On Fri, 19 Dec 2025 20:01:13 +0000 Paolo Vincenzo Olivo <[email protected]> wrote:
> I'm not enough into NetBSD development to provide a reliable statement > on this, but the available TODO [4] seems clear about what needs to be > done still. As for updating our codebase, NetBSD will have to decide > sooner or later whether to rebase on OpenZFS or illumos ZFS. As emerged > in this thread, NetBSD's ZFS port is becoming increasingly outdated. In > my opinion, updating ZFS on NetBSD should be top priority. Otherwise, > the fewer users choose NetBSD for ZFS storage, the fewer bugs are > reported and fixes contributed...it becomes the usual vicious cycle, > and the risk is that of letting a fully functional port of the most > advanced and widely available file-system bitrot. I use a lot of outdated software and hardware and it still works. The main goals should be stability and reliability. So importing ZFS from other projects can be just as much effort as fixing what NetBSD already has and it could potentially introduce many new bugs and regressions. There is no bitrot if a software component is correctly implemented and then validated and free of bugs (mostly). The never ending cycle of adding more and more new features is not something that every project has to follow. NetBSD developers could finish its ZFS implementation and then freeze it in time. We can talk all day about doing this or that, but unless somebody puts extra time and effort into improving ZFS on NetBSD, the schedule could stretch well into the distant future. The best way to achieve this would be to define specific goals and hire developers to do the job. Paying somebody money to do the work they enjoy is the best motivation there is. Since the NetBSD foundation is a charitable organisation, they may be constrained in what they can do. It would be good to have commercial versions of "enterprise" NetBSD where users who paid for a valid license have access to support and advanced features not available in the free version. This can fund the development of features like ZFS and these can later be backported to the free version once the return on investment has been achieved. Without significant commercial involvement, NetBSD will always lag behind Linux and others.
