Hello David Sommerseth,

On Wed, Apr 03, 2013 at 06:02:07PM +0200, David Sommerseth wrote:
> On 27/03/13 16:25, Florian La Roche wrote:
> >> And yeah,  as I thought given TUV stance on things:
> > 
> > Some sort of side-repo testing rpm fo the newest e2fsprogs-release
> > would be cool. There have been so many bug-fixes and actually so many
> > performance-fixes that I wondered if there is enough interest for a few
> > people who admin "real big filesystems" to start using it...
> 
> Performance fixes are generally hitting the kernel, at least that's
> where the performance will really hit all users.  And those will
> normally be added to the kernel, not e2fsprogs.

I was talking about e2fsck and co, so the user spcae tools.

> 
> Otherwise, all important fixes to ext4 in the kernel or e2fsprogs will
> be backported into the TUV's packages if deemed needed and the risk for
> data loss is considered low enough.  But if any patch may increase the
> risk of data or performance loss, they will not be part of an updated
> package.
> 
> But that's why we all love Enterprise Linux, right?  We've been given a
> rock solid and stable OS, which will stay so for many years forward -
> due to this conservatism and thorough testing done by TUV.

Red Hat is perfectly ok and does this conservative and very good decision
for RHEL and co.

I am talking about updating this one rpm to current upstream versions
for these very limited number of people who want to try out such an update.
For these, having one such binary and a common testing rpm plus sharing
information about it would add some value.

This is the same as CentOS providing en extra side kernel or elrepo extending
this even further to current upstream kernels for CentOS. It's the ecosystem
around RHEL and ScientificLinux/CentOS.

best regards,

Florian La Roche