Re: [tlug] Peeling onions.

[ijw@example.com]

>>>>>> "ijw" == ijw  <ijw@example.com> writes:
>
>     >> On the other hand, while you clearly can write operating
>     >> systems in C++, Modula-3, and Lisp, where are you going to get
>     >> the reasonable-quality implementations except from GNU?
>
>     ijw> Pay money? ;-)
>
> Paying money is acceptable to me; the licenses that go with price >
> zero mostly are not.  (I don't say "non-free" because I don't have a
> problem with many non-free licenses, a classic example being the
> Aladdin Ghostscript license, while on the contrary some "free"
> licenses are abhorrent, specifically the GFDL.)

There's a trade-off between paying money and not being able to fix bugs
when you find them, which is a consequence of the unavailable source.  On
the other hand, I'll pay money happily if I get a piece of software that
does its job well and isn't full of such bugs (and, if it does have bugs,
the company is responsive).  On the other other hand, it's always
difficult to find out if a piece of software is going to be OK until
you've actually paid the money, by which time it's too late.

>     ijw> - The HURD, for all its faults, is a great environment for
>     ijw> experimenting with user-level APIs and filesystems, because
>     ijw> you can effectively replace OS functionality.
>
> You can do that with Darwin, too, since it's also based on the Mach
> kernel.

The translation mechanism in Hurd means that FSes run as user processes. 
I'm not sure that the same is true of Darwin - the FS may be a Mach-level
process unit but I'm not sure how privileged it is.  Besides, Unix has no
means to convey "hm, your FS just died", so you either have to extend Unix
semantics or rethink your FS call interface, and I'm inclined to believe
that an interface rethink is called for at this point.