Closed gcjenkinson closed 2 months ago
FYI the changes I have made to protobuf. I'll have another look if I can minimise the set of changes required to suppress warnings about ambiguous provenance, these warnings are benign so it seems overly disruptive to have conditional code with different types.
I've tried compiling mysql80-server and I get a Bus error running the protobuf compiler. So there's clearly more work required here.
@gcjenkinson Is it still a draft?
I've lost a bit of the context here, but the changes produce a working protobuf and shouldn't be too controversial.
I'm fine with merging cheribsd.patch as is. The code is appropriately annotated and shouldn't break aarch64 packages. Could you squash commits and reference a repository that includes these changes in a commit message?
FYI, I'd expect PRs in CheriBSD ports to review changes specific to FreeBSD ports, not ports' code base changes. I think a review of patches in port's source code would work better as part of a PR in a port-specific repository.
A set of very explicit changes for CHERI. Some of the changes are required due to the reluctance to provide alignment greater that 8 bytes assumed by ubsan and tcmalloc debug allocation logic.
With these changes the protobuf port compiles but as the package archive does not contain the git submodules the tests cannot be ran.