[openbeos] Re: OBOS Security
- From: Ingo Weinhold <bonefish@xxxxxxxxxxxxxxx>
- To: openbeos@xxxxxxxxxxxxx
- Date: Thu, 8 Aug 2002 12:16:09 +0200 (MET DST)
On Thu, 8 Aug 2002, Tom Young wrote:
> I think the point is that OBOS is designed to be a desktop OS and
> hence the ability to have multiple users logged on at once is not a
> big priority. Trying to keep focused on the goal is important,
> expecially at this stage before even R1 is achieved.
Multi-user in whatever form is not an R1 goal. So I think noone is losing
focus here.
> I'm sure if someone went and wrote a good true multiuser
> implementation that was stable, fast and secure then it would be
> adapted into the OS, and if not there is always dear i say it, forks
> and patches.
Implementing true multi-user is a big effort, as many parts of the system
have to be touched. Some will even need to be completely
redesigned/rewritten. It's nothing one person can do by just forking the
tree, implementing it and merging it back into the original tree a week
later.
Instead it will take months to mature and it needs to be compatible with
changes done to the original tree. To me forking seems not to be an
option, at least not, if one intends to re-merge later on.
> But actually encouraging this feature at this stage, or any stage in
> the immediate future is going to be counter productive to the other
> areas.
Noone thinks of implementing it till R1, but it is quite sure, that some
kind of multi-user will be implemented earlier or later. And it doesn't
harm to already think about it.
> Many universities and schools survived using win9x as a multiuserOS,
> it was far from perfect, but was widely used. I would highly suspect
> that if/when BeOS gets its pseudo Mutli-user abilities it would be a
> lot better than win9x.
I don't know about schools, but believe, that german universities (at
least the ones I know) attach some importance to security. OSs other than
Solaris or Linux are rare, and those that don't support true multi-user
are not really integrated into the networks.
You might say, that universities are not a primary target for OBOS, but I
believe universities would be more than happy with an easy to use, easy to
maintain (and reasonably secure) OS.
> Good integration with somthing actually designed to be a server such
> as a BSD or linux could achieve a high level of security as a
> workstation/desktop solution.
Mmh, I have no idea what you mean.
> I think its best not encourage OBOS to slip into the one size fits all
> aproach linux has and instead adopt the right tool for the right job.
Personally I feel, that true multi-user is a standard feature of a modern
operating system, regardless of whether it is a server, real time or
desktop OS.
If I can't ssh into my box then this is a major flaw. And even if I can do
this, but only one user can do this at a time, then it's not possible set
up a sane ssh based CVS repository. And if multiple users are allowed to
log in at a time, then I also want security. Nothing big -- I give a shit
about the SYN flood protection of my Linux box -- but basic protection of
the system's and other users' files and processes.
> Its also worth remembering at the end of the day, no system is 100%
> secure. It comes down to how much of an emphisis on security we
> believe is required for a Desktop OS.
Exactly. And IMHO this includes what I wrote above.
CU, Ingo
Other related posts:
