>(since this hasn't gotten a response over on the EPS board yet I thought I'd post it here)

They're usually pretty good about responding within a day or so - and usually it's in fits and spurts - so if you get a response, reply back ASAP with a followup if need be.

Regarding the reverse engineering, it's functionality is currently quite limited and somewhat unpredictable. Supposedly that is changing very soon with the next release. I had been emailing back and forth with EPS for a week or two about suggestions for changes, and I know they made a number of modifications - we'll have to wait and see how deep they go. If you'd like a summary, go to the EPS site, and search for my name. A thread sometime in the last month has a response from EPS with the updates. But alas, that answers none of your issues, so I'll give them a try:

>1) the two classes on the class diagram have not changed
>2) the superclass still shows the 3 methods

The reverse engineering wizard only brings in new methods and properties - it does not synch up the VFP and VUML classes. In my opinion, this is one of the most serious shortcomings of the wizard, and I believe (hope) it is one of the items they have fixed in the new version.

>3) the subclass doesn't show the methods

I would definately post this on the EPS site, and just wait for an answer. I have observed the same behavior, and it is quite frustrating. It has forced me to add the methods to the subclass. The really frustrating part is that it is sporadic, and I have not figured out how or when it happens. Definately a bug, and I haven't figured a work-around.

>4) the Classes items in the left pane show the same don't reflect the changes either.
>

I'm not sure I understand this issue. If you give me more detail, I may be able to help.

>If I create a new model from scratch using the classlib the model shows the methods in the correct classes. The class diagrams on these classlibs are so complex I surely don't want to have to build new models from scratch all of the time.

>
>Am I doing something wrong or is there just a problem re-reverse-engineering the classlibs?

As you have discovered, there are some serious shortcomings to the engineering process (both directions). However, this is the only VFP enabled tool out there, so we just have to live with it. It is nice to be able to bring in a single class created in VFP, and then manually set the associations and generalizations. I have developed a habit of working with both products open simultaneously (not recommended with Win98), and manually making changes in VUML as I make the changes in VFP. Despite the inconvenience, I think this actually makes me a better developer. I frequently find myself questioning my coding practices when I add the changes to the model by hand.

As a final thought, I will say that the folks at VUML and EPS have been quite responsive to the product's issues. Based on my flurry of emails with EPS, I suspect that they will continue to strengthen the VFP portion of the product, which can only help the community. So I've decided to be patient and just try to give them some feedback along the way. Markus himself frequently responds, which to me shows some degree of commitment. I think that in time, this could become a very good product. And after getting over it's quirks, I have come to like it quite a bit.

Hope this helps (and forgive the italics if they're annoying - a little test I'm doing).

David