On 5/26/21 17:23, Francois Gouget wrote:
On Wed, 26 May 2021, Paul Gofman wrote:
It seems wrong to me to claim a Windows behavior is broken if:
- It is the behavior of all recent Windows versions.
Not that I see here. And the result doesn't make sense by itself, not sure why the size for all features should be zero. Sometimes we mark test output possible on Windows as broken because it looks weird compared to other possible results.
Options:
- Drop the test If we don't know of Windows applications that depend on this aspect and don't expect them to, then maybe the test is not needed.
At this point I think it is the best way. I'd personally prefer it to stay in the current state at least for now but if that causes concerns I think this test doesn't worth it, I highly doubt anything depends on it in the wild given it is not consistent.
- Investigate more to actually understand what is going on. Then we can decide what to do.
I can't do that without direct access to a machine reproducing this behaviour. If I had one maybe I would be curious to find out.