GFS removed??? (was: rawhide report: 20050315 changes)

[Dave Jones]

On Wed, Mar 16, 2005 at 09:13:13PM +0100, Axel Thimm wrote:
 > On Wed, Mar 16, 2005 at 03:00:21PM -0500, Elliot Lee wrote:
 > > On Wed, 16 Mar 2005, Axel Thimm wrote:
 > > 
 > > > On Tue, Mar 15, 2005 at 07:57:47AM -0500, Build System wrote:
 > > > > Removed package magma
 > > > > Removed package ccs
 > > > > Removed package GFS
 > > > > Removed package magma-plugins
 > > > > Removed package iddev
 > > > > Removed package gulm
 > > > > Removed package fence
 > > > > Removed package rgmanager
 > > > > Removed package ccs
 > > > > Removed package magma
 > > > 
 > > > Why?
 > > 
 > > Because the kernel support is not included yet. The maintainer of these
 > > packages asked for them to be removed until that happens.
 > 
 > All they needed as a mandatory kernel patch was the distributed flock,
 > which was there since 2.6.9 or so.
 > 
 > Or do you mean that the GFS' kernel support (which was in these
 > packages) is going to become a patch to the kernel? IMHO it was
 > cleaner as an external kernel module solution and would also give a
 > nicer layering idiom to RHEL4 w/o having to replace the kernel.

FWIW, I'd rather see this done as an external module also.
If for no other reason than as a demonstration that external modules
can be packaged correctly. If they can't, lets fix that.
Rik van Riel has been picking up the slack on kernel-devel the last
few days, as I've been swamped with other stuff.

Dropping the module seems to imply I'm about to get a patch
adding GFS to the rawhide kernel.  I'm not thrilled by that idea at all.
Being part of the kernel tree means that if I rebase to a -bk snapshot
that makes it not compile, it gets dropped until it gets fixed, so
being merged there is no better off than the current situation
of needing rebuilding each time theres a new kernel.

Unrelated sidenote: I'm also less than excited about including stuff in
our kernel (even though it is shiny new Red Hat toys) that hasn't had any
upstream review yet.  When is this getting proposed for inclusion in -mm ?

Lets fix this problem properly.

		Dave