On Thu, 17 Apr 1997, Larry W. Virden, x2487 wrote:
> The first thing he wanted to do was to turn on the grid manager. Remember,
> we are using 1.09 here. He tried 'pressing the manager' buttons on the
> top of the geometry window. It _appears_ that these were just 'indicator
> lights' showing which manager was in effect, becausing pressing grid
> would not change to the grid manager - instead, we had to go to the main
> tool window and press grid there.
>
> Two other areas gave a bit of problem. The first was figuring out,
> after getting grid to activate, how to get the 'geometry' of the window
> to work. He would press north/south/east/west or sticky, and sometimes
> the buttons would highlight and sometimes they would not. It wasn't
> obvious to me what needed to be done. However, I suspect it was a
> paradigm conflict that we were encountering.
The Geometry Manager is gone for 1.10 which reduces the confusion
here :) In it's stead is a much richer attribute editor. This is
more in line with what programmers expect from GUI builders. It
also allows for plug-in geometry managers.
> Finally, a very interesting bug was encountered. Before we realized that
> we were not getting into the Grid Manager, my friend had tried setting some
> of the values in the geometry window. He set several to 0, trying to get
> a label widget in a toplevel to move to that position. What he ended up
> doing, by accident, was getting some variable in vtcl set to zero that
> began causing divide by zero errors. _Every time_ from that point on that
> _anything_ was pressed in vtcl, a divide by zero error occured. We had to
> kill the process to get rid of vtcl - even trying to quit caused this
> error.
I would dearly like to know how this happened so that I can fix it!
Do you have any clues?
> One thing I found myself 'longing' for was a log function in vtcl, where it
> would record every event that was occuring. In that way, I could send along
> the log as a bug report.
>
> With the proper extension, it might even be possible to use the log as a
> playback mechanism to 'drive' vtcl in a regression mode test type of way!
great ideas. I'll add it to the hot list.
-stewart-
