On Wed, Jan 05, 2000 at 06:05:27PM +0100, Thomas Michael Wanka wrote:
> Hi,
> 
> I discussed that with Manuel Bouyer, and understand that it is a 
> problem of the FS implementation of linux. Manuels first message left 
> the impresson it did not have to be OS related, and I defenitely did 
> not want to blame application programmers of doing a bad job!
> 
> But I do not think, that this is a real reason for or against an OS. It 
> defenitely is not an everyday problem, although I have to be grateful 
> to Manuel for this info, as I will check twice when choosing an OS 
> for critical systems. I think, that I should be able to handle most OS 

I think we finally agree :)
Sure this is not an everydays problem but, at last for me, I *want* my
servers to come back up without troubles after bad external even - just
because after a power shutdown there may be nobody around to hack the
servers to get them back up.

> around (I had my problems with QNX, but in the end it did what I 
> wanted) so I HAVE the choice. What about the everyday user? Most 
> of the people I know simply could not handle Linux or a BSD, I have 
> to recommend to these people to use the preinstalled MS OS on 
> their machines, because for them it is the best choice! I cannot 
> request them to use an OS they need a technician for every little 
> change!

Such peoples unsually don't have to make changes (at last it's what I
see in my environement), they just need a robust system.

> 
> The open source movement will not benefit from frustrated end users.
> 
> A good recommendation in this topic should sound like "The file 
> system implementation of linux may ristrict its use in critical 
> environment, and NetBSD is speedier on older less performant 
> systems ... "
> 
> I guess we need better recommendations than to convince people. 
> In the end it is their choice and they have to live with it and pay for it!

Sure, but a demonstration along with a recommendation is better.
My problem is that I don't always find the nicer words :)

--
Manuel Bouyer, LIP6, Universite Paris VI.           Manuel.Bouyer@lip6.fr
--