[02:24] <thumper> axw: got some time?
[02:24] <thumper> axw: I need to talk through some crazy
[02:24] <axw> thumper: sure
[02:25] <thumper> axw: https://hangouts.google.com/hangouts/_/canonical.com/broken-magic
[02:54] <thumper> axw: you might be right on that stashing thing
[02:54] <axw> thumper: what did you find?
[02:54] <axw> I haven't found anything yet...
[02:54] <thumper> I just ran again on lxd
[02:55] <thumper> and before the machine could come up and write in, both had been set
[02:55] <thumper> so something else must cut in for it to operate differently
[02:55] <thumper> it is just a deduction thing,
[02:55] <thumper> there must be something
[02:55] <axw> thumper: can you pastebin that output for me?
[02:56] <axw> the ops you showed me before
[02:56] <thumper> axw: https://pastebin.canonical.com/206750/
[02:56] <thumper> oh... yeah
[02:56] <axw> ta
[02:57]  * thumper gets them
[02:57] <thumper> https://pastebin.canonical.com/206751/
[03:14] <axw> thumper: yup, the "u" field gets decoded as a bson.M. https://paste.ubuntu.com/26323241/
[03:14] <thumper> wow
[03:14] <thumper> the stash'll do it then
[03:14] <axw> it's a wonder it ever worked
[03:14] <thumper> yeah
[03:15] <thumper> probably a "feature"
[03:15] <thumper> for the bson.D that is
[03:18] <thumper> axw: pretty weird though that what it said it did, and what it actually did were still different.
[03:20] <thumper> axw: https://github.com/juju/juju/pull/8259
[03:56] <thumper> axw: mgo.v2/bson/encode.go:219
[03:56] <thumper> slices of DocElem types are treated like key, value pairs of a dict
[03:57] <thumper> then bson.decode gets the bson.M back out
[03:57] <thumper> anyway, patch landed
[03:57] <thumper> I'm at EOW
[03:57] <thumper> have a good one
[03:57] <axw> thumper: thanks, you too