[05:31] <jtv> bigjools: do we need a big flashing “confined to the cluster controller” warning on the registries?
[05:35] <bigjools> jtv: we need a build error
[05:35] <bigjools> to *prevent* it
[05:35] <bigjools> it's too easy to import
[05:36] <bigjools> I have half a mind to set up the import fascist thingy and prevent imports from provisioningserver inside maasserver
[05:39] <jtv> So far we haven't  needed it simply because Django fails when you try to import maasserver code into provisioningserver.
[05:40] <jtv> I'd be tempted to say "prefix the name with an underscore," but the problem there is that export happens at several levels.
[05:41] <jtv> The idea that a bit on a symbol's name can say whether or not it's imported is an oversimplification...
[05:41] <jtv> *exported
[05:43] <bigjools> jtv: this import problem is the other way around
[05:43] <jtv> Yes.  What I'm saying is: I'd love to have something like the import fascist.
[05:43] <bigjools> ok. we are in violent agreement
[05:43] <jtv> Although we've been getting by without it through a form of luck.
[05:44] <bigjools> I'm not convinced the luck has held out everywhere, I suspect a hand grenade awaits somewhere
[05:45] <jtv> Quite possible.
[05:54] <jtv> Why can't Chrome make up its mind about whether to present a "link" context menu or a "page" context menu on a link?  It's probably subjective, but the browser's quality seems to have slipped dramatically of late.
[05:55] <bigjools> yeah it feels like it
[05:56] <jtv> It's also been failing to update the mouse cursor properly as the page loads.
[06:25] <bigjools> jtv: are you looking at the add-osystem-to-node branch?
[06:25] <jtv> Sort of.
[06:25] <jtv> My thoughts were veering towards food though.  :)
[06:26] <jtv> Ah, I was having a look at osystem-preseed-cleanup...  Not too happy about the "check if object has method, and if so, execute it."
[06:28] <jtv> I'll write that up now.
[06:30] <bigjools> ok
[06:31] <bigjools> this falls into your KISS comments
[06:34] <jtv> Yes, it looks like a sign that that class hierarchy really is needed — assuming templating isn't enough to do the job.
[06:54] <bigjools> jtv: hey can you do a backport to 1.5 of that api fix please?
[06:55] <jtv> Sure.
[06:55] <bigjools> I added the bugtask
[06:55] <jtv> Client, by the way.
[06:55] <bigjools> yeah, what I meant
[06:55] <jtv> On it.
[06:58] <rvba> jtv: thanks for fixing the API bug (1313556)!
[06:58] <jtv> Hi rvba
[06:58] <jtv> Took me insanely long to figure it out, given how easy it actually was to fix...
[06:58] <jtv> Wow, tons o'lint in trunk.
[06:59] <jtv> Oh, it's 1.5.
[11:48] <gmb> rvba: Can you fact-check something for me? From what I’m reading on t’interweb, it’s not possible to use PostGres array fields in Django without the djorm-ext-pgarray package… which would require us to have that as a dependency, right?
[11:48] <gmb> (Asking because allenap suggested using arrays as part of the bootresources work, and it looks like we can’t)
[11:48] <rvba> gmb: from memory, we already use that package
[11:49] <gmb> rvba: ORLY
[11:49] <gmb> HOW INTERESTING
[11:49] <gmb> rvba: Bugger me, so we do.
[11:50] <gmb> See, that’s why one should grep first, ask the Djangoist after.
[11:50] <gmb> rvba: Thanks :)
[11:50] <rvba> no worries
[11:56] <gmb> allenap: As far as the BootSource model is concerned, ISTM that Cluster<->BootSource is a 1-Many relationship, so we should have a cluster FK on BootSource, no? Or should all BootSources go to all clusters?
[11:59] <gmb> (In which case there’s no need for an FK at all… Or we could have a table linking the two to make it N:N, but I don’t see what that gets us.)
[12:36] <jtv> gmb: we even got a patch into djorm-ext-pgarray...
[12:36] <gmb> jtv: So I see.
[12:43]  * gmb lunches
[12:50] <allenap> gmb: For now, don’t customise per cluster, and default to getting everything.
[12:52] <jtv> Is our 1.5 lander broken?
[13:24] <magicrobotmonkey> I'm glad the metrics got all smooth as soon as I showed up
[13:24] <magicrobotmonkey> super helpful
[14:18] <rvba> allenap: this is the branch in question: https://code.launchpad.net/~rvb/maas-test/power-type-support/+merge/217350
[14:38] <jtv> gmb: you looked into the landers recently... does the 1.5 one seem to be down?
[18:00] <smoser> allenap, https://code.launchpad.net/~blake-rouse/maas/virsh-probe-and-enlist/+merge/216632
[18:00] <smoser> am i wrong there ?
[21:19] <FunnyLookinHat> I've added an external network ( i.e. other than the private configuration network ) and added it ( by mac ) to a node - but it's not showing up or assigning an IP when I commission or start a node.  Any ideas?