Part of extending the domain XML to show an entire backing chain
is splitting the RelaxNG definitions so that the portion related
to a single source within the chain can easily be reused.
I'm still debating about the way to handle storage volumes using
<format type='qcow2'/> vs. domain disks using <driver
name='qcow2'
type='qcow2'/>. The majority of the <driver> attributes are
associated with the one disk device presented to the guest, rather
than attributes for a file in the chain. So I'm thinking of
having domain disks add a new <format> element; where we output
the format under both elements but accept either spelling for
new input (back-compat reasons require that we always provide
the old spelling).
Eric Blake (6):
conf: create common storage RNG grammar file
conf: better <disk> interleaving in schema
conf: tighten <domainsnapshot> schema
conf: split <disk> schema into more pieces
conf: move storage source details to common RNG file
conf: move <auth> and <encryption> to disk source
docs/schemas/Makefile.am | 5 +-
docs/schemas/domain.rng | 15 +-
docs/schemas/domaincommon.rng | 268 ++----------------
docs/schemas/domainsnapshot.rng | 85 +++---
docs/schemas/storagecommon.rng | 312 +++++++++++++++++++++
docs/schemas/storageencryption.rng | 33 ---
docs/schemas/storagefilefeatures.rng | 24 --
docs/schemas/storagevol.rng | 3 +-
.../qemuxml2argv-disk-drive-discard.xml | 8 +-
.../qemuxml2argv-disk-source-pool.xml | 12 +-
.../qemuxml2xmlout-disk-drive-discard.xml | 37 +++
.../qemuxml2xmlout-disk-source-pool.xml | 44 +++
tests/qemuxml2xmltest.c | 4 +-
13 files changed, 477 insertions(+), 373 deletions(-)
create mode 100644 docs/schemas/storagecommon.rng
delete mode 100644 docs/schemas/storageencryption.rng
delete mode 100644 docs/schemas/storagefilefeatures.rng
create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-disk-drive-discard.xml
create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-disk-source-pool.xml
--
1.9.0