUIed ones, and both have tended to be obsoleted by "web app" ones.

  In order for the 4GL to buy "ease of use," it has to provide a very specific development approach, and that results in it being very "brittle" in a changing environment.

• Incomplete Results

  The "benefit" of a 4GL is that it makes certain things (perhaps creating certain forms of reports) very easy.

  Unfortunately, this may only extend to covering 60% of system requirements. Which is wonderful when prototyping, when you only need to fulful 40% of system requirements. But it is not so wonderful when you're fighting to get the rest of the way, and have to fight with a tool that just wasn't meant to add in whatever that last 10% is.

  I worked with a group that was building a GIS system atop PowerBuilder; they got about 1/2 done, and then had to discard it and recreate the whole application using MFC/C++ because they just couldn't push PowerBuilder far enough.