Checking software (Was: Re: XLF optimizer unreliable ...)

[Bjorn Engsig]

Article <9599 at ccncsu.ColoState.EDU> by flatau at handel.UUCP (flatau) says:
| I can't spend my life
|checking answers on another computer each time I run application
|on RISC6000.
Well, you HAVE to (maybe not a lifetime long).  5 to 10 years ago we struggled
with bugs in compilers, today we struggle with bugs in optimizers.  And it
is our job to make sure our software is reliable when we send it out - also if 
that means using workarounds around bugs in the vendor software.

Testing software is a never ending process, the major decision is when you
stop your testing and let the customer take over.
|
|One would hope that IBM puts its act together and fixes the optimizer.
Yes, and one would hope that NCR, HP, DEC, you name it do the same thing.
Our experience with at least 30+ HW/compiler combinations tell us that there
is no such thing as a bug-free optimizer.
-- 
Bjorn Engsig,	Domain:		bengsig at oracle.nl, bengsig at oracle.com
		Path:		uunet!mcsun!orcenl!bengsig
		From IBM:	auschs!ibmaus!cs.utexas.edu!uunet!oracle!bengsig