qemu-devel
[Top][All Lists]
Advanced

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

Re: [Qemu-devel] [PATCH] Add HTTP protocol using curl v2


From: Daniel P. Berrange
Subject: Re: [Qemu-devel] [PATCH] Add HTTP protocol using curl v2
Date: Wed, 6 May 2009 14:39:20 +0100
User-agent: Mutt/1.4.1i

On Wed, May 06, 2009 at 08:12:56AM -0500, Anthony Liguori wrote:
> Avi Kivity wrote:
> >Daniel P. Berrange wrote:
> >>Then I'd prefer we add a protocol=XXX option for magic protocols. 
> >>This would be easier to use & clearer than requiring escaping of 
> >>magic characters, eg
> >>
> >>   -drive file=/some/path:with:colons,protocol=file
> >>   -drive file=http://some/path,protocol=uri
> >>
> >>For compatability, we could make it such that if protocol=XXX was 
> >>left out,
> >>it could try and "guess" it, in same way QEMU does if format=XXX is 
> >>left out
> >>for content format.
> >>  
> >
> >How about
> >
> > -drive file=file:///some/path:with:colons?
> >
> >Libvirt would use this unconditionally, command-line users can choose.
> 
> Why does libvirt care?  That confuses me.

I don't have any problem with QEMU supporting http, nbd, or other clever
file access schemes. I'll let others debate its merits vs using FUSE 
http filesystems :-)

All I care about from libvirt POV, is that there is a way to give QEMU an 
absolute file path for a disk, and guarentee that QEMU will treat this as
a local file path, and not try any access protocols, other than 'open(2)'.
Various options from this thread....

 - Treat any path starting with / as local file
 - Allow file:  as a prefix
 - Allow file:/// as a prefix, real URI style
 - Add a protocol=file  flag to -drive

Pick one, pick several, suggest more. Any of these options would work as 
far as I'm concerned, and we could easily support several. I think the
first is desirable because that's the natural thing users will try when
launching QEMU directly. I see the merit in also having file: or file:///
as an explicit protocol too. I think the 4th protocol=file is redundant
really, given that we have precedent of using XXX: prefixes in nbd driver

Regards,
Daniel
-- 
|: Red Hat, Engineering, London   -o-   http://people.redhat.com/berrange/ :|
|: http://libvirt.org  -o-  http://virt-manager.org  -o-  http://ovirt.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505  -o-  F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|




reply via email to

[Prev in Thread] Current Thread [Next in Thread]