20160904 Log

[18:02:41] #startmeeting

[18:02:49] oh, there's no chaosgoblin :)

[18:02:58] well, consider that line to be the "meeting strated" line :)

[18:03:11] Hi

[18:03:17] hey, larjona too!

[18:03:24] that makes three of us. a good start! :)

[18:03:55] how about we Just-In-Time some meeting topics :)

[18:04:02] * Pubstrate / ActivityPub updates

[18:04:14] * When's paroneayea working on MediaGoblin again????

[18:04:42] * GSoC wrapup

[18:04:52] anything else? :)

[18:05:01] 1.0?

[18:05:09] * 1.0 !

[18:05:12] o/

[18:05:21] hey breton ! yay, we're at 4

[18:05:24] yeeey, i am almost not late

[18:05:28] :D

[18:05:42] breton: want to add any topics?

[18:05:56] paroneayea: nope

[18:06:07] cool, ok

[18:06:21] #topic Pubstrate / ActivityPub updates

[18:06:51] as I think everyone here knows, I've been focused on getting the federation spec work tsyesika and I are editors on for the W3C, which is ActivityPub

[18:07:10] we basically want to get to Candidate Recommendation state by the end of TPAC, if not before

[18:07:24] maybe not everyone knows what tpac is :)

[18:07:33] -*- breton doesn't know

[18:07:57] https://www.w3.org/2016/09/TPAC/

[18:08:05] so TPAC is the big w3c technical conference thing

[18:08:12] I'm going to be attending in-person

[18:08:20] it's in Lisbon

[18:08:51] so, the good news is that AcitvityPub is moving well along. We just put out a new Working Draft a week ago: https://www.w3.org/TR/activitypub/

[18:08:58] earlier in the month, my friend Amy Guy stayed with us for a week

[18:09:10] she's also part of the w3c group (currently works at the w3c part time even)

[18:09:15] we knocked out a ton of updates

[18:09:27] including more cleanly separating the client-to-server from the server-to-server stuff

[18:09:36] anyway, the good news is: the spec is in very good shape

[18:09:47] the bad news is: the deadline is tight, and *serious*.

[18:10:00] the group only is chartered to be around till end of this year

[18:10:22] if we want the spec to go through and not just be a "note", we have to get everything to that point by year's end, and that's a lot of process

[18:10:23] and... we need implementations

[18:10:42] luckily, there are two implementations in progress: Amy's site runs one, and I'm writing Pubstrate.

[18:11:12] http://dustycloud.org/tmp/pubstrate_2016-08-17.png

[18:11:22] I can now log in and post notes

[18:11:29] it's very similar in design to pump.io

[18:11:41] but written in guile scheme, and a bit more demo'y

[18:11:55] so, no integration with mg yet?

[18:12:03] not yet. so

[18:12:09] I think porting GMG's federation code to this

[18:12:11] will not be hard.

[18:12:23] I'd like to have that done by end of 2016.

[18:12:52] that's a good tie in to the next topic, but maybe there are questions :)

[18:13:21] (I'm hoping to release a Pubstrate 0.1 release in the next few weeks so people can try it)

[18:13:55] https://github.com/cwebber/pubstrate it's currently (unfortunately) hosted on github. I'll move it to gitlab before release.

[18:14:12] that was originally for perceived-synchronization-with-group reasons, but I no longer believe that matters.

[18:14:26] any questions? or should we move on

[18:14:30] sorry for textwall :)

[18:14:48] -*- breton has none

[18:14:51] ok :)

[18:15:46] #topic When's paroneayea working on MediaGoblin again????

[18:15:57] good question/topic! :)

[18:16:06] I've been noticably absent, because of all the above, and tight deadlines

[18:16:30] I'm hoping that by late october I'm back on GMG to a reasonable degree.

[18:16:59] which is maybe also related to 1.0

[18:17:11] maybe we should switch to that topic since what else is there to say on this one ;)

[18:17:23] and we'll do gsoc after

[18:17:33] #topic 1.0

[18:17:52] 1.0! Right around the corner, several months ago, I thought, but then this happened...

[18:18:24] the main thing that needs to happen next is that we need to review, make sure it's stable, and merge tsyesika's federation code.

[18:18:36] isn't it merged yet?

[18:18:36] tsyesika: maybe you might have comments on what you remember the current state of that is?

[18:18:42] is it?

[18:18:46] i thought that all federation stuff is in

[18:19:04] nope :)

[18:19:23] it's still unmerged.

[18:19:31] nupe it's not merged yet, there are a few little tasks that need finished and then a full review

[18:19:33] is the branch with federation in the main repo?

[18:19:42] nope, it's on my notabug

[18:19:50] -*- tsyesika finds the link

[18:19:58] https://notabug.org/Tsyesika/federated-mediagoblin

[18:21:07] so

[18:21:16] if someone wants to do review and help this move along towards merge

[18:21:22] that *can* be done without me, I think

[18:21:31] though it would be helpful if there's a list of "things that need to be done"

[18:21:33] do we have that tsyesika ?

[18:21:37] tsyesika: which branch?

[18:21:54] i'd love to do review

[18:21:58] paroneayea: i can make one up and post it to the dev mailing list later :)

[18:22:18] in fact i'd do one if i knew that it needs reviewing a lot of time ago

[18:22:20] breton: "federation" i believe

[18:22:33] tsyesika: that would be great

[18:22:40] breton: that would also be great! :)

[18:22:56] before a review those little tasks need fixing, some a serious but quick to do but they definitely definitely need doing

[18:24:52] I want to add one more topic

[18:25:20] * Making breton co-maintainer (at least while SocialWG stuff is wrapping up)?

[18:25:31] not to switch topics now

[18:25:34] just queue it :)

[18:26:12] i have a question about 1.0. What about deployment stuff?

[18:26:21] breton: it's a good question...

[18:26:27] we had that as a target during that campaign

[18:26:29] breton: we need packages for distros.

[18:26:52] breton: we never hit that goal

[18:26:56] http://mediagoblin.org/pages/campaign.html

[18:27:01] but nonetheless

[18:27:05] it's still important to me

[18:27:11] that was going to be at the $100k

[18:27:12] level

[18:27:15] well

[18:27:27] we have another target

[18:27:30] which is premium hosting

[18:27:37] ironically, deployment has been a pre-requisite for launching that!

[18:27:42] oh, yeah, that one too

[18:27:51] so we never officially hit "deployment".

[18:27:57] but

[18:28:02] so I can't give details

[18:28:21] there's some vague possibility of funding debian packaging but I think it's low chance.

[18:28:28] but regardless

[18:28:31] I think it needs to happen.

[18:28:40] simonft got debian packaging pretty close already

[18:28:52] I have a guix package near-done

[18:28:54] Debian freezes 5 november

[18:29:01] I was working on making things more easily packaged before I got caught up in all this stuff

[18:29:15] larjona: it's unlikely we'll hit that freeze by my own work, but maybe if someone else can help

[18:29:31] I feel like our build system, in trying to make things easier, has also made things a lot harder

[18:30:07] probably the autotools part, but definitely the npm/bower part

[18:30:25] though, I hear jquery 2.0 is packaged in debian now

[18:30:29] actually i never figured that part out

[18:30:37] because i haven't done js in ages

[18:31:16] I think it should be torn out and replaced with something simpler: a directory where either distros can put javascript files, or we can pull them down and unpack the prebuilt ones

[18:31:22] via a script

[18:31:52] I think every javascript dependency we use already has prebuilt versions anyway

[18:32:09] so if we let distros provide the full version, or for developers let them run some script to get something

[18:32:14] or at the very least

[18:32:15] what do others do?

[18:32:19] for example, wordpress

[18:32:21] not make the bower stuff a core part

[18:32:27] breton: I don't know about wordpress, it's a good question

[18:32:52] breton: it's hard to answer, because what examples of web applications aside from wordpress do you see successfully packaged with distros anyway :\

[18:33:42] i also wonder about, say, django-cms

[18:33:47] even the php things are getting just as complex to deploy now

[18:33:54] breton: yeah a good question

[18:34:01] we should probably research before we make decisions

[18:34:12] probably we should try to package, and see how hard our stuff currently is, and figure out how to change things around that.

[18:34:31] I'll be interested in working on this once I'm done with my socialwg focus-mode madness

[18:34:44] even php is hard now with Composer

[18:34:50] https://lwn.net/Articles/670752/

[18:34:59] see the owncloud / debian rift

[18:35:36] what about guix package?

[18:35:42] breton: that one's already almost done :)

[18:35:45] I packaged all our deps

[18:35:53] and I was using it

[18:35:59] I just wasn't doing the final mediagoblin package

[18:36:00] excepting the javascript stuff

[18:36:05] heh!

[18:36:23] I have no idea how the js stuff will be done. we might need to provide another repo that does "evil" things and pulls in the pre-minified js pakcages

[18:36:28] though!

[18:36:30] oh, good news

[18:36:40] we had a "packaging npm for guix stuff" gsoc student

[18:36:50] I was technically a mentor, though the student did quite well without my help

[18:37:07] so, it turns out packaging jquery is insane, but hey, some things can be packaged using their tools now :)

[18:37:15] so there's hope? maybe? :)

[18:37:32] to meet the "premium hosting" target, we don't really need debian packages, we might provide the hosting with guix, right?

[18:37:41] I have high hopes and optimism for anything that doesn't need npm (and maybe even npm can eventually happen in guix)

[18:37:43] breton: yup

[18:37:57] no need to use debian, and I can do guix easily, even if it means mediagoblin's js deps won't go into guix proper

[18:38:10] we could get mediagoblin in guix proper, and provide a supplementary package that provides the js stuff

[18:38:25] maybe we should tear out all this js stuff. Who needs javascript libraries anyway! :)

[18:38:28] There is a sandstorm package but outdated (0.7?) If anybody can have a look at it, maybe it's easy to update it

[18:39:16] larjona: sandstorm support would be good... I wouldn't be using it for premium hosting I would myself run, but I would want us to be up-to-date in sandstorm-land

[18:39:33] breton: ugh, one more item not even on the agenda, what to do about this kombu / celery stuff :(

[18:39:46] breton: I have a client that paid me to do some consulting on their GMG deployment (which was nice!)

[18:39:52] and they got bit by it :(

[18:40:09] paroneayea: yep, lets get to it in the end

[18:40:12] ok cool

[18:40:21] * kombu / celery

[18:40:28] one more topic to queue for later :)

[18:40:52] so it sounds like tsyesika will post stuff to list, breton will review, and paroneayea will come back in late october to help? :)

[18:41:02] and larjona will look at sandstorm state? :)

[18:41:13] sounds good

[18:41:22] --> neiljp joined #mediagoblin

[18:41:29] (you don't have to do anything larjona, but maybe just nice to see what the state is !)

[18:41:38] though if you want to, you're more than welcome to :)

[18:41:47] yep

[18:41:56] cool, is it next topic time?

[18:42:07] (I'll probably just dump these minutes on the mailing list btw)

[18:42:27] #topic GSoC wrapup

[18:42:36] Yes, I'll report to the list

[18:42:40] larjona: great!

[18:43:20] so saksham1115 got their video subtitling stuff in

[18:43:20] <-- afuentes left (Ping timeout: 258 seconds)

[18:43:43] looks like results are announced

[18:43:53] we passed saksham1115 :)

[18:44:11] thanks to saksham1115, and to breton for co-mentoring (and doing bulk of mentoring, TBH!)

[18:44:45] saksham1115 wrote an update to the mailing list...

[18:44:52] maybe I should pull it up

[18:45:09] http://lists.mediagoblin.org/pipermail/devel/2016-August/001465.html

[18:45:10] there we go

[18:45:40] breton: do you think we could/should maybe get this merged in for 1.0 also?

[18:46:24] paroneayea: i'll review more thoroughly and tell after that

[18:46:46] breton: cool, if you think it's close

[18:46:54] tag for 1.0 on the tracker please!

[18:47:06] cool, any questions?

[18:47:10] should we move on?

[18:47:26] congrats again to saksham1115, and thank you, from GMG land! :)

[18:47:26] yep

[18:47:44] #topic Making breton co-maintainer (at least while SocialWG stuff is wrapping up)?

[18:48:12] so, breton has already been doing a lot of work keeping things moving smoothly in mediagoblin land... more than me while I've been in focus on the socialwg!

[18:48:17] and long-term, I'm not going anywhere

[18:48:31] but I think it would be a good idea to officially appoint breton co-maintainer, at least for now

[18:48:46] +1

[18:48:46] we'd have to make it official with GNU too, but how do people feel?

[18:48:54] +1 here too :)

[18:49:03] +1

[18:49:12] yay!

[18:49:14] let's make sure breton +1's ;)

[18:49:23] +1 :)

[18:49:24] whew that was the most important vote :)

[18:49:30] ok, so I'll start that process :)

[18:49:34] thank you breton !

[18:49:45] Thanks breton

[18:50:22] whee!

[18:50:28] ok, next topic

[18:50:29] that's a good present for my todays birthday :p

[18:50:35] :D

[18:50:38] happy birthday breton :)

[18:50:39] happy birthday breton!

[18:50:52] happy birthday!

[18:50:52] . o O (I have just the ascii art to close today's meeting...)

[18:51:07] #topic kombu / celery

[18:51:21] breton: want to explain the state of affairs?

[18:51:45] yep

[18:52:01] celery dropped sqlalchemy as a broker

[18:52:07] with a bunch of other brokers

[18:52:20] and they support only 4 brokers: http://celery.readthedocs.io/en/master/getting-started/brokers/index.html

[18:52:35] I mean, to be fair, sqlalchemy was never a "great" broker, but it made it easy for people to get started

[18:52:41] esp for small installs

[18:52:47] the reason is that it was hard to maintain and had some issues

[18:52:51] yes

[18:53:03] we could do a few things:

[18:53:21] - maintain the broker ourselves? (yuck, maybe temporarily?)

[18:53:40] - switch from celery to an actor model system... hm, that's maybe too long term :)

[18:53:52] - write a broker that uses unix named pipes?

[18:54:00] https://github.com/celery/celery/issues/3274#issuecomment-228324848 if someone's interested

[18:54:31] one more

[18:54:43] - we could switch to recommending one of the official brokers

[18:54:53] like rabbit/redis/etc

[18:55:06] i'd chose the last one as the simpliest one.

[18:55:40] the only thing that bugs me is uncertanty around redis

[18:55:59] breton: what's the uncertainty?

[18:56:21] paroneayea: the bug ^ initially was about issues with redis broker

[18:56:29] paroneayea: and they wanted to remove it too

[18:56:38] oh right

[18:56:45] but people from redis came and said that they are willing to work on the issues

[18:56:50] breton: maybe we should choose rabbitmq then

[18:57:05] breton: it's something to research, for now, I think

[18:57:24] paroneayea: yes. I don't thing that it's going to be hard, because it's well packaged everywhere

[18:58:00] breton: rabbitmq you mean?

[18:58:20] paroneayea: yes

[18:58:50] breton: ok, I'm ok with going rabbitmq

[18:58:54] it will take work

[18:59:29] here is our ticket https://issues.mediagoblin.org/ticket/5464

[18:59:42] I'll add a comment summarizing our thoughts from the meeting

[19:00:23] and... I think that's it from the meeting right?

[19:00:34] Yes

[19:00:52] ok, in that case :)

[19:00:58]   HAPPY BIRTHDAY

[19:01:02]

[19:01:02]   .-||-.

[19:01:05]  |'---______---'|

[19:01:08]  |.            .|

[19:01:11] _| '__' |_

[19:01:14] ( '--__________--' )

[19:01:17] '---__________---'

[19:01:20]       BRETON!

[19:01:24] thanks for coming, breton, larjona, tsyesika :)

[19:01:40] #endmeeting