The System360 Model 20 Wasn't As Bad As All That 3881

snip

There seems to be something vaguely wrong with the above paragraph ("the software"? "same results between two compilers"?), but I think I get the idea -- that there are low-level differences between machines that can affect application-level software, and a really smart user will test for these differences? and that different compilers given the same source code might produce executables that give different results?

Yeah .... (So, this thread may be just about to dwindle away, because this might be the kind of question that for me requires some time on a mental back burner.)

snip

Ah, we might be talking at cross purposes here.

I was talking about testing *the GUI-based program itself* -- e.g., if you put the same formulas and values in the spreadsheet, does it give the same results?

You're talking about -- something else, something lower-level, I think. Okay.

Which is why I'm inclined to think that there are such programs: I can't quite believe that, however buggy some GUI programs are, they aren't tested at all, or that they're tested exclusively by sitting people in front of them. I'm sure I could find this out but am not quite curious enough right now ....

With extra software that captures and "plays back" user input? or that allows driving the GUI programs using simulated input and recording output so it can be compared to previous-correct output?

I think we may be talking at cross purposes here.

However, I think we probably agree that the problem of coming up with repeatable tests is more difficult ....

snip

Semantics, then. By "output" I meant, hm, pretty much everything that can be observed from outside the program -- what it puts in files, what it shows on the screen, both including error messages.

snip

My point was that I don't have-make a lot of opportunities in my day-to-day life to observe different clbuttes of computer users -- which I *think* is what's needed in order to do what you describe. Hanging out in Usenet is meant to be, among other things, a way to partially remedy that.

-- B. L. Mbuttingill ObDisclaimer: I don't speak for my employers; they return the favor.