Page 1 of 1

Test Execution - Loop counter

PostPosted: 8. Dec 2008, 11:53
by AlphaTester

a counter for the amount of performed loops would be helpful.
This number shall also displayed in the Report.


Re: Test Execution - Loop counter

PostPosted: 11. Dec 2008, 19:25
by cgittinger
the whole loop-business opens the door to a bunch of new requirements and questions:
I often dont need (want!) the trace and logs of all iterations; actually I plan to add the following additional execution flag settings to steps which have a loop pin:
- dont trace at all
- only trace iterations with error
- only trace the first (first-N ?)
- only trace the last (last-N ?)
- only trace the first with error (first-N ?)
- only trace the last with error (last-N ?)
(any more ?)

that looks like we either need 6 more checkboxes :oops: in the step-menu, or a more flexible, general solution. (how about a smalltalk expression, which gets the loop-count, the iteration-count and the error status as arguments, and has to return true/false to control if a log-entry is to be generated ?)
The first would be easier to use for beginners - the second would be what advanced users want.

I tend to ask for the following compromise:
- add a multiple-choice pull-down item to the steps menu, which offers one of the following choices:
1) no trace
2) error trace only
3) control trace of individual iterations via a boolean expression

is this a good idea - or do we need more ?


Re: Test Execution - Loop counter

PostPosted: 16. Dec 2008, 09:54
by AlphaTester

I think that here are mixed up two things...
1) I asked for a simple Loop counter - Everytime the testplan will be executed again, this value shall be increased. Finallaly this number shall be displayed in the report ( and now we are at your points...)
2) The report opportunitys (which you're listed) are great. They could be very helpful for me.

Here is my automotive point of view:
Every TestStep which was performed during the test execution must be visible in the testreport.
Sometime the customer wants to see that a issue was tested x-tousand times. And therefore we need
to have a report in detail. (Current implementation: Only the last loop will be displayed - and that is not good)

For internal testing only a small report with failures could be from interest (and this is working fine).