[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help
From: |
Anthony Liguori |
Subject: |
Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help |
Date: |
Mon, 26 Jul 2010 14:35:42 -0500 |
User-agent: |
Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.1.10) Gecko/20100528 Lightning/1.0b1 Thunderbird/3.0.5 |
On 07/26/2010 02:19 PM, Avi Kivity wrote:
Is what we are supporting just what libvirt expects there to be or
what any tool out there expects there to be?
We should try to support all users, prioritized by the number of end
users they represent. If this patch broke some other large user we'd
be in a bind. But likely this isn't the case so we aren't.
As I've said, I'm pragmatic and that's why I've argued for these changes
in the past. But libvirt should have changed a long time ago to using
something more reliable (like version).
It's not. Our help output is unreadable. The (artificial)
restrictions we're putting ourselves with respect to the help output
prevents it from being improved.
Are you saying
" [,cache=writethrough|writeback|unsafe|none][,format=f]\n"
is more readable than
" [,cache=writethrough|writeback|none|unsafe][,format=f]\n"
? if so I disagree.
Absolutely. But I meant in the general sense.
Backports are such a special case. Honestly, we're talking about
RHEL and it's trivially easy for libvirt to just special case RHEL.
As it happens the patch came from a Novell employee, so it isn't about
RHEL.
SLES doesn't carry many backported features. The reason this hit Bruce
is that libvirt is detecting using help because of RHEL.
I applaud your choice of enterprise operating systems, however.
Regardless, outside of Windows users qemu will mostly be consumed via
distribution branches, with different levels of backport happiness.
We should recognize that and work with it, not against it.
And it's trivially easy for libvirt to deal with this. They simply have
to do: `cat /etc/redhat-release` or `cat /etc/suse-release` and adjust
version logic accordingly.
It's an easy change for libvirt to make, and the problem goes away for
the future. If such a change was made, I'd be inclined to take a patch
like this now to make up for the difference.
But as I said, we've been accommodating libvirt's use of help for a long
time now. We shouldn't let perfect (omnipotent capabilities system)
stand in the way of good (version/feature matrix).
I don't see what we gain by not doing this.
We're losing the ability to make *any* change to our help system by
encouraging it to be used in this fashion.
If we could point libvirt towards a better way of doing things (not a
better regexp, which could just break under different circumstances; a
supported interface) I'd agree. Go RTFM chapter-this-or-that and
don't bother me no more. But we can't.
Again, give me an example of where versioning isn't simple and robust
that applies to things as they stand today and I'm happy to agree with you.
If you want libvirt to do the right thing, provide a proper
capabilities interface. Using the version has its downsides as much
as the help text.
That's simply not the case. Please, provide an actual example where
version is not reliable and backports aren't trivially easy to detect.
t=0 starting point, cache=unsafe is unknown
t=1 qemu upstream adds cache=unknown
t=2 libvirt adds support for cache=unsafe, releases
t=3 evil distro backports cache=unsafe, releases qemu-kvm-1.2.3.4
t=4 user tries libvirt from t=2 with qemu from t=3, cache=unsafe not
detected
version numbers force a libvirt update every time a feature is
backported.
That's a bogus scenario because said evil distro would also enhance
libvirt to detect the feature properly.
Regards,
Anthony Liguori
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, (continued)
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Markus Armbruster, 2010/07/22
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Daniel P. Berrange, 2010/07/22
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Anthony Liguori, 2010/07/22
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Markus Armbruster, 2010/07/23
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Anthony Liguori, 2010/07/26
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Avi Kivity, 2010/07/26
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Anthony Liguori, 2010/07/26
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Avi Kivity, 2010/07/26
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Anthony Liguori, 2010/07/26
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Avi Kivity, 2010/07/26
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help,
Anthony Liguori <=
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Avi Kivity, 2010/07/26
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Anthony Liguori, 2010/07/26
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Markus Armbruster, 2010/07/27
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Jes Sorensen, 2010/07/27
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Cole Robinson, 2010/07/27
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Anthony Liguori, 2010/07/27
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Cole Robinson, 2010/07/27
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Markus Armbruster, 2010/07/27
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Avi Kivity, 2010/07/26
- Re: [Qemu-devel] [PATCH] move 'unsafe' to end of caching modes in help, Anthony Liguori, 2010/07/26