Re: Reiser4. BEST FILESYSTEM EVER - Christer Weinigel

[Richard Knutsson]

johnrobertbanks@xxxxxxxxxxx wrote:
> On Mon, 09 Apr 2007 00:58:53 +0200, "Richard Knutsson"
> <ricknu-0@xxxxxxxxxxxxxx> said:
>   
> > Wow, I'm impressed. Think you got the record on how many mails you 
> > referenced to in a reply... 
> >     
>
> TWO actually. I guess you are easily impressed.
>   
Oh, took it to be from 5-6 sources...
> > + you have repeated the same statement several times, that is 
> > not the best way of convincing people.
> >     
>
> I know you DON'T believe that, as you are about the tenth person to
> repeat that "repeating stuff has no effect."
>   
Why should we change our response to the same error? The only solution 
to this loop is when people stops answering you and you "lose".
> > I believe you picked up the "anti-Reiser religion"-phrase from previous 
> > rant-wars (otherwise, why does that "religion"-phrase always come up, 
> > and (almost) only when dealing with Reiser-fs), and yes, there has been 
> > some clashes caused by both sides, so please be careful when dealing 
> > with this matter.
> >     
>
> NO. You people simply come across as zealots who work together, against
> Reiser4.
>
> Hence the term "anti-Reiser religion."
>   
Please, don't address someone you meet for the first time as "you people"!
Yes, we do _work_ together, it is a community and as a community you 
have to follow the social rules agreed upon. Without all those 
pro-Reiser peoples who knew how to work with the rest, there would not 
be a ResierFS/Reiser3 in the kernel. Unfortunately, Hans is in this case 
his own worst enemy and has ruffed quite a few feathers over the time. I 
don't think you would like someone who tells you "if you do it my way, 
then you are doing it wrong"...

But personally, even if I find Hans a bit too strong-headed, he got some 
interesting design-ideas and the Reiser-filesystem is something I think 
many find interesting as a concept but not yet trust-worthy for their 
own machines.
> > Would you be willing to benchmark Reiser4 with some compressed 
> > binary-blob and show the time as well as the CPU-usage? 
> >     
>
> I might be. I don't really know how to set it all up.
>
> Perhaps if you guided me through it.
>   
Am not sure how much help I would be but from the responses to your 
benchmark-list, there seems to be many who could help you. But first I 
think you should set up a system to test on, and then after some tests 
and made the result public, there will (most likely) be people who ask 
you to test it in some specific way.
> > I may have missed something, but if my room-mate took my harddrive, 
> > screwed it open, wrote a love-letter on the disk with a pencil and then 
> > returned it (ok, there may be some more plausible reasons for 
> > corruption), is the OS really suppose to handle it? 
> >     
>
> Yeah, I can't see how the OS could read the love-letter either.
>
> But one thing is for sure. The FS ain't responsible for reading it.
>   
And no-one has asked the file-system to _read_ the disk, but to be 
designed to help restore the file-structure. This I have found to be the 
main-point people complains about.
It is like arguing against air-bags in a car. Of course the car should 
not be responsible for preventing accidents, but they are designed so 
_if_ it happens, you should not be totally screwed.
> > Yes, it should not 
> > assign any new data to those blocks but should it not also fall into the 
> > file-systems domain to be able to restore some/all data?
> >     
>
> It's a tough ask of any FS. 
>
> Microsoft's filesystem checker totally roasted all my data on an XP-box
> last night. 
>   
Sorry to hear that, but two wrongs does not make it right.

Richard Knutsson

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/