1. 2

  2. 1

    When you’re holding a hammer, all problems look like nails.

    1. 2

      Is Linux the hammer? Are unikernels? It’s unclear what you’re saying.

      1. 2

        He means that Linux is not designed for this purpose and that it is better to use systems which are. We had a recent discussion on this subject where I held the pragmatic view that Linux will do just nicely until a better alternative comes along which achieves the same (or at least a sufficient) level of performance and support such as Linux does in opposition to his principled statement that Linux is not suitable and the choice should instead fall on verified system like seL4. While such systems might have theoretical benefits they lack the level of performance and hardware support offered by Linux (et al, the same goes for the *BSDs or any of the commercial closed systems).

        I think both sides of this argument have their merits but in the end it is the result which counts. It would be splendid if seL4 (or something similar) made it into the limelight, gained adoption and ushered us all into a brave new secure verified world. Thus far this has not happened for some reason, whether it is just lack of exposure or the mentioned lack of hardware support and performance. Until such a time I think it does not make much sense to grumble about ‘Linux this monolithic that’. Start a project which shows how seL4 (or similar) can be used in real life applications and convince the world that way. Running code tends to win over users, misgivings on bulletin boards generally don’t.

        If you - that is ethoh - have some pointers to projects showcasing practical applications of seL4 (et al) please share them, that is a better way of convincing pragmatics like me of the viability of these systems. Until such time I will most likely continue with Linux or one of the BSDs, warts and all.

        1. 1

          Is Linux the hammer?

          Yes, that’s what I meant.