Fix bug GH-14456: Attempting to initialize class with private constructor calls destructor #14469
Conversation
There was a problem hiding this comment.
I'm not completely sure if the existing behavior was done on purpose or not.
The proposed change looks right to me, and I don't object against the PR.
However, some user code may relay on the existing behavior.
@iluuu1994 please take a final look.
There was a problem hiding this comment.
It does seem to me that the destructor should not be called in this case. The class may make assumptions about fields being initialized. Hence, calling the destructor might be unsafe. I'm not sure how much this matters, given that this code already throws. Either way, I think not calling the destructor seems more reasonable.
|
Oh, I missed that this was proposed for 8.2. I would have preferred to merge it for master and document the change. But I hope it doesn't matter in practice. |
No description provided.