On 14/02/18 16:49, Fabio Estevam wrote:
On Wed, Feb 14, 2018 at 2:41 PM, Richard Fitzgerald rf@opensource.cirrus.com wrote:
This is the actual problem you are trying to fix? The comment about { 0 } only clearing the first member is probably bogus, your actual problem is that it should have been { NULL } ?
Changing to { NULL } fixes the sparse warning, but that would be a fragile fix because it relies on the order of the struct members.
If someday the struct changes in a way so that the first member is no longer a pointer, then we will have issues again.
Using {} is more error prone.
I agree but your commit message didn't say that. It's irrelevant now Mark has merged the patch, but for anyone looking at the commit message later, the structure of your commit message implies that the problem is that {0} doesn't work correctly (probably untrue and certainly not the actual problem.) and your actual fix relies on precisely the behaviour that the first line of your commit message implies is broken.
I just like commit messages to be accurate about what the problem was and why. It's often said that an incorrect comment is worse than not having the comment at all, and the same could be said for commit messages,