The type parameter is not needed anymore because the service type
can be extracted by only using the unique vmid.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>

This allows the Backup to work with ExtJS5

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>

As the size parameter is now deprecated and was replaced with the
new rootfs parameter, use that one when creating an container.
So we don't get an 'parameter verification failed' error.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>

this is necessary because we use image files for both.

The original fix using a config Object was a bit overkill, this works well too , requires less code in the child classes, and is more consistent with rest of the code we have.

We already did this for ExtJS4, just copying from there

It is currently unclear howto restrict storages for container usage.
The 'rootdir' property does not work anymore, because we can now create
containers on normal VM image files.

Ext.applyIf(me, {
	    title: gettext('Search')
});
in initComponent() is not safe anymore with ExtJS5.
Since the 'title' property is existing at that stage (even if with a null or undefined value), Ext.applyIf will not update it.

ExtJS refuses with version to have an id set to an empty string, hence we can't
use our "key" property as a custom idProperty (an empty string for key is ok for us, and is used to set back a PVE property to its default value)

We always access the KeyValues entities via their Key property, so this change should be safe.

Since ExtJS5, when doing such a declaration,

	Ext.applyIf(me, {
	    items: {
		xtype: 'pveKVComboBox',
		comboItems: data,
	    }
	});

the comboItems property is passed in the parent class as a config object,
instead of being directly available via this.comboItems

Since using a config object is the recommended way of passing parameters
int the ExtJS class model, adapt the parent class to use the new paradigm.

in the pvesm and pvesh we are giving the correct size of the used space from a disk.
this change make the output of Gui and CLI consistent.

Fixes #681

The previous code stripped *all* combinations of zero or
more letters and digits followed by a dot of an interface
name to retrieve the VLANID. At the same time the supposedly
opposite code for IF_VLAN_RAW_DEVICE only actually extracts
the part up to the *first* dot. Thus an interface named
'a.b.0' would have a VLANID of 0 and a raw device name of
just 'a'.
I also don't see a reason to limit the removed portions to
alphanumeric characters. I.e. an interface named tap-test.0
would result in a VLANID of 'tap-0' and an
IF_VLAN_RAW_DEVICE of 'test'.

A simple shell substitution seems to do a much better job
and is more efficient than forking out two processes with a
pipe.

if the packet on the opt list is not installed, it will not shown.