Re: Interactive program testing

[Vern Hamberg <vhamberg@xxxxxxxxxxx>]

Hi Dave

This sounds like regression testing, where the operations of an entire 
application are tested when changes are made. This is because a new 
feature or a fix can affect things outside the narrow focus of the change.

It is very typical to focus only on the new feature or change - of 
course, setting up a regression test is a really big task for an 
application of any complexity. And it probably needs some tools to help 
automate the process.

So is regression testing the reason for the new rules? If so, it's not a 
bad thing - but it does take longer to add or change something.

I don't know how to set up all this - I hope there is a balance 
somewhere. That one starts with unit testing, trying to find the 
possible side effects.

Can you imagine the scope of testing efforts at IBM that are related to 
the i5/OS? They have to add new features to scripts, etc. I'm aware of 
some of this from the periphery. But it's one of the reasons we trust 
this system so much.

If this IS what I think - your company should have a separate group 
doing this testing - it shouldn't be you that runs the entire test 
suite. If something is broken because of a change, then it will come 
back to you, but you are able to work on other things in the meanwhile.

At least I hope that is the plan!!

Good luck!
Vern

On 4/2/2012 6:32 AM, Dave wrote:
> Thanks, all
>
> Perhaps I should have said why I was asking. We didn't used to use a test
> library for the interactif part. This was not a big problem for me as I
> tried to externalise as much of the logic as possible. Now, test librairies
> have become compulsory for an interactif program. Which means if I make the
> slightest modification to a huge monolith, I will have to get out the huge
> corresponding folder containing the photocopies of the 5250 screens and
> then go from the start to the finish in the right order and test everything
> in the monolith. And no point in trying to rush it, the slightest mistake
> and you have to start over again. The cost of development has thus gone up
> overnight! I wonder though if we could benefit from this rule to encourage
> people to make smaller programs.