Ordering of nodes

[tbsuht]

Hi,

are there any guarantees about existing node ordering when the YANG schema changes (adding an augmentation in this case)?

In my case I have a set of YANG schemas loaded into the context. When traversing the tree I have a given fixed ordering.
After loading another augmentation into the context, I can observe that some nodes in the tree change their ordering compared to before loading the augmentation.

Thanks!

[michalvasko]

Could you please attach the YANG modules you are using and the exact observed and expected behavior?

[tbsuht]

See the attached reproducer, let me know if something is not working. I wasn't able to shrink the reproducer down, so there are quite a few models involved.

The output of the program should give an impression of what is expected.

You can start the reproducer by calling libyang-reorder-reproducer $path-to-models/

repro.zip

XPath: /ietf-interfaces:interfaces/interface/ethernet
Before children():
  1. auto-negotiation
  2. duplex
  3. speed
  4. flow-control
  5. max-frame-length
  6. mac-control-extension-control
  7. frame-limit-slow-protocol
  8. capabilities
  9. statistics
  10. auto-configuration
  11. poe
  12. mirroring-sink
  13. base-t1
After children():
  1. auto-negotiation
  2. duplex
  3. speed
  4. flow-control
  5. max-frame-length
  6. mac-control-extension-control
  7. frame-limit-slow-protocol
  8. capabilities
  9. statistics
  10. base-t1
  11. auto-configuration
  12. poe
  13. mirroring-sink

E.g. base-t1 is reordered although the newly added model should not influence the /ethernet nodes?

In general my wish would be that an augmentation only inserts nodes but does not reorder existing nodes.

[michalvasko]

Right, there was a non-determinism causing the augments of a module to be applied in a different order. Should be fixed in the latest devel.