=== dmsimard|afk is now known as dmsimard
[01:05] <harlowja_> smoser rharper if u get some time https://gist.github.com/harlowja/d63a36de0b405d83be9bd3222a5454a7
[01:06] <harlowja_> https://gist.github.com/harlowja/d63a36de0b405d83be9bd3222a5454a7#file-gistfile1-txt-L271 being the main complexity
[01:06] <harlowja_> trying to understand the 3 different network configs that get mutated and passed around makes this a PITA, lol
[01:06] <harlowja_> openstack network data --> some intermediate format --> some other final format
[01:10] <harlowja_> == pain
[02:06] <rharper> harlowja_: nice;  IMO, the complexity comes from having to understand the mapping between underlying devices and the upper layers;  I've not yet looked at the code in ifupdown2; but that cloudious ifupdown2 written in python claims to have a model that uses a graph for dependencies between interfaces;  they might have a cleaner internal state representation that we could map the various inputs (network-config.yaml, netw
[02:06] <rharper> ork_data.json) and then it certainly handles writing out eni since it's an ifupdown replacement;  and I don't think networkd will be that hard after handling eni and sysconfig