This is to wake up all you developers out there, who work by the rule that debugging code should be removed from the release version.

You probably are wrong!

My point 1: If your application faults, the code to display any error message would execute, and it would not slow down the application in any way. It would certainly be in a if/then/else clause, that never would affect the main program loop anyway, so why not add some more info into it?

My point #2: "The debug level"
Many developers seems to think that they should hide any low-level error messages from users, because they could contain critical data, like passwords and such.

Guess what: WE DON'T CARE, if we just can switch debugging on and off as we please!
I'm perfectly happy if I can get an error message in the syslog, even if it contains a password, if I just can disable the error logging somewhere on a production server, it does not matter.

I'm a software developer myself, and one of the things I usually do in my code, is to create debug routines as often as I can, to resolve my own errors.
Many of you developers out there has such debug routines in your code, but what if the "common user" also could benefit from those, especially when things go wrong.

And, yeah, I know I'm fighting the guys trying to get the size of the Debian/Ubuntu CD's down to a minimum.
Get a grip on reality, and consider DVD instead if it's THAT important.

Just my 10 cents, any comments are welcome..

--
This message was entered after 20+ hours of wasted time debugging failures in nfs4/kerberos and nis on my home LAN without any solution.

I'm with you.

All my non-noddy scripts a) do logging b) report the output from all failed external commands c) have an option to turn on debugging which prints a function call trace and key values. They are this way because I reckon the lifetime effort involved is less than doing it any other way.