>Unless my beta testers' reports improve sharply, my next post is liable to be
>a Deadly Serious plea for help ...

I came to the conclusion the other day that along with my application, I 
should have shipped all my beta testers a plentiful supply of 
tranquilizers. Well, actually just _one_ of my beta testers... The 
frightening thing is that along with some real bugs and desirable tweaks, 
he's coming up with some excellent points about which I can do little or 
nothing, short of rewriting large chunks of the Mac toolbox, or coding 
sections of my application in totally different ways. And he's coming up 
with these points at about 100x the rate I can even analyse and respond, 
much less fix code!

Then there are the testers that get errors no one else gets, that I can't 
replicate no matter what I do, that looking at the code tells me "can't 
happen". Yet I believe they really got that error, and they say they can 
replicate it reliably. Now what? I've tried having them boot with all 
their extensions off, and walking them through step by step. And gone and 
found someone else with an identical Mac model, memory, OS, etc. (it 
works there).

At my "day job", on in-house applications, when someone calls with a bug, 
90% of the time they're "local" and I can walk or drive over to squash it 
in person if I have to. (And I've been flown to Tucson from Dallas once 
when nothing else worked.) That's just not possible with a "commercial" 
application, even during beta testing, at least not at "home or school 
market" prices. (I've demanded - and got - vendors to fly in and 
investigate/fix bugs on site for some high-dollar applications before, 
even on the Mac.)

I already had a great deal of respect for technical support types, and 
understood the reasons for the high turnover in those positions... but 
until you've had an app in widespread beta test, you don't _know_ what 
it's like!

>Error 56989: YOUR DATA INVALID

I used to think the Mac with it's cryptic "Type 3" errors was bad, before 
I started using Oracle. At least there's only a handful of errors 
_possible_ in the Mac OS, albeit a large handful.

And FB usually "points" at the spot in the line that it doesn't like on a 
compile error. (Although figuring out what to do about it is sometimes 
tough!)

Getting a "ORA-ERR-1604-Unable to connect" is not at all unusual; and if 
you look at the "Errors Handbook", which is a couple of inches thick, for 
1604, you get the enlightening message that this means "Oracle was unable 
to connect." Duh!

Every time I see a "non-crash" error that doesn't give me some clue as to 
what to do to _fix_ the error, I resolve yet again to write more detailed 
error messages myself, and check for more possible error conditions...

But I also remind myself of the COBOL programmer at a place I worked who 
got fired for a too-detailed error handler. The first time in one 
execution of the code that the user did "X", the message was "Error - You 
did X". The next time, it was "X is not a legal option at this point." 
The next time was "X doesn't work. Try something else." I believe he had 
eleven levels, each getting more explicit. #10 was something like "You 
can do Y, you can do Z, you can even do A or B; you can give up and quit 
this program. But you CANNOT get X to work here, no matter how hard you 
try!"

#11 was judged by his manager to be insulting and/or obscene, and his 
employment was terminated. (Yes, a user was, um, mentally challenged 
enough to have reached that message, _not_ on purpose, and filed a formal 
complaint...)

Bill