Paul Vriens Paul.Vriens@xs4all.nl writes:
Hi,
added some tests with NULL parameters.
Changelog Added some NULL testing
I appreciate your willingness to write tests, but please don't spend too much effort on testing NULL inputs, it's not an interesting case unless there's a specific app that depends on it. Otherwise we end up adding NULL checks all over the place for no good reason. It's much more interesting to test valid inputs.
On Fri, 2006-07-07 at 18:24 +0200, Alexandre Julliard wrote:
Paul Vriens Paul.Vriens@xs4all.nl writes:
Hi,
added some tests with NULL parameters.
Changelog Added some NULL testing
I appreciate your willingness to write tests, but please don't spend too much effort on testing NULL inputs, it's not an interesting case unless there's a specific app that depends on it. Otherwise we end up adding NULL checks all over the place for no good reason. It's much more interesting to test valid inputs.
It's just that I'm going through a lot of the Coverity reports. A great deal of them mention NULL parameters passed. That why I started writing these checks and found that we're not always a 100% in line with M$.
I will try and not change 'real' code to check that much for NULLs but couldn't we leave the test cases (with todo_wine of course) for documentation sake?
Cheers,
Paul.
Paul Vriens Paul.Vriens@xs4all.nl writes:
It's just that I'm going through a lot of the Coverity reports. A great deal of them mention NULL parameters passed. That why I started writing these checks and found that we're not always a 100% in line with M$.
Yes, but that's deliberate, we don't want to add too many checks. Windows in many cases uses exception handlers, but that's quite expensive without compiler support.
I will try and not change 'real' code to check that much for NULLs but couldn't we leave the test cases (with todo_wine of course) for documentation sake?
If you have written them already, sure, but we don't want to systematically add test cases for NULL.
-
Alexandre Julliard -
Paul Vriens