[Libvirt-cim] [PATCH] [TEST]Fix HostSystem-01_enum.py with sblim cmpi base provider installed, add Linux_ComputerSystem to enumclass

[Guo Lian Yun]

libvirt-cim-bounces at redhat.com wrote on 2008-09-23 08:46:00:

> > @@ -43,27 +44,35 @@ def main():
> > 
> >      status = FAIL
> >      keys = ['Name', 'CreationClassName']
> > -    try:
> > -        hs = enumclass.enumerate(options.ip, 'HostSystem', keys, 
> options.virt)
> > -        name = get_typed_class(options.virt, 'HostSystem')
> > +
> > +    linux_cs = enumclass.enumerate(options.ip, 'ComputerSystem', 
> keys, 'Linux')
> 
> I think this would be a good chance to fix the behavior of the 
> enumerate() function.  Instead of passing in both the base name and the 
> virtualization type, the test itself should call get_typed_class() to 
> get the proper classname.  That way, this function only needs to take a 
> classname param.

  Good idea. But a lot of tests have to update with the behavior of 
enumerate() function fix.
  At now, I put fixing HostSystem-01_enum.py with sblim cmpi base provider 
installed in high
  priority, so I sill use current enumereate() in my patch. Once it 
applied, I will start 
  to cook up new enumerate() function and update all related tests.

  Thanks!
 
> 
> You'll also need to set the namespace appropriately before calling this 
> function.  Since you'll need to do this in several places, you could 
> create a function that checks whether the system has the SBLIM providers 

> or not. A function that returns a True if the enum of Linux_CS returns 
> an instance or returns False if Linux_CS returns an error / 0 instances.
> 
> > +    hs = enumclass.enumerate(options.ip, 'HostSystem', keys, 
options.virt)
> > +    if len(linux_cs) == 1 and len(hs) == 0:
> > +        return PASS
> > +    if len(linux_cs) == 0 and len(hs) == 0:
> 
> A debug statement here would be good.
> 
> > +        return XFAIL_RC(bug)
> > +    elif len(linux_cs) == 0 and len(hs)== 1: 
> 
> What about the case where len(linux_cs) == 1 and len(hs) == 1?  This 
> scenario should be an error.
> 
> So fixing this if / else block would be good.  You could have something 
> like:
> 
> ret = using_sblim()
> if ret:
>      #verify HostSystem enum didn't return an instance and return 
> accordingly
> 
> try:
>     #rest of the test here
> 
> > +        try:
> > +            hs = enumclass.enumerate(options.ip, 'HostSystem', 
> keys, options.virt)
> 
> This is no longer needed - you already called enumerate() above.
> 
> > 
> > diff -r 892ce3fce234 -r 83e6f32be2ae suites/libvirt-
> cim/lib/XenKvmLib/classes.py
> > --- a/suites/libvirt-cim/lib/XenKvmLib/classes.py   Fri Sep 12 14:
> 35:12 2008 -0700
> > +++ b/suites/libvirt-cim/lib/XenKvmLib/classes.py   Sun Sep 21 22:
> 54:28 2008 -0700
> > @@ -23,6 +23,8 @@ virt_types = ['Xen', 'KVM', 'XenFV', 'LX
> > 
> >  def get_typed_class(virt, basename):
> >      if virt not in virt_types:
> 
> I'd remove this if statement.
> 
> > +        if virt == "Linux" and basename == "ComputerSystem":
> > +            return 'Linux_ComputerSystem' 
> 
> Which would make this part unnecessary.
> 
> >          if virt != "Virt" and basename != "MigrationJob":
> >              raise ValueError('Invalid class type')
> 
> This is obsolete now - so we need a separate patch to remove this piece.
> 
 
  Follow patch I will sent out later. 

> > 
> > diff -r 892ce3fce234 -r 83e6f32be2ae suites/libvirt-
> cim/lib/XenKvmLib/enumclass.py
> > --- a/suites/libvirt-cim/lib/XenKvmLib/enumclass.py   Fri Sep 12 
> 14:35:12 2008 -0700
> > +++ b/suites/libvirt-cim/lib/XenKvmLib/enumclass.py   Sun Sep 21 
> 22:54:28 2008 -0700
> > @@ -68,6 +68,9 @@ class CIM_ComputerSystem(CIM_MyClass):
> >      pass
> > 
> >  class CIM_System(CIM_MyClass):
> > +    pass
> > +
> > +class Linux_ComputerSystem(CIM_MyClass):
> >      pass
> 
> The MyClass extension is very useful, but I concerned that the list will 

> just continue to grow. I don't think it's very scalable at this point. 
> and I'd like to keep the amount of SBLIM specific bits to a minimum.
> 
> This test case doesn't make use of the CIM_MyClass pieces from what I 
> can tell.  So I think it's safe to remove this.
> 
> -- 
> Kaitlin Rupert
> IBM Linux Technology Center
> kaitlin at linux.vnet.ibm.com
> 
> _______________________________________________
> Libvirt-cim mailing list
> Libvirt-cim at redhat.com
> https://www.redhat.com/mailman/listinfo/libvirt-cim
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/libvirt-cim/attachments/20080923/b9e2b514/attachment.htm>