[libvirt] [PATCH 2/3] conf: Clean up nodedev code

[Daniel P. Berrange]

On Thu, Mar 02, 2017 at 10:04:24AM +0100, Michal Privoznik wrote:
> On 03/02/2017 09:58 AM, Peter Krempa wrote:
> > On Wed, Mar 01, 2017 at 19:27:15 -0500, John Ferlan wrote:
> >> Alter the static functions from virNodeDev* to just nodeDev* as a visual
> >> cue to determine which are local or not when reading code.
> >>
> >> Cleanup spacing between functions, function defs, etc. to match more modern
> >> techniques used in libvirt
> >>
> >> Signed-off-by: John Ferlan <jferlan at redhat.com>
> >> ---
> >>  src/conf/node_device_conf.c | 476 +++++++++++++++++++++++---------------------
> >>  src/conf/virnodedeviceobj.c | 128 ++++++------
> >>  2 files changed, 322 insertions(+), 282 deletions(-)
> >>
> >> diff --git a/src/conf/node_device_conf.c b/src/conf/node_device_conf.c
> >> index bc36527..09e815a 100644
> >> --- a/src/conf/node_device_conf.c
> >> +++ b/src/conf/node_device_conf.c
> >> @@ -72,9 +72,9 @@ VIR_ENUM_IMPL(virNodeDevDRM, VIR_NODE_DEV_DRM_LAST,
> >>                "render")
> >>  
> >>  static int
> >> -virNodeDevCapsDefParseString(const char *xpath,
> >> -                             xmlXPathContextPtr ctxt,
> >> -                             char **string)
> >> +nodeDevCapsDefParseString(const char *xpath,
> >> +                          xmlXPathContextPtr ctxt,
> >> +                          char **string)
> > 
> > Please don't remove the vir prefix. The coding style tries to converge
> > to having them everywhere.
> > 
> 
> Why? If a function is static, we can be sure it's not called from
> outside of the file. Moreover, I'd direct your attention to recent
> commit of f557b3351e0b6d for instance. In fact whole qemu driver serves
> as a great example: it's "static int qemuDomain*()" not "static vir
> virQEMUDomain*()".
> 
> In fact, I'd suggest the opposite rule - use "vir" prefix only if
> function is shared between modules. For instance  virFileCopyACLs should
> have the vir prefix because it's exported. virFileRewriteStrHelper
> should not have the prefix because it's static.
> The advantage of this approach is that one can immediately tell just
> from the name if the function is exported or not.

No, this is a bad rule because it causes us to rename code when we
inevitably make static functions non-static. We want a naming rule
that is standardized & stable long term. Convering to use 'vir'
prefix everywhere gives us that.

Regards,
Daniel
-- 
|: http://berrange.com      -o-    http://www.flickr.com/photos/dberrange/ :|
|: http://libvirt.org              -o-             http://virt-manager.org :|
|: http://entangle-photo.org       -o-    http://search.cpan.org/~danberr/ :|