Closed trws closed 4 months ago
maybe the hardening flags in the buildfarm did.
I certainly didn't change anything by hand but maybe something in the buildfarm changed? I don't know the buildfarm that well.
We pull the hardening flags from the platform specfiles, so it could have been something we did, or even an update from RHEL. 🤷
All modified and coverable lines are covered by tests :white_check_mark:
Project coverage is 74.1%. Comparing base (
eca89f0
) to head (686f0cc
). Report is 168 commits behind head on master.
problem: a few places were using
&(vec[0])
to get an address of the start of a potentially empty vector. This is almost always actually safe in practice, but it's not allowed and technically could be treated like a null dereference as well.solution: use the
vec.data()
method that exists specifically for this purpose@milroy tagging you here since we happened to be talking about this code just the other day. As far as I know this hasn't been touched recently, but maybe the hardening flags in the buildfarm did. 🤷