> Though "Philosopher without Character" sounds better. "Bad Philosopher!" only makes sense if "Philosopher" is your dogs name.

:)))

>I've been trying to get away from the soft landing concept with error handling; or rather I want the error documented as soon as it occurs and then allow the calling object to handle the landing.

I'm having trouble with this only in the sense I agree with you and yet want to talk more about it... which is sometimes hard without making it sound like a dispute. So I'll just try to add a few cherries on the cake.

>I really don't like
>

>if  vartype(tMan) <> “C”
>   * error
>   return .f.
>endif
>

>Because
>1. The program knows that there is an error that is just as serious as a syntax error.

...and yet does absolutely nothing about it. This is nearly equal to a boobytrap.

>2. and if the program does crash later then it will be much harder to track the error down.
...which is the sole blame of the guy who wrote the error concealment code in the first place. That seems to be the literal following of the rule "do not write error handling code for errors you can't handle".

>3. If it doesn't crash then one could have a worse situation: bad data in the data base or incorrect results from the users point of view. I would rather see a hard crash than ship a 20,000$ chunk of metal to somebody who didn't ask for it.

But the coder who writes a thing like this did ask for... well, whatever the trouble he gets in :).

>ASSERT really doesn't help at all for this situation if you ask me. Its good in debugging but not once a bug has made it into production.

ASSERT is just a debugging tool, it simply doesn't work when .exe is launched outside VFP. The nice side of it is that it's never going to throw the "feature not implemented" error as a forgotten "set step on" or "suspend" would.

>Thanks for putting up with my rambling!

I asked for it :)