16:58:53 <meejah> #startmeeting
16:58:53 <MeetBot> Meeting started Wed Jun 11 16:58:53 2014 UTC.  The chair is meejah. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:58:53 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:59:05 <meejah> so for agenda, i have:
16:59:08 <meejah> 1. updates
16:59:28 <meejah> 2. more-open merging for "next", or another branch?
16:59:35 <meejah> anything else?
17:00:02 <karsten> anything we need to discuss regarding deployment? (I have no news though.)
17:00:20 <meejah> 3. deployment stuff?
17:00:34 <karsten> nothing else from me.
17:00:37 <meejah> okay. lets start with 1.
17:00:40 <meejah> #topic 1. updates
17:01:10 <meejah> i'll start. i merged two of lucyd's branche (search, onionoo_wrapper) and pushed to a branch in my repo
17:01:22 <meejah> and sent a couple emails about deployment
17:01:41 <meejah> #action meejah needs to merge pull-req from jondbaker from yesterday
17:01:42 <lucyd> #info added couple of small checks like 'is_stable' and 'is_hibernating' to onionoo-wrapper utilities
17:01:43 <karsten> did you merge anything into next which I should pull-and-push?
17:01:51 <meejah> karsten: no
17:01:54 <karsten> ok
17:02:19 <lucyd> oops karsten go ahead
17:02:27 <meejah> #info merged lucyd's search, onionoo_wrapper
17:02:28 <karsten> ah no
17:02:33 <meejah> #info mailed about deployment
17:02:36 <karsten> lucyd: didn't want to interrupt.
17:02:44 <meejah> okay, lucyd go ahead
17:03:43 <lucyd> ok after making minor modifications to the wrapper, I rewrote the tshirt script(#9889)
17:04:19 <lucyd> #info also changed the relay-search script(#11956)
17:05:00 <meejah> lucyd: tshirt thing, is there a pull-request or merge I haven't done for that?
17:05:32 <lucyd> well, the ticket is not a part of the weather-rewrite so I didn't update it anywhere
17:05:55 <lucyd> i dont think there is a ticket for the rewrite specifically
17:06:04 <lucyd> should we add one now in the wiki?
17:06:23 <meejah> #action lucyd add a ticket for tshirt to wiki
17:06:37 <lucyd> #info i think some front-end tweaking is all it takes to make this relay-search script integrated with the new weather
17:06:46 <meejah> great!
17:07:10 <meejah> does it have tests? (unit or an integration?)
17:07:17 <isis> nickm: i am revising bridgedb's concept of "what a bridge is" and "what a hashring is", and we're currently using the `ring = set([elements]); key = sha1(client); pos = key % len(ring)` method… do you think using consistent hashing would work better? http://www.tomkleinpeter.com/2008/03/17/programmers-toolbox-part-3-consistent-hashing/
17:07:34 <lucyd> well not yet...i'm working on updating(and testing) the wrapper first
17:07:41 <meejah> okay
17:07:58 <meejah> do you feel you're on schedule, overall progress-wise?
17:08:00 <lucyd> anyways, the interaction between the relay-search script and the wrapper wouldn't change though
17:08:40 <lucyd> meejah : yeah, it looks ok but I'll be chipping in more hours so we have ample time to test and debug later
17:09:04 <meejah> okay
17:09:15 <meejah> any more updates? or can we move to 2?
17:09:19 <lucyd> btw, meejah karsten : have you looked at the wrapper?
17:09:36 <meejah> i have
17:09:37 <lucyd> after meejah's reply yesterday i'm working on updating the caching module
17:09:46 <karsten> lucyd: unfortunately not. :(
17:10:10 <meejah> i would say just rip it out for now, and once the system is working add back in?
17:10:11 <karsten> glad that meejah has.
17:10:29 <lucyd> meejah : okie dokie
17:10:51 <meejah> i.e. i'd rather see a working (with tests) version first, then add (back) the caching. and/or use luketheduke's
17:11:29 <meejah> (using memcached seems like the right choice here, since you've got web stuff and cron-run scripts etc)
17:11:39 <meejah> ...but let's worry about that feature later :)
17:11:41 <luketheduke> oh hi.
17:11:51 <luketheduke> was there a meeting scheduled for today?
17:11:52 <meejah> #topic 2. next branch
17:12:21 <meejah> luketheduke: yes, i didn't get out a timely reminder this week...but it's same-time-same-place each week
17:12:30 <luketheduke> the wiki page still says next meeting is in may >_>
17:12:45 <karsten> luketheduke: want to update it? :)
17:12:49 <meejah> ooo, that's a fail for me from last-time's #action to update wiki page :/
17:12:57 <luketheduke> anyway I'm here (:
17:13:08 <luketheduke> karsten: I don't think I have all the info needed to update it
17:13:08 * meejah was away 5 days this week...
17:13:34 <meejah> i'd like somewhere we can merge changes that won't get auto-sucked into production
17:13:34 <karsten> luketheduke: ok. sounds like meejah was planning to do it? otherwise I can do it.
17:13:52 <meejah> #action karsten update wiki page w/ meeting schedule
17:13:54 <meejah> :)
17:14:03 <luketheduke> meejah: into a branch called "development"? >_>
17:14:14 <meejah> i.e. either karsten agrees not to merge anything from "next" or ... ^^ what luketheduke said
17:14:50 <meejah> okay, i'll make a branch called "devel" and more-rapidly pull things into there
17:15:08 <meejah> ...and merge "ready to go into production" things to next (as per current plan)
17:15:15 <karsten> ok
17:15:15 <meejah> sound good?
17:15:23 <karsten> (wiki page updated.)
17:15:26 <meejah> #agreed create and use "devel" branch in tor-weather
17:15:43 <meejah> oops, didn't mean to hit return just yet on that ;)
17:16:06 <meejah> okay, if nobody has more on that, let's move to 3
17:16:18 <meejah> #topic 3. deployment updates?
17:16:25 <karsten> so, we still have no reply from our sysadmins.
17:16:36 <meejah> #info no real update, karsten and i emailed, and waiting for sysadmin reply
17:16:43 <karsten> how bad would it be to develop with just-wheezy-packages in mind for now?
17:16:52 <karsten> wheezy+wheezy-backports
17:17:01 <meejah> it makes "using OnionPy" harder
17:17:04 <luketheduke> no objections, as long as -backports stays in
17:17:23 <karsten> -backports should stay in, yes.
17:17:37 <meejah> ...but we're already defering that to later anyway at this point
17:17:41 <karsten> ok
17:17:48 <luketheduke> (python-requests in wheezy is an unusably old version)
17:18:13 <meejah> #info we do at least need wheezy-backports i'd say
17:18:26 <karsten> I'm optimistic about that.
17:18:32 <meejah> #info at least for python-requests
17:18:42 <karsten> also, will ping sysadmins later today.
17:19:22 <meejah> #action meejah to read more about recommended django deployment
17:19:44 <meejah> #info currently, it's behind apache (right, karsten?)
17:19:58 <karsten> yes, and I think we'll want to keep it behind apache in the future.
17:20:23 <karsten> btw, I was wondering if we could turn this "research how to deploy something on a tor machine" into something more general than just weather.
17:20:27 <meejah> #info will be deployed behind apache
17:20:39 <karsten> I bet other tor devs frequently wonder about these questions, too.
17:20:50 <meejah> yes, i can write something like "how to deploy a python-thing to a tor-machine"?
17:20:58 <luketheduke> is there a rationale behind using apache|
17:21:01 <luketheduke> ?
17:21:18 <karsten> meejah: I think that would be quite useful. using sysadmin input.
17:21:19 * meejah defers to karsten
17:21:29 <karsten> luketheduke: our sysadmins know apache pretty well.
17:21:35 <meejah> #action meejah to write up deployment findings as HOWTO
17:21:41 <luketheduke> ok
17:22:03 * meejah agrees defering to sysadmin's choice of webserver makes sense
17:22:09 <HB33> Does anyone have time for a quick tor question?
17:22:23 <luketheduke> HB33: -> #tor
17:22:26 <meejah> i think that closes the meeting, unless there any more topics?
17:22:37 <meejah> #action meejah needs to merge jondbaker pull-req with more tests
17:22:39 <lucyd> meejah: on a side note, developing the backend scripts requires using the cache feature
17:22:51 <meejah> lucyd: why?
17:23:28 <lucyd> meejah: for example, the hourly script has to keep downloading onionoo docs
17:23:40 <lucyd> and the protocol insists usage of 'If-Modified-Since' header
17:23:41 <meejah> how many docs?
17:23:56 <karsten> well, you can of course send requests without that header.
17:24:08 <karsten> if you don't have any older documents available.
17:24:21 <lucyd> karsten : ok and then update them once the feature is implemented?
17:24:27 <luketheduke> since OnionOO updates hourly, an hourly script doesn't really need to use caching.
17:24:41 <karsten> don't worry too much about overloading onionoo. it has moved to a new server which should handle the load.
17:24:55 <meejah> lucyd: for now, we use live onionoo req's for everything.
17:24:58 <lucyd> luketheduke: consider the daily script as well
17:25:02 <karsten> lucyd: just let me know when you think you broke it. ;)
17:25:06 <luketheduke> lucyd: that even less.
17:25:06 <meejah> if weather is pounding it, we do caching sooner ;)
17:25:16 <luketheduke> it's a very good idea to have caching in place when you go into production, but for testing, meh, who cares.
17:25:33 <meejah> luketheduke: yes :)
17:25:49 <lucyd> alright then
17:25:50 <karsten> lucyd: maybe restrict your test set to just a few relays, so that your tests don't run forever.
17:26:05 <karsten> but otherwise, just do the requests you need.
17:26:12 <meejah> i believe the vagrant setup includes some test-data?
17:26:51 * karsten doesn't know that.
17:27:09 <meejah> lucyd: for testing (e.g. unit-tests) you'd likely want to mock out the onionoo wrapper and feed it the test data (i.e. no onionoo req at all)
17:27:58 <karsten> ah, yes, no onionoo requests in unit tests, please.
17:28:31 <meejah> #endmeeting