Even though I fully know that I'm being superficial and silly, I cannot help but find it hard to take seriously projects that do not use either git or mercurial for the sources in 2015.
What will git or mercurial gain them? There is no feature they want from git or mercurial. None. Zero. Zip.
DVCS? Easy branching and merging? No. No. No.
Why not? Because that's not how they code. It would actually promote bad behavior.
The OpenBSD way is to write your code correct the first time. Commit often, incrementally. Do not introduce large, sweeping changes that leaves things half broken. Be mindful that everyone else is going to compile and run your next commit. The result is that their HEAD/trunk nearly always compiles and works. They don't have to deal with huge code drops that now make it questionable whether or not they can safely cut a release every 6 months because the code hasn't been tested or reviewed well enough.
Their code practices are better than probably any other project and the lack of features in CVS are the reason.
Pretty much as I described -- developers making too many changes without anyone else seeing them and then later doing a big push and now lots of stuff that has changed since is broken.
The devil you know...
Clearly their current process is working for them though. On a big infrastructure project like that, is it really worth it for them to switch? They like what they've got.
Imagine the linux kernel development process pre-bitkeeper, mailing lists and patches. That's close to the openbsd process. CVS is mostly used as a record of changes.
