david,

i know when the exec is NOT connected to the UI
that Trace works fine.    if it's possible to use the UI
to make a .net file, and then run dx -script foo.net
in batch w/o the UI and watch the trace messages,
then i'd suggest that.

but from context i'm guessing you are trying to debug
display problems.  that might still work in script mode,
but might not depending on how much interactivity
you need.

i apologize that i don't remember the details, but
i think the reason the messages don't show up is because
of the way DXDebug() calls DXqmessage().   i think the
first parameter which is some sort of label.   when
the dxexec runs with the UI and sends messages to
the UI, the label determines what the UI does - errors,
warnings, etc have different labels.   i see from the
code that DXDebug messages have NULL as the label.
for your own debugging, you might find another
DXqmessage which does show up in the UI's message
window (like a warning or error or something) and
use the same label from it and see if that makes them
visible.

the other question is what the eventual "right" thing to
do with them is - most of them are pretty ugly messages
only intelligible (or not!) to the code writer.   i'm not sure
exposing them in the general case is necessarily a good thing.

nancy
[EMAIL PROTECTED]

Reply via email to