Software

Ask something
Discussion Locked

Correct me if I'm wrong

0 helpful
24 replies

by ed woychowsky

Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.

I just received an e-mail from an outside organization regarding the last round of integration testing. The impression that the e-mail gives is that I am expecting too much when I expect the data to pass validation. Because, the author states, it is only test data and test data never passes validation. In addition, validating the data is unnecessary because users will be trained to enter valid data and once that training occurs all data will be valid and that validation is a waste of both time and resources.

I am reminded of a sign that I once saw in a bar in Alabama; "Trust in the Lord, all others pay cash."

24 replies

by fregeus

...that sounds like a recipy for future problems. There is nothing better than a user to find undocumented bugs!!!

TCB

by charliespencer

[snicker]

"...validating the data is unnecessary because..."

[giggle]

[giggle]

Okay, calm face. I can do this.

"...validatingthedataisunnecessarybecauseusers ... becauseuserswillbetrained ... trained ... TRAINED ..."

BWAAHAAAHAAAAHAAAAHAAAHAAAAAHAAAAHAAAAA!
BWWAAAAHAAAAHAAAAHAAAAAHAHAHAAAAAAHAAAAHAAA!
BWAAAHAAAAHAHAHAHAHAHAAAHAAAAAAAAAAAAAAA!

BWAHAHAAHAAAAHAAAA!

BWAHAAHAAA!

Ha!

Oh, man; my ribs can't take much of that.

by jamesrl

I'm holding up a project right now because of these kinds of issues.

Does this organization presume that no old data will be imported from any other applications? That the users will simply start from scratch?

James

by ed woychowsky

A number of managers and developers actually seem stunned when I suggested validating data. They seemed to be of the "train the users" school of thought.

On the plus side, these managers and developers aren't involved with developing the control software for a nuclear power plant.

by jamesrl

I did work for a company which designed and built nuclear plants.

But I worked in IT, so my projects were more around the business systems, not the control software.

I did have regular interaction with those types though, they were some of our more demanding customers.

I'm pretty confident that my friends in the control software teams were all pretty much sticklers for intense validation.

I'm in commmercial software now. I know if we don't validate, I will get the calls. The support supervisors know who to talk to.

James

by ansugisalas

But someone just like them might be.
Over here, the finnish state commissioned a "new type" of nuclear power plant...---... from the french ...---...!
Luckily it seems the governments technocrat supervising organ is being suitably hyper-anal... the project is years overdue because they insist on checking the weldings and castings for hairline fractures, which they always seem to find. Despite the fact that the french feel it's a waste of time to "validate" the procedures.

Trust, but make sure. It's the only way to go.

by gdb758

Far be it from me to talk badly about another consultant, as I've been a consultant for years, but validating data is what tells you that what you are doing, was done correctly. Even if you are validating "incorrect" data, it should be the data you are looking for, no? Expecting people to ONLY put in valid data, is a misconception that not only sets you up for failure, but is supremely naive at best.

Remember this and you will always be fine, "I can fix most things as more times that not, they need fixing. But, you can't fix stupid!"

Jerry

by tink!

That is so unbelievable I don't know whether to laugh about how idiotic that response is or cry about the fact that there ARE such idiots out there!

They obviously assume that the users will lose all humanity and become perfect with training therefore eliminating all possible TYPOS or other human mistakes.

LOL!

by notsochiguy

They want to get their money and get the h_ll outta Dodge as quickly as possible.

If they are a small/boutiquish type firm, I'd guess they already have the next client engagement lined up, and the resources they need to start that would be tied up longer doing the data validation.

by deadly ernest

that test data was created solely to force faults in the validation system, and if the test data didn't pass, neither did the system as it proved the system was faulty.

by nicknielsen

Murphy?

Cash their check quickly, Ed.

Or apply the first rule of consulting: http://www.despair.com/consulting.html

etu

by tig2

Um-kay. So it doesn't matter if your test data will be valid because it is a certain bet that your production data will be equally invalid?

End users can be trained???

Ed, baby... Run far and fast and plan to meet up with us in Louisville. You have to autograph my copy of your book. You promised.

by charliespencer

"Groo went to that DeVry class when we started the cave painting department. Isn't he showing you how to do these?"

"Well, yeah, but he showed me some shortcuts he says work better than what they showed him in class, and Accounting says three legs uses less charcoal than four, and hey, the system didn't reject it so it must be okay, right?"

by ansugisalas

Chernobyl was human error combined with failing machinery, just like three mile island. This takes the cake:
http://news.bbc.co.uk/2/hi/asia-pacific/461446.stm
Far as I recall, they decided to forego the designated transfer vessel for a soluted plutonium liquid or other such, and instead go with a more handy one... a bucket.
They figured that more of the stuff fits in the bucket.
That's right, just enough for critical mass... god sakes, the poor bastards saw the blue glare of raw nuclear reaction right under their noses. Most uncool.

This discussion is closed to new replies.