On 2008-02-27, Alexander Shishkin <[EMAIL PROTECTED]> wrote:
> I'd like to continue this conversation more subject-wise, like
> collecting ideas and opinions and with a higher seriousness than that
> of "hey, I'm writing a new operating system, it'll be mostly like
> linux, except for I'm the author" approach that I can possibly call
> for. Thanks for bringing this up.

Yeah, I think it is lame to (seriously) work on clone software with 
just minor tweaking (Yet Another Distro/WM/whatever). Instead one 
should take an entirely new approach, or at least turn a research
project into more practical software. In large scale systems, such
as an entire OS, that does seem to require too long an attention 
span for the modern culture, however. Not that it wouldn't be possible
to get to the desired end result with incremental improvement, but
first one has to be able to define that, to not end up with a 
complete clusterfuck.

Anyway, I'm all for a simple low-level core system (such as an 
exokernel) with capability-based security and various high-level
abstractions on top of it (typed pipes/shells, a decent config 
system abstract to applications, decent decentralised package 
management, vis, etc., just to throw some ideas around).

-- 
Tuomo

Reply via email to