[libvirt] [Xen-devel] [PATCH 00/12] Replace Xen xl parsing/formatting impl

[Ian Campbell]

On Mon, 2015-01-12 at 09:23 -0700, Jim Fehlig wrote:
> Ian Campbell wrote:
> > On Fri, 2015-01-09 at 22:03 -0700, Jim Fehlig wrote:
> >   
> >> The first attempt to implement support for parsing/formatting Xen's
> >> xl disk config format copied Xen's flex-based parser into libvirt, which
> >> has proved to be challenging in the context of autotools.  But as it turns
> >> out, Xen provides an interface to the parser via libxlutil.
> >>
> >> This series reverts the first attempt, along with subsequent attempts to
> >> fix it, and replaces it with an implementation based on libxlutil.  The
> >> first nine patches revert the original implementation and subsequent fixes.
> >> Patch 10 provides an implemenation based on libxlutil.  Patches 11 and
> >> 12 are basically unchanged from patches 3 and 4 in the first attempt.
> >>
> >> One upshot of using libxlutil instead of copying the flex source is
> >> removing the potential for source divergence.
> >>     
> >
> > Thanks for doing this, looks good to me, FWIW.
> >
> > Is the presence/absence of xen-xl support exposed via virsh anywhere? If
> > so then I can arrange for my Xen osstest patches for libvirt testing to
> > use xen-xl when available but still fallback to xen-xm. I've had a look
> > in "virsh capabilities" and "virsh help domxml-from-native" but not
> > seeing xen-xm, so assuming xen-xl won't magically appear in any of those
> > places either.
> >   
> 
> AFAIK, the only place the supported native formats are listed is in the
> virsh man page.

Not to worry, I think I'll just use xen-xl everywhere then, osstest's
handling of test failures and regression detection will do the right
thing with versions of libvirt which don't have this applied.

>   But thanks for the question, else I would have missed
> adding xen-xl to the man page in 12/12.

No problem ;-)

Ian.