On 4/10/20 4:33 AM, Esme (they/them) wrote:
On Thu, Apr 9, 2020 at 3:22 PM Nikolay Sivov nsivov@codeweavers.com wrote:
In my opinion we should get rid of such macros, not add more. Especially for floating point tests.
I think you're right in this case. There's no point in tracing the pointer when we already know it's non-NULL.
But what's the problem with the existing macros?
Not a problem that requires immediate attention, but something to consider. For example expectf() uses arbitrary absolute difference threshold, which isn't great in general, because it should depend on magnitude of values being compared. And expect() does not offer anything more than straight ok() call does.