Re: Journalling filesystems

[Ognyan Kulev]

Bas Wijnen wrote:
> On Mon, Jun 14, 2004 at 07:24:25PM +0200, Alfred M. Szmidt wrote:
>
> >   I think they should be implemented in the Hurd.
> >
> > Would you like todo this?
>
> Yes, I would.  But as Ognyan says:
> But we are far away from all that, and IMHO we should first finish
> porting to L4 before dealing with such stuff.

My opinion shouldn't stop you.  We all are happy when somebody improves 
the Hurd.

When I work on ext2 or ext3, bugs in Hurd/GNUMach often slow me down, 
sometimes a lot.  What I want to say is that stable foundation is much 
more important than fancy stuff, because fancy stuff is developed very 
hard when foundation is not reliable.

Of course, GNU Mach is OK for doing such transaction things.  "Porting 
to L4" is something like symbol of reaching stable foundation, again, in 
my opinion, which is highly influenced by bad practical experience with 
the Hurd and GNU Mach.

> I just wanted to mention it at this early stage, because it may be very hard
> to implement it later if it wasn't considered as a posibility.

I don't think there is great difference in effort to implement this now 
and, say, L4 port completion (which itself will change the Hurd, 
probably a lot, I'm not tracking L4 efforts closely).

In conclusion, get your hands dirty by working on what you want, and 
you'll better understand what I and Alfred are saying.

Regards,
ogi