[13:50] <benji> gary_poster: ...and we're back
[13:51] <gary_poster> benji, none the worse for wear?  any idea what went wrong?
[13:51] <benji> my machine wasn't aknowleging that any USB devices exist; lots of things are usb, like WiFi and keyboards and other important things
[13:52] <gary_poster> frankban, did you have plans as to what PPA to use for the haproxy?  I was thinking a juju-gui PPA might be good fwiw
[13:53] <gary_poster> benji, so what did you do?  Did you have to perform the Sacred Rite of Beseeching the USB Gods to Intervene, involving bananas and chocolate sauce?
[13:53] <benji> I think it was yesterday's update that got me.  After connecting an ethernet cable directly into my cable modem I was able to get a kernel update and it was happy after that.
[13:53] <benji> And what you said.
[13:53] <gary_poster> :-)
[13:53] <gary_poster> cool
[13:54] <benji> mmm, that makes me think of the fried banannas in the Thai place in Fredericksburg
[13:54] <gary_poster> heh
[13:55] <gary_poster> teknico, submitted code review of your branch; trying it out now in practice
[14:07] <frankban> gary_poster: I am testing with my own PPA. IIRC, in launchpad is quite easy to move a package from one PPA to another. +1 on juju-gui PPA
[14:07] <gary_poster> cool frankban, sounds good
[15:03] <teknico> gary_poster, thanks for the reviews and the comments. I implemented your suggestions and proposed again
[15:03] <gary_poster> great teknico.  would you like a follow-up?
[15:05] <teknico> gary_poster, yes, thanks
[15:05] <teknico> gary_poster, as far as linking the code docs from the project ones goes, I feel I'm missing something simple
[15:06] <gary_poster> teknico, ack.  I'm guessing simple links like ../../../ don't work.  Even if they did, they are fragile.  I suppose we could also make a symlink from the yuidocs to within the html output, as part of the doc makefile?
[15:07] <teknico> gary_poster, yes, I guess going up the dir tree would be fragile, but I can't even come up with a syntax for it :-)
[15:07] <gary_poster> heh
[15:07] <teknico> gary_poster, on the other hand, the symlink is the simple thing that I was missing :-)
[15:07] <gary_poster> :-) cool
[15:12] <teknico> gary_poster, uhm, no, not really. how do I express "current dir" with a file:// URL? is it even possible?
[15:12] <bac> benji: when you were investigating saucelabs did you encounter any mention of capybara as an abstraction above the web drivers?
[15:13] <benji> bac: nope, what does it do?
[15:13] <gary_poster> teknico IME you are already in the file:// namespace, so relative links work great
[15:13] <teknico> without protocol qualifier? ok, I'll try it
[15:14] <bac> benji: it is a more abstract way to express the tests rather than using the web driver.  i just watched the 'retrofitting an existing web app with selenium' talk and he was really in favor of it.
[15:14] <bac> benji: i can't tell if it is too ruby-centric, though.
[15:14] <bac> benji: actually it is called "Better Late Than Never: Integrating Selenium After The Fact"
[15:15] <benji> bac: I'm looking for info about it now.  If it requires a language other than a real programming language to describe the tests, I would be very skeptical.
[15:15] <benji> I'll take a look.
[15:17] <bac> benji: it is pretty unreal
[15:19] <benji> Capybara appears to be Ruby-only.
[15:19] <teknico> with "py" in the name?!? appropriation! mistification! sacrilege!
[15:20] <benji> all my future python projects will include references to precious stones in order to confuse rubyists
[15:21] <teknico> that'll teach them ;-)
[15:29] <frankban> gary_poster: do you have a minute, after the call, for a quick hangout?
[15:29] <gary_poster> frankban, y
[15:30] <frankban> gary_poster: thanks
[15:30] <gary_poster> bac benji bcsaller1 hazmat frankban Makyo teknico goodspud call now :-P
[15:35]  * hazmat plays hangout ping-pong
[15:43] <benji> grr, my two factor auth is screwed up again
[15:51] <gary_poster> teknico, juju-ui?
[15:51] <teknico> gary_poster, yep
[16:04] <gary_poster> bac, juju-ui?
[16:58] <gary_poster> teknico, don't hate me for abusing your poor code review :-D
[16:58] <gary_poster> which is to say, sorry
[16:59] <teknico> gary_poster, no worries :-)
[16:59] <Makyo> Better to ask forgiveness, etc. :)
[17:00] <gary_poster> :-) exactly Makyo, thank you teknico 
[17:30] <gary_poster> didn't work :-/
[17:31] <gary_poster> no dkim from rietveld, despite google connections, unfortunately
[17:31] <gary_poster> if we had it, it would work
[17:31] <gary_poster> but no biggie
[17:31]  * gary_poster goes to lunch
[17:56] <teknico> gary_poster, landing my branch, could not use the symlink but had to go for the fragile link solution
[17:57] <teknico> gary_poster, also changed the Makefile doc targets and added a few, I remain awaiting the flames ;-P
[18:16] <gary_poster> teknico, heh, ok thanks.  have a good evening
[18:28] <benji> does anyone that is around know how long the charm usually takes to start?
[18:35] <benji> it finally started, so I guess the answer is "five minutes longer than you are willing to wait"
[18:38] <gary_poster> benji how long did it take you? My experience is 10-15 minutes IIRC
[18:40] <benji> gary_poster: that's in the range, I was pretty bored waiting so I don't know for sure
[18:41] <gary_poster> :-) k
[18:50] <gary_poster> benji, would you agree that it would be a valuable task to teach the charm how to start and expose the test server?
[18:51] <gary_poster> Then we would be able to write a selenium test against that, to run unit tests on different browsers
[18:51] <benji> gary_poster: yep (we may do something else in addition, of course; but having an easy way to say "oh, run the unit tests too" would be a big win)
[18:51] <gary_poster> cool
[19:17] <gary_poster> Makyo, if you are looking for something to do at some point let me know.  I have something that would be probably fast for you, and less so for others
[19:18] <Makyo> gary_poster, Alright.  Halfway done with #1099921.
[19:18] <_mup_> Bug #1099921: Dragging services fails intermittently <juju-gui:Triaged> < https://launchpad.net/bugs/1099921 >
[19:18] <gary_poster> Awesome Makyo!
[21:44] <benji> bac: I am trying to remember how we got around SSH's fingerprint verification prompts from Juju.  Was the best we came up with setting UserKnownHostsFile and StrictHostKeyChecking in .ssh/config?
[21:44] <bac> benji: i thought we just used the appropriate command line switch to tell it to leave us alone.
[21:45] <bac> benji: a quick look at the buildbot charm should show, no?
[21:45] <benji> bac: if we are directly using SSH that works, but this is Juju envoking SSH on our behalf
[21:45] <benji> bac: good idea
[21:45] <bac> oh yeas
[21:50]  * Makyo dogwalk.
[21:57] <benji> bac: I haven't attained enlightenment.  (but I have been reminded how many moving parts that project had)
[21:57] <bac> amazing, huh
[22:05] <hazmat> benji, that's one way..
[22:05] <hazmat> benji, it was actually in juju for a little while before people freaked out about the security implications ;-)
[22:06] <hazmat> so now everyone does it individually.. and blindly says yes.. which is not really an improvement
[22:06] <hazmat> gojuju ends up doing tls with certs to avoid this fwiw
[22:07] <hazmat> ie, they pass the server cert down via user data, its an insecure channel..  but its effectively the only proc against the pristine image.
[22:07] <hazmat> digression
[22:07] <benji> I could have sworn we found a better way, but I can't remember/find it now.  Maybe it will come to me later.