Weaving WebSphere: WDSc 5.1.2 Quick Look--EGL, the Next Generation

[MCWebsite.Staff]

** This thread discusses the article: Weaving WebSphere: WDSc 5.1.2 Quick Look--EGL, the Next Generation **
** This thread discusses the Content article: Weaving WebSphere: WDSc 5.1.2 Quick Look--EGL, the Next Generation **
0

[Guest.Visitor]

Weaving WebSphere: WDSc 5.1.2 Quick Look--EGL, the Next Generation

** This thread discusses the article: Weaving WebSphere: WDSc 5.1.2 Quick Look--EGL, the Next Generation **
Joe, I don't quibble with your general issues about 4GLs, but I'd like to say that in the 80's, Cognos' PowerHouse 4GL was dynamite. It truly was at a higher level than a 3GL and therefore had good cross-platform success. Speaking from personal experience, productivity was phenominal. The makers identified the differences between non-procedural reporting (QUIZ), CRUD templates (QUICK), and batch processing (QTP) and had a different 4GL tool for each within the one PowerHouse product. You could modify the "generated" code in QUICK, but even that was above the 3GL level. In general, any tool that understands the logic of your database or the structure of your software can do an amazing amount of work for you. Regardless, your points about the problems with 4GLs are well-taken, and your article informative. Thanks! --Chapin Kaynor

[J.Pluta]

Weaving WebSphere: WDSc 5.1.2 Quick Look--EGL, the Next Generation

** This thread discusses the article: Weaving WebSphere: WDSc 5.1.2 Quick Look--EGL, the Next Generation **
Chapin, I think a 4GL can be made to do good things, but there are two definite prerequisites. First, the tool has to be designed by people who write application software. The actual implementation can be done by CS majors, but the actual design of the tool needs to be by and for application programmers. Second, and most important, the tool needs to be used every day, both in production of applications and by the folks who write the tool. AS/SET was a perect example of a tool written by really bright people who never wrote an application in their lives (which is why it really was never more than about a 3.25GL or so). As I alluded to in my article, I personally think that the right kind of tool will allow you to design spheres of functionality and define the communication between them, and then let you decide how to implement the details. Once you've defined a CRUD server, then you've pretty much designed them all, and all that's left is the details. Get that part correct, and all that's left is the really tough stuff, like batch balancing and route analysis on the business logic side and really nice graphical interaction on the UI side. Make a tool that allows you to push a button and generate servers in either RPG or Java, and generate either thick or thin client, and now you're talking about productivity. Joe

[bharder@nlrha.ab.ca]

Weaving WebSphere: WDSc 5.1.2 Quick Look--EGL, the Next Generation

** This thread discusses the article: Weaving WebSphere: WDSc 5.1.2 Quick Look--EGL, the Next Generation **
I was always taught that a 4GL was a non-procedural language. You specify What to get, but not How to get it. That's what makes SQL a 4GL: Select * from tablename where fieldname = something There is nothing explicitly there in terms of an I/O cycle--it's all implicit. My understanding of a 5GL seems to be a little different from yours. My background gave me to understand that these were rule driven languages. I can't think of many off hand other than Prolog. These languages had an implicit logic cycle that accomplished "rule firing" in the background. Again, the programmer did nothing to create this logic cycle. It's implicit in the language, thus creating higher levels of abstraction and justifying the 5GL name.

[J.Pluta]

Weaving WebSphere: WDSc 5.1.2 Quick Look--EGL, the Next Generation

** This thread discusses the article: Weaving WebSphere: WDSc 5.1.2 Quick Look--EGL, the Next Generation **
I'm less worried about the strict definition of a term than the mainstream definition, Brian. James Martin is the guy who said 4GLs are non-procedural, but that was 20 years ago. Since then, the term 4GL has been used for just about any CASE tool, including things like AS/SET. As for fifth-generation, there is no mainstream. How many Prolog programmers do you know? If the definition of a language requires you to know what a Horn clause is, then it's probably not mainstream. That doesn't mean it's to be disrespected or dismissed, simply that its time hasn't yet come, and certainly not to our little neck of the woods. Joe