I've communicated with several commercial organizations, but they
tend to be much more tight-lipped than those working on academic or
open source projects. Maybe they're just more busy :)

The primary reason the abbot project was started was that we had
investigated several commercial tools, and found them all lacking in
one respect or another. That will *always* be an issue with any non-
trivial problem. No solution fits all, so your question has to be,
how easily can I reach my solution? We needed focused testing for
custom UI components, and broader regression testing on the whole
system. I always expected the tests to be constructed manually, or
at least tweaked after a recording session generated an outline of
the user actions.

The strategy for recognizing and tracking individual components
within a large, complex hierarchy often depends on characteristics of
that hierarchy. A dialog-based application may require a different
approach than something forms-based. The strategies built in to a
tool will be based on what the authors have conjectured or had
available as an example.

Automated UI testing will never replace a good QA team; at best it
frees them from common, repetitive tasks so they can focus on more
exploration and analysis. One of my hopes for abbot is to increase
the general level of understanding about where and when such
functional testing tools are effective.

(cross-posting to java-gui-testing since this more of a general
testing issue)


On Jul 7, 2006, at 9:08 AM, Bergmann, Dr.Egon wrote:

> Hi,
>
> after (unsuccessfully) trying to use abbot for testing a real
> commercial heavy weight java / database application and going
> through the discussion in this mailing list, I am not sure that
> anybody has been using abbot for an comparable project as we
> consider to do. Our app consists of more than 100 dialogs and about
> 500 database tables. We need regression testing because of our
> development process
>
>
> Please give me an idea of the kind of projects abbot has been
> successfully used in.
>
>
> Should we use another open source project or even an commercial tool?

Timothy Wall
http://abbot.sf.net