29 Replies to “UHS-1 support in DragonFly”

  1. This is nice, but I think most people are waiting for some word on the status of HAMMER2.

  2. Are you going to post that comment on every non-HAMMER2 post I make from now on?

  3. Hi Justin,

    First many thanks for your excellent blog, I read it every day! Your answer made me laugh… must be kind of depressing though…

    At the risk of asking a naive question I would be interested in what UHS is, is it a hardware driver or rather a file system?

  4. It is depressing; there’s not much value in non-sequitur comments – but I keep getting them.

    It’s funny; I tried typing “UHS” into Google Search and most of the initial results were about University Health Services at a local school; searching for “UHS” and “Secure Digital” gave much better results.

  5. If Dfly had a release roadmap, a ton of these blog comments wouldn’t exist.

    Please ole please Dfly, just lost a simple release roadmap.

  6. Why does someone have to chase down each developer?

    Just have a wiki page for Release Sxhedule that allows developers to add to it.

    Don’t accept commits unless developers add to the wiki Release Page.

  7. At the risk of sounding pedantic, I too think that if Dragonfly had a development roadmap, even if a *rough* one, it would mean a lot to the community. I mean its probably a bit tiring for you Justin as well as the developers to get these comments, emails, IRC questions etc. asking the same things. It wastes everyone’s valuable time. If a roadmap were available one could just direct all these questions to one link.

    So +1 for the roadmap

  8. Saying “but you just have to do this” is not suddenly giving me time to do it. Aside from the fact that wiki and redmine pages have been set up for this before, and also the fact that making it harder to commit doesn’t help developers, I’m saying: if it matters to you, do something about it.

    That goes for all of you posting “+1”. If you were looking for the absolute minimum effort you could make – the least helpful assertion possible – that would be it. This is leaving me disappointed.

  9. Andrew – II’ve written multiple times that doing something is better than commenting about how someone should be doing something.

    So, maybe reread what you linked to and realize that all those descriptions are about work where there were a larger number of people putting in effort. And then go put in some effort.

  10. Hi,

    While not comprehensive there are some ideas about forward development that can be gleaned from here:


    A fair bit of it seems to be porting work which any number of people could immediately begin with should they have interest in Dragonfly’s progress. I think the supporting more than 63 CPUs bit is probably going to be quite important seeing as how AMD is releasing a 64 core / 128 thread thread dual socket system in the next month or so.

  11. Honest question: is DragonFly meant to be used in a commercial setting (production use) or is DragonFly a hobbyist OS for people to hack on for fun?

  12. I’ve always looked at commit messages and mailing lists to give me a rough idea of what’s being worked on and if you follow closely you can get a rough idea of timeframes.

    I’d much rather have developers developing instead of documenting.
    But that’s just me.

    I’d also guess that Matt et al would be delighted to have someone say “How can I help get Hammer2 completed?”

  13. Just an opinion, but it would be very difficult for me to believe that Dragonfly is just a hobby project. There is some very serious muscle in Dragonfly.

    IIRC there have been some questions from third parties concerning what can be done to contribute to HAMMER2, but I get the impression that its such a complex effort that there aren’t many with the necessary coding skills who would be willing to do it “just because”.

    HAMMER2 compression, encryption and copies have been on that gsocprojectspage for quite a while now.

    If we are waiting for some volunteer to come along and do it, it might not happen – ever.

    I would suggest having some kind of fund-raiser/crowd sourcing effort to get these bits off the ground. HAMMER2 benefits the entire BSD ecosystem. I’d be willing to hand in a minimum $200 USD contribution towards that. I am sure the folks over at OpenBSD might also be willing to help out with some of the encryption stuff because they badly need a new filesystem.

    Does anyone else out there think this is a fair idea? Can we bring this to Matt Dillon et al?

  14. Ok, you guys should really go to bugs.dragonflybsd.org . There is a roadmap in the redmine instance there and there has been for years. It doesn’t always directly reflect everything in current status as not everything is always tracked as a bug there formally and it’s not completely used as the ultimate source of truth for releases, but there has been a roadmap that goes along there for years now, at least 4 years. If you’d like to make any commentary about it, go make an account there.

  15. Maybe we can make our own roadmap :)

    DragonflyBSD 5.0 RELEASE

    – A usable HAMMER2 ready for widespread testing for local mounts. Rollout features being compression, deduplication, copies and filesystem healing for disks in a local cluster. This by itself gives us ~80% of what the vast majority of people want/use/need ZFS for and provides a real alternative to FreeBSD/ZFS. Local redundancy is fine for now, being able to have copies=2 spread out over two+ boot volumes effectively gives us a mirrored boot volume with failover without needing to rely on obsolete RAID cards and buggy firmware. OpenBSD folks can begin salivating now.

    DragonflyBSD 5.2 RELEASE

    – Clang as one of the two default compilers

    – Import npf (the multithreaded firewall from NetBSD)

    – Support for >64 CPUs (AMD’s EPYC CPUs are almost here; 64 cores / 128 threads in a dual socket setup)

    DragonflyBSD 5.4 RELEASE


    DragonflyBSD 6.0 RELEASE

    – The network clustering features for HAMMER2 are ready for widespread testing

    – A port of Dragonfly to ARM?

    Now we should ask ourselves not what DragonflyBSD can do for us, but what can we do for DragonflyBSD? Mr. Dillon et al., what can the community do to collectively to help DragonflyBSD be the very BSD it was destined to be?

  16. Interesting.

    I wonder if the developers actually read the comments posted here? Perhaps users@ might be a better venue to start the conversation.

  17. really hoping that hammer2 is being written with portability in mind. several attempts to port hammer1 were met with really non-trivial porting issues.

    is there an expected timeline for hammer2?

  18. If you’re going to make a roadmap, go ahead… but list only the things either you will work on or that someone else has committed to work on. Otherwise, it’s not a roadmap, it’s a wishlist.

  19. Wow. “I want X to be done”, “I want HAMMER2 to be done”, “I want a roadmap”, “I decide what makes an OS professional and therefor you should..”

    Come on. If you want that do something for it! And no, trolling in the hope that someone else does it for you is not doing something for it.

    Most people using any OS or software professionally aren’t even interested in a roadmap. You either are involved enough to see what’s going on, or you don’t change it and see what’s new and might be interesting to you anyway once the release come up.

    Most of the time you don’t care at all, cause stuff will simply be faster, saver, most stable, more convenient, etc.

    Sure, if you are a blogger you care and at your meetup group you care, but then you’ll likely look at stuff on your own instead of whining for someone else to do it.

    And seriously, it’s not too hard to do it. Just check the git log or mailing list, write a summary or even base upon this great blog, and every time you read about commit X make a note and you will see what was done.

    As for roadmaps. They are usually just interesting for devs anyway, cause else you won’t even know what will get done. Look at Fedora, look at Postgres, etc. They have lots of big features that might have been in a release years earlier, but they ended up doing it later cause they care about doing things right.

    And don’t act as if your opinion matters that much. Either you really do care, then you would do it, you use DagonFly professionally and contribute, or you use it professionally or as a fun OS to use, which is certainly fine, but you don’t add anything. Then why would anyone care about your opinion?

    Sane people make OS choices not about some random person whining in a the comments section of a blog. And nobody really cares for what non-sane people do. If a project ends up caring for people that just want X without contributing they will end up only having whining people and nobody profits.

    I know that might be harsh, but it really is how things tend to be in real life.

    And usually if you add something to a project, everyone is incredibly welcoming, helpful, digs into problems in their free time. Again that’s true for all big communities, be it OS, Programming languages, Databases, etc. Those just following some hype usually live less than five or ten years.

    Really, if you want HAMMER2 to be done, learn C, learn how File Systems work, and make some actual contributions. Yeah, people do stuff like that. That’s how people end up being OS developers and yeah, there are many very successful and famous kernel hackers who do that for a living, yes, as a hobby and still create stuff that Enterprises rely on.

    Look at Con Kolivas in the Linux world for example. Look at big contributors of various big projects.

  20. By the way.. not speaking for anyone, but I just don’t get how people can be complaining about X not existing on something that they don’t contribute to.

    That does work nowhere. Yes, you can make feature requests, but that feature is already planned and yes you can make wishlists… some people even turn those into talks at conferences, but that “most people wait for feature X” is horrible.

    The author of this blog probably is excited for it to, but if you want a status update: Write it yourself! If you want it to exist: Write code!

    Other things are bugging and certainly not helping to get there.

    But again: Not trying to speak on behalf of anyone. Not the author of this blog and certainly not a developer.

  21. About those who state “stop the complaining and contribute code” … note: not all of us who read this blog are developers who have those skills.

    Take me as an example. I’m a CTO of a smaller company. I can contribute financially (use my corporate budget) to Dragonfly but I can’t contribute technically.

    I feel uncomfortable contributing financially to Dragonfly when I don’t even know what you’re working on.

    That’s where a roadmap helps

  22. If you paid someone to work on DragonFly, for a feature that you needed, you’d know what they were working on.

Comments are closed.