> How do you write unit tests for applications written using
> a VB-style IDE? If there aren't any tests, this sounds
> like a step backwards.

That's an interesting question. I think the components themselves would usually be written as classes, and as a paying customer of a component, I'd expect the components to be well-tested and robust. I would then tie components together graphically, in VB style, viewing property lists and setting them, etc. Other parts of my app that are written as classes I could test in the traditional way. When I write an event handler, though, I'm curious how I would test that.

I'm not sure writing tests is something VB programmers often concern themselves with. I'm also not sure to what extent automated tests matter for the kinds of applications VB is used for. Plain old testing may suffice. My perception is that VB is often used for internal apps that have a small number of users. VB is so hugely popular, that I'm sure it's used in lots of ways. But the smaller enterprise app, used by dozens of people in one department and written by folks who don't necessarily think of themselves computer scientists, is I believe the main market project Rave is trying to target.

The event handler code that is typed in dialog boxes in VB style does end up in a class, so in theory it is testable in the usual way. But I suspect most VB programmers would see that as more trouble than its worth.

> How do you write unit tests for applications written using
> a VB-style IDE? If there aren't any tests, this sounds
> like a step backwards.


I think components of the kind this JSR talks about are just plain old Java classes + design-time metadata that allows their manipulation at development time. That manipulation, in turn, generates plain old Java code in the background. The tools help write that code in response to visual gestures performed by a developers on a design canvas. For instance, you can indicate that you want to link beans together, etc. That rich design-time manipulation ability makes these classes components.

The difference that Joe Nuxoll hints at is that we Java kinds tend to just tuck a few JAR files into a directory, add those JARs into our classpath, fire up our editor, and proceed to write code that interacts with the classes inside those libraries.

A VB person, on the other hand, would stick a component into a visual tool's component palette. The component itself might come in a JAR file, of course. But, instead of firing up an editor and writing code to manipulate the classes inside those library files, he fires up a design canvas, drags design-time representations of those components onto it, and assembles applications via visual gestures, i.e., dragging, dropping, etc. Then, on occasion, he might write small snippets of code here and there to enchance the basic interaction model defined via the visual gestures. The end result in both cases, though, is just plain old code that can be tested.

If your memory goes back to the bad old days of Fido Net and 300 baud modem you might remember the state of Windows shareware before Visual Basic and after Visual Basic.

Before VB you had to program in C or C++. The average quality of the programs was much higher than those produced after VB was released to the world. It got so bad that I didn't even try downloading any program that required vbrun.dll.

I supect similar things from this.

> Oh look - programming tools for non-programmers! Again.
> Great.
>
These tools are not exclusively for non-programmers. I only tried VB once, but did do some work with Delphi and quite liked it. And I'm a "programmer." I recently saw a demo of Sun's Java Studio Creator app, and was impressed with the speed. They could make a little web app much faster than the way we do them at Artima. Although I don't that kind of tool is not appropriate for what we're doing most of the time here, it seems a good fit for a large number of problems. So I think that tools like VB can help programmers move faster and are very appropriate in certain tasks.

The other thing that this Design Time API JSR reminded me of is more dynamic languages like Python. Such languages tend to help programmers move faster, and are easier to learn than Java or C++. They also focus more on building programs by interacting with objects, not classes, just like VB and Studio Creator.

> Clue - non programmers DON'T PROGRAM.
>
> OK, so maybe non-programmers is too harsh. Lets say that
> they are really just algorithmically challenged. So now
> we have a tool that automates construction of bad
> applications. Nifty. I can hardly wait to use some of
> these wonder-apps.
>
I think there are a lot of people who aren't programmers but who could benefit from doing some programming: scientists, accountants, engineers--lots of people. And I don't think it is a bad thing to try and figure out how to make programming accessible to them.

VB programmers have a bad reputation in the programming community. I spent about 5 years programming MS Visual C++, and in those days we all turned our noses up at the VB programmers. They were considered not as smart as C++ programmers. We felt they were programming in VB because they couldn't handle C++.

I didn't really spend time in the VB world so I don't really know the culture, but I don't see why using a tool like VB is inherently going to produce bad programs. Maybe what's missing is a set of best practices for VB programming that is part of the VB culture. I'm not sure what is valued in that culture. It does make sense to me that tools like VB could enable otherwise incompetent programmers able to produce something that kind of works on the surface, but is really crap and unusable underneath. But I doubt that's the main use case of VB, because otherwise wouldn't those 4 million or however many VB programmers lose their jobs?

I'm also curious what the demographics actually are for VB. Who are these programmers, and what are they actually building and deploying. I always imagined most of them building small enterprise apps used by 20 to 80 people, things that aren't mission critical. Does anyone know what the demographics are? My understanding is that Sun wanted to target these programmers with project Rave, and I believe Java Studio Creator is part of that strategy. Who are these people, and what are they building?