Re: [libvirt] [PATCH 1/8] latency: Define new public API and structure

[Daniel Veillard <veillard redhat com>]

On Fri, Sep 02, 2011 at 08:05:14AM -0600, Eric Blake wrote:
> On 09/02/2011 04:06 AM, Daniel Veillard wrote:
> >On Wed, Aug 31, 2011 at 04:26:06PM +0800, Osier Yang wrote:
> >>---
> >>  include/libvirt/libvirt.h.in |  111 ++++++++++++++++++++++++++++++++++++++++++
> >>  src/libvirt_public.syms      |    5 ++
> >>  2 files changed, 116 insertions(+), 0 deletions(-)
> >>
> >>+/**
> >>+ * virDomainBlockStatsFlagsStruct:
> >>+ *
> >>+ * Struct filled by virDomainBlockStatsFlags() providing information
> >>+ * about the block device.
> >>+ *
> >>+ * Hypervisors may return a field set to ((long long)-1) which indicates
> >>+ * that the hypervisor does not support that statistic.
> >>+ *
> >>+ * NB. Here 'long long' means 64 bit integer.
> >>+ */
> >>+typedef struct _virDomainBlockStatsFlags virDomainBlockStatsFlagsStruct;
> >>+
> >>+struct _virDomainBlockStatsFlags {
> >>+    char field[VIR_DOMAIN_BLOCK_STATS_FIELD_LENGTH];
> >>+    long long value;
> >>+};
> 
> Are we positive that all useful block stats will always be in long
> long format, or should we reuse virTypedParameter here to also allow
> other typed objects in the future?

  Ouch, very good point ! We can't guess so we should not take the risk !
we really ought to align to the other APIs and use virTypedParameter

Daniel

-- 
Daniel Veillard      | libxml Gnome XML XSLT toolkit  http://xmlsoft.org/
daniel veillard com  | Rpmfind RPM search engine http://rpmfind.net/
http://veillard.com/ | virtualization library  http://libvirt.org/