Are you an Excel guru?

[Mich the Tester]

That's just bad design and bad testing. Its only doing what it was told to do. If it had been built in SQL server it could still have been built with that logic error in there

Oh, you're right;it doesn't matter what the app is, I'm just using it as an example of data having real life consequences where somebody blindly believes what it says on the screen. Seeing as it was built by Accidenture and Ordina and audited by KPMG, it doesn't surprise me. They probably ran a load of TMap style glorified happy flow checks and said 'yep. it works', because that kind of tester 'earns' his rate by demonstrating that a system seems to do what it's supposed to do so that a PM gets sign-off. Context Driven Testers would approach it differently with an attitude of 'how can I make it do something it's absolutely not supposed to do'. But context driven testing isn't popular yet because it eschews 'best practices', faulty metrics and procedure and all the other guff that managers buy into.

[darmstadt]

I use it as an alternative to tables in Word. I now know how to make different cells different colours, cool!

[bless 'em all]

That's just bad design and bad testing. Its only doing what it was told to do. If it had been built in SQL server it could still have been built with that logic error in there

Not nessessarily bad testing - a badly designed system is inherantly difficult to test. On the other hand the entire essence of an asylum system would rest on the setting of the deport/not deport flag - perhaps they didn't take a risk-based approach to focus their tests?

[doodab]

Not nessessarily bad testing - a badly designed system is inherantly difficult to test. On the other hand the entire essence of an asylum system would rest on the setting of the deport/not deport flag - perhaps they didn't take a risk-based approach to focus their tests?

It's more likely to be user error IMO. I doubt they would leave something as important as ticking that box to a machine.

[Mich the Tester]

Not nessessarily bad testing - a badly designed system is inherantly difficult to test. On the other hand the entire essence of an asylum system would rest on the setting of the deport/not deport flag - perhaps they didn't take a risk-based approach to focus their tests?

'Risk based' approaches fail here too because the people making the judgment call on risk levels generally don't include the ones who suffer the consequences of a fault. You really need a clever tester who starts from the premise that it doesn't work until we have evidence that it does, and the lack of evidence for faults is not evidence for a lack of faults, and then expressly tries to make something do hideously wrong. You also need managers with enough neurons to understand what I've just said and enoug conscience to care.

[Mich the Tester]

It's more likely to be user error IMO. I doubt they would leave something as important as ticking that box to a machine.

Why do you doubt that? I think you struggle to believe this because you, yourself, can see the human consequences of such a checkbox and understand that a fully fledged human will suffer if it goes wrong, and of course that an asylum seeker is a fully fledged human being. Since this case it turns out it's gone wrong in at least 300 cases, because the check box is filled by an algorithm depending on the answers to a list of questions. Some people in the Dutch immigration service have less conscience than you.

[BoredBloke]

Well the plus side is that his life was not lost in vain. No doubt it will cause them to fix the checkbox in the database.

[doodab]

Why do you doubt that? I think you struggle to believe this because you, yourself, can see the human consequences of such a checkbox and understand that a fully fledged human will suffer if it goes wrong, and of course that an asylum seeker is a fully fledged human being. Since this case it turns out it's gone wrong in at least 300 cases, because the check box is filled by an algorithm depending on the answers to a list of questions. Some people in the Dutch immigration service have less conscience than you.

I'm surprised it's automated, primarily because I find it hard to imagine petty bureaucrats ceding control to an algorithm. That strikes me as exactly the sort of thing that would have the unions crying foul.

[Pogle]

Ah i remember the day several Eurofighters were grounded, because someone sorted one column in the EXCEL spreadsheet and it ended up showing that several engines needed to be replaced as they'd exceeding their flying hours.

On the subject of testing - i totally agree with Mich. We have had the most anal annoying testers at clientco driving me nuts, and I totally respect them and am very grateful for their work. I know that the code produced has been tested thoroughly.
Sadly, the current bunch don't even do regression testing, so lets just say i have less faith in this part of the project.

[vetran]

That's just bad design and bad testing. Its only doing what it was told to do. If it had been built in SQL server it could still have been built with that logic error in there

Unfortunately users tend to prefer expediency over accuracy.

This morning a colleague contacted me and said 'my Boss has spent all day uploading files to a Sharepoint that I haven't got access to' I replied send me a link.

The link arrived and it was to a sharepoint site someone created yesterday in totally the wrong place (I'm not on the Sharepoint team otherwise I would have had enforced multi level categorisation so we don't have 1000+ sites in the root Directory). The Sharepoint site I created 6 months ago at the behest of the EMEA VP in the correct place and advised them of is still there and untouched.

My colleague's boss says 'oh we will use the new site because I have uploaded 10 documents there.'