[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: ext3 0.9.12 for 2.4.10-ac11



Hi,

On Thu, Oct 11, 2001 at 02:35:50PM +0200, Tim Tassonis wrote:

> > Personally, I believe that panicing kernels cause more hassle for
> > users!
> 
> You are certainly right, but that wasn't the point. The point is your
> concentration on the -ac tree, which is not the official tree.

What?  *What* concentration on the -ac tree?  We maintain both.  For a
long time the primary ext3 patches were on the linus-tree, not the -ac
one.  The support of the -ac tree is a recent addon, it's not where we
concentrate.

However, right now the linus tree is unsettled, and ext3 on the linus
tree is *known* to have been broken by this, and we're still testing
the fixes.  Ext3 on -ac is still thought to be rock solid right now.
So until we've done more testing on the linus tree, the main supported
ext3 patches will be on -ac, not on linus's tree.

That's a recent change, and it's short term situation while linus's
tree is in flux.

Your claim that we concentrate on the -ac tree is false, and your
assertion that it's somehow due to both Alan and myself working for
the same company is bizarre.

> If course
> it is your decision which kernel to follow primarily, you wrote the code
> and I am only a non-contributing user of your stuff. I run 2.4.10 with
> ext3 since it came out and did not have one panic.

You have been lucky --- fine.  I'm a little more cautious, and I don't
like releasing supported patches to end-users when I know for certain
that they contain regressions.  The e2fsprogs tools are totally broken
on 2.4.10, for a start.

> > > I guess it's a some kind of political thing (Red
> > > Hat using -ac kernels and ext3 development payed by Red Hat?).
> > 
> > That's just insulting speculation.  We have kept the -ac and Linus-
> > versions of ext3 both uptodate until now, and in fact until fairly
> > recently the linus-tree was the primary focus of development.
> 
> Maybe, insulting, but rather obvious, don't you think?

Huh?  It's plain false.  We tracked the linus tree primarily while it
was the most stable tree, and only recently switched to -ac for
stability's sake when the linus version started getting serious
internal rewriting.  I don't know how you can read anything political
into this.

> Well, I don't get any. I guess I wouldn't get any with Alan's tree, since
> I don't run VM benchmarks very often, just normal work.

Simply running tune2fs on a mounted filesystem will be enough to
encounter one of the 2.4.10 breakages.  Converting a mounted ext2
filesystem to ext3 is impossible on 2.4.10, and setting other fs
parameters with tune2fs will silently fail too.

Cheers,
 Stephen





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]