Re: [PATCH v5 09/45] Revert "block: Let replace_child_noperm free children"

[Hanna Reitz]

On 30.03.22 23:28, Vladimir Sementsov-Ogievskiy wrote:
> We are going to reimplement this behavior (clear bs->file / bs->backing
> pointers automatically when child->bs is cleared) in a nicer way.
>
> This reverts commit b0a9f6fed3d80de610dcd04a7e66f9f30a04174f.

This doesn’t really explain why it’s fine to revert this commit here.  
As far as I understand, the bug that was fixed in that commit will 
resurface when it is reverted without the proposed reimplementation, so 
technically, we cannot revert before reimplementing.

As far as I can guess, it’d be unwieldy to do the reimplementation while 
these existing changes are in the way, and it’d be one bomb of a patch 
to squash these five patches (9 to 14) into one, and that’s why you’ve 
chosen to do it this way around.

But technically, we can’t willingly break something just to keep patches 
nicer.  We can make exceptions, but then there needs to be justification 
here in the commit message.

(Or perhaps I’m wrong and it is fine at this point to revert the patch, 
but then I’d like to see the explanation for that, too, because I can’t 
see it myself.)

Hanna