Subject: Re: CVS commit: src/sys/arch/x86/x86
To: None <source-changes@netbsd.org>
From: Christos Zoulas <christos@astron.com>
List: source-changes
Date: 08/15/2006 23:09:47
In article <20060816000426.I28872@cargo-cult.k.bsd.de>,
Christoph Badura <bad@bsd.de> wrote:
>On Tue, Aug 15, 2006 at 02:03:09PM -0700, Jason Thorpe wrote:
>> Yes, I realize that's what they're for. However, it seems to me that
>> they just don't belong there. It's already screwy having the fields
>> change meanings for FFS vs LFS.
>
>That is like saying that deriving multiple classes from a base class
>is "screwy" because the implementation stores different information
>at the same offset in the objects storage.
Most languages don't do this... They append the derived class' data.
>Now, there may be other pressing reasons why we want to give up this feature.
>However, I haven't seen them yet.
There is another use of the fields: fsck uses them to compute the prototype
superblock. But it does not seem to break anything when you get rid of them.
christos