Re: Changing from L4 to something else...

[Jonathan S. Shapiro]

On Fri, 2005-10-28 at 17:12 +0200, Alfred M. Szmidt wrote:
>    >The claims that Mach is `very buggy' are all false, the only
>    >reason why it requires `alot of work' is simply because of hard to
>    >find bugs, but that isn't specfic to Mach.
> 
>    Sure it is. Bugs are hard in proportion to lack of quality design.
> 
> Mach has its qualities, it has its warts (it has more qualities than
> warts).  The same deal for EROS/Coyotos, L4, and any other kernel out
> there.

Absolutely. If EROS had started as an academic research project, and had
not inherited its initial architecture from the KeyKOS work, I'm sure
that it would have come out almost as bad as Mach. Graduate students
have many jobs to do, writing code is just one of them, and it isn't the
most important one. Actually, it's not even the second most important
one.

But EROS *did* inherit. I took a fair bit of friendly hassle from my
adviser for being too committed to production quality code, but I
couldn't see any point to breaking something that was working.

shap