Questions of meaning

[sawyl]

A subtle semantic point arose today concerning the precise difference between a test and diagnostic, whereupon I drew a rather Jesuitical distinction between the two, pointing out that diagnostics are designed to help locate a specific fault, whereas tests are designed purely to confirm functionality. All of this suggests to me that diagnostics should be designed in the most general way, should make as few assumptions as possible about the knowledge of the user and should complain loudly about every potential problem, whereas a test should be able to take shortcuts, should assume that the user knows what they are doing and should be as quiet as possible if all is well.

Why all this sophistry? Well, I have a test script (it's name ends in _test.sh so it must be a test, right?) which I wrote a few years back specifically so that I could run a few simple, hassle free, tests of the disc paths. When I wrote it, I knew exactly what I wanted to do - check all mounted file systems of a particular type using IO chunks large enough to test every disc in the logical volume - so I decided to rip a list of file systems from mtab rather than fstab. Fair enough. Anyway, somehow, this test script escaped into the wild and the guys in Sector 7G got their hands on it and now, whenever we have any sort of two banana IO problem, they run it and act all surprised when it doesn't complain about file systems that aren't mounted or that are mounted NFS or something.

I mean, what's the problem? It's a test, therefore it's DWIS not DWIM. Like the varnish advert says: it does exactly what it says on the tin.