Hi, On Thu, Apr 21, 2022 at 03:13:05PM +0200, Peter Krempa wrote:
On Wed, Apr 20, 2022 at 21:07:57 +0200, Victor Toso wrote:
Either create or append to existing docstring, the version (git tag) that a given macro was introduced in the format:
Since: v1.2.3
Signed-off-by: Victor Toso <victortoso@redhat.com> --- include/libvirt/libvirt-domain.h | 663 +++++++++++++++++++++++++++++- include/libvirt/libvirt-host.h | 123 ++++++ include/libvirt/libvirt-network.h | 24 ++ include/libvirt/libvirt-nodedev.h | 3 + include/libvirt/libvirt-secret.h | 3 + include/libvirt/libvirt-storage.h | 3 + 6 files changed, 810 insertions(+), 9 deletions(-)
While based on the method some of these may be inaccurate it's better than what we had before.
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
Yes, I think it is clear that false positives will exist. Hopefully, they are not too far of and shouldn't be a real issue for software that will rely on this metadata. A quick check on the versions provided in the xml now: * Exported types per major release Major # % v0 1254 57.3% v1 479 21.9% v2 89 4.0% v3 100 4.5% v4 37 1.6% v5 108 4.9% v6 47 2.1% v7 52 2.3% v8 21 0.9% Total 2187 100.0% (~) * Number of types at the moment Type # Enum 1028 47.00% Function 553 25.28% Functype 53 2.42% Macro 283 12.94% Struct 50 2.28% Typedef 219 10.01% Variable 1 0.04% Total 2187 100.00% (~) I'd guess that, if problems arise, they would come from v5 onwards and that's about 10% of all exported types. Cheers, Victor