From: Herbert Poetzl (herbert_at_13thfloor.at)
Date: Sat 27 Mar 2004 - 12:53:51 GMT
On Sat, Mar 27, 2004 at 10:49:22AM +0100, Georges Toth wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> > > I enabled the block device option using menuconfig.
> > > Do I need to
> > > 1.) Install any additional patches for these quota items,
> > > if I am using 1.26?
> > if you want per vserver disk limits and quotas on a
> > shared partition (more than one vserver per partition)
> > then you need to add the quota patches ...
> isn't it possible that you plz include those patches in
> your releases ?
yes, it would be 'possible', and depending on the way I'd
include them, the implications would be:
a) include them unconditional
- if somebody doesn't want 'per context' stuff, just
select context tagging none, and add a context hash
- normal quota won't work anymore, but who cares,
those who want to use that feature do not need to
add a simple patch anymore ...
- enlarges the patch by ~120k
b) include them #ifdefed
- there is a 'master switch' turning on/off the
quota hashes, context tagging
- you have to 'choose' between normal and context
quota as an kernel option (at build time)
- would enlarge the patch by ~250k
> if you consider them too experimental (allthough i thought
> they were stable?) you could add a note next to the option
> (in kernel config) that that option is experimental...
this is not the reason, it's just that many people do not
use quota or context quota (I'd say 80% or more), and why
should they be forced to download and apply a patch, which
is about twice as large as necessary, with the following
- there might be a critical bug in the quota code
- some kernel change might harm the 'old' quota code
> what do you think ?
so I'm not convinced that 'including' a perfectly separable
patch unconditionally into another separate one is such
a good idea, btw, the following command should help you here
# cat patch1.diff patch2.diff >patch-all.diff
> - --
> Georges Toth
Vserver mailing list