Mediagoblin community meeting notes (was: MediaGoblin community meeting this weekend - 4pm Sat Feb 15 UTC-5)

[Ben Sturmfels]

Hi Folks,

We've just wrapped up the community meeting. Thanks to those were able
to attend, and thanks so much for giving up time on your weekend! :)

We covered the short-term section of the agenda and will hold the
longer-term items over for another meeting (TBA). Meeting notes are
below - thanks very much to Chris for scribing!

Regards,
Ben

==== AGENDA (updated) ====

Intros
 - briefly tell us something about you

Short-term:
 - appreciation for the great work in the last few years (see "Thanks to" below)
 - a minor release ("we're still here")
 - a forward-looking blog post or two
 - ensuring issue tracker is working smoothly for new contributors
     - unintuitive compared to Github/Gitlab, hard to register, hard to sumbit 
a patch, went unseen [from Michael Lynch]
 - bug squashing parties [from Amelia Rose]
 - CI
 - urgent/blocking issues if any

Appendix - Thanks to:
Contributors since aroundabout 2016 (highly likely incomplete, apologies):
   - Andrew Browning (aleph): bug fixes and improvements, merging contributions
   - Boris Bobrov: bug fixes and improvements, merging contributions
   - Chris Lemmer Webber: bug fixes and improvements, community organising, 
ActivityPub
   - Jessica Tallon: bug fixes and improvements, ActivityPub
   - Vijeth Aradhya: multiple-resolition video transcoding/switching
   - Simon Fondrie-Teitler: upgraded/migrated/managed hosting for website, 
Trac, mailing lists and wiki
   - Saksham: lightbox, video subtitles
   - Amelia Rose: bug squashing parties
   - Ben Sturmfels: bug fixes and improvements, migrated mailing lists
   - Iyzsong: bug fixes
   - Berker Peksag: Python 3 bug fixes
   - Johnathan Sandoval: login validation improvement
   - Leah Velleman: localisation
   - Loic Dachary: cleanups
   - Matt Dealsy: UI improments (PyCon AU 2016 sprint)
   - Kesra: bug fixes (PyConAU 2016 sprint)
   - Josh Crompton: UI improvements (PyCon AU 2016 sprint)
   - Olivier Mehani: OSM tiles, datetime format config
   - Alexandre Frankie: docs improvements
   - Stéphane Péchard: collection option to addmedia
   - Robert Smith: bug fixes
   - Simen Heggestøyl: bug fixes and improvements
   - Romain Porte: tags in API
   - chrysn: EXIF rotation
   - DPG: license option updates
   - Dod: bug fixes
   - Michael Lynch: dev setup improvements
   - Ian Kelling: set up and imported our mailing lists (twice!)
   - undoubtedly others


====== MEETING LOG =====
sturm: thanks for coming everyone, weekend meetings are difficult, so thank you 
everyone to making it.
this is my first time doing a larger mumble meeting
let people pause after they speak to avoid interruptions
I've put the agenda on the etherpad (above)
anything else to add at this point?  No?  ok!

TOPIC: Introductions
sturm: Thought it woudl be good to go around and do introductions, I'm Ben, 
from Australia, getting tea and saying hello to my kids.  I'm a software 
engineer, doing intermittent mediagoblin work over the years.  I used it a bit 
early on to host photos of my kids and for usergroup things and now I'm using 
it for indigenous community looking to publish their media.
ayleph: Hi!  Yes I'm Andrew, I'm in Seattle, Washington, about 1pm here since 
I'm in the west coast in the US.  In my dayjob I'm an engineer for aerospace 
stuff and robotics.  Started poking at and fixing bugs in python in 
mediagoblin.  I've run Goblin Refuge, probably one of the biggest instances.  
It's been hitting some issues with queues and etc recently but still running

cwebber: one of the three creator of Mebiagoblin; got pulled away from it for 
other activities; currently working on Spritely, co-authored/edited ActivityPub 
(originally for MediaGoblin's use)

lnxw37ipad: I’m Walt. I’m in Southern California. I’ve been a user of federated 
socnets since 2009 and hosting my own since 2010 or 2011. I do tech support and 
Windows administration for a US federal agency. I’m here mostly to listen and 
see how GMG is going to proceed in the future.

shtrom (Olivier Mehani): I'm now based in Hobart (Tasmania, Australia), started 
hearing about / playing with MediaGoblin in 2013, didn't want to get involved 
in proprietary platforms and when pump.io I got excited by it being 
decentralized and distributable.  Tried to set up Medagoblin for his partner' 
painting workk buh didn't quite take off. Later on, since we're spread out from 
family mediagoblin was a good solution to share pictures of our kids.  Been 
running it since our first kid was born.  Mostly runs ok, try to fix bugs when 
they come up

R13ose: My first name is Amelia Rose and from Toronto, Canada. I have been a 
user of mediagoblin for a few years. I am doing freelance web developer and 
environmental work. I want to see if the project can come back alive and the 
issues tracker can be updated and maintanced well.

simonft: Hi, I'm an infrastructure engineer at a company in New York.  I 
started getting involved in MediaGoblin in 2011(?), realized Chris was also in 
Madison, met up and said hello, have been maintaining MediaGoblin servers since 
about then as well as other work.  Tried to get MediaGoblin into Debian but 
that ran into issues... last few years haven't had a lot of time/energy for the 
servers as much as before

sturm: Great thanks Simon, so I updated the agenda a bit...
[call for agenda edits?]
Looks like no, so two big blocks, short term / long term
First of all want to say thanks for all the work that's been done over the 
years.  huge appreciation to andrew and boris who have merged lots of and fixed 
bugs and reviewed things, which takes time and is a regular activity.  Chris 
and Jessica have obviously done a bit of work.  Vijeth did the multi-resolution 
video transcoding.  Simon's obviously done a lot of work admin'ing the servers. 
 This list is in no particular order but there is a large and incomplete list 
of people who have helped, of which there are many in terms of code and other 
things.  R13ose has done a lot of work to corall bug-squashing.  I've moved the 
mailing list over the years.  (goes through list of all the helpful people 
above!)

sturm: next thing I'm thinking of in terms of showing people we're still here 
is a minor release.  Doesn't have to be complete or a load of new things, but I 
think a minor release before any major release might help, especially because 
our current docuemntation refers to the stable release

ayleph: I've been running master with some modifications with other people... 
after I updated some merges from master things became very unstable for me, 
things get stuck in the queue.  It might not be related to that, it might be 
the celery queue or something else.  Or maybe it's my own modifications?

shtrom: I have kind of a similiar situation... have been running master with a 
few fixes.  I've updated a few things from it, I guess there's the known issue 
of (audiolab?).  I have 3 or 4 sets of python libs in vairous states of 
disrepair.  I think it is a good time to get people upgraded, try to do at 
least one release a year to not get too difficult.

cwebber: I think it's a good idea to do a release, will regain interest.  if we 
have showstopper bugs, let's maybe look at those... ayleph, did you file it as 
a bug?

ayleph: don't know if I filed a specific bug, will have to look at the tracker 
which I haven't done in a while

sturm: sounds like a good idea for the release for marketing purposes and etc, 
but maybe there are some issues with bugfixes.  I think there's some work for 
the minor release or etc

lnxw37ipad [chat]: A patches and bug-fix release would be a good thing, IMO. I 
occasionally see people asking if the project is still alive.

R13ose [chat]: The blog needs to be updated as well.

sturm: a release and blogpost are low-barrier things to help people see that 
the project is still alive
what's appropriate?  how far is it to go into?  maybe just a post about things? 
 I'd be happy to write something, or if someone else has questions... I have 
been doing a project for work, which I'd be happy to write about... or write 
about what people have been working on

cwebber: blog post is a good idea; other ideas: install day to get developers 
going; encouraging people to get involved by talking about a bug-fixing day; 
acknowledge quietness

sturm: yeah I think I like that you said that acknowledgin "it's been quiet 
lately"... just acknowledging that can help.  What do you think about the 
install day?

cwebber: We actually did this bacgk in the day: made announcements saying ... 
and I think it was (asheesh) that recommended it, and Rose was running 
bug-triage day. We could announce that it would happen over a full day, and we 
could pell that off since sturm is in Australia. We don't know if other people 
are going to show up, but it's a signal, and an opportunity to try.. to do two 
things at once: acknowledge quietness, but point out what got out of it 
(ActivityPump standardisation with Tsyeika; no shame about that). What do you 
think sturm?

sturm: I think yeah that's great, acknowledging that, and it's so important as 
I've learned as a small business owner to market and show what's there.  I 
think the world still has a hole in it that mediagoblin fills, and we can go 
into that later.  I'll write an outline and send it to the mailing list

cwebber: yes and then I can review and help co-author, we can develop on the 
mailing list

TOPIC: issue tracker

sturm: simonft and chris and I had a conversation about the issue tracker, we 
had a conversation and I thought "oh we don't need to change it, maybe I can do 
work to just get it up and running again".  But hearing other feedback, while 
I'm fine with Trac and have used it and since I can commit to the repo there 
are some hurdles I don't encounter that others have.  Michael said that they 
found it very hard to deal with compared to github/gitlab/etc.  the experience 
is probably a bit poor compared to what people are used to.  I've been kind of 
burning time moving issues around.

shtrom: yes I think gajim has migrated from Trac to GitLab, and it's pleasant 
... I've been using Trac for years, think it's been used for years.  Not sure 
about how to integrate with the GNU system / forge.  Not sure about merge 
requests and etc.  Would be worthwhile.  Limited enough resources where it 
could be problematic to file Merge Request  for issues and integrate with the 
code repository.  Wonder if we could go halfway and just make Trac read-only, 
and as issues come up move them to a new instance and start adding issues to 
the issue tracker on demand, so we can roll in issues that are alive, and ones 
that aren't are a backlog

sturm: Thanks shtrom, that's a good approach.  Chris, other obligations?

R13ose [chat]: Would updating tracker be helpful?

sturm: yes I think it would help, it's certainly behind, having to paste 
patches into trac is a bit clumsy for the github generation.  Probably doesn't 
feel right anymore.

cwebber: a new issue tracker and just copying old issues over would be a good 
idea, even if just creating a new issue and puting a link to track. this would 
allow to triage/remove the chaff. GNU.org i currently looking at other option 
at the moment: GitLab, GOGs, SourceHub. GitLab is the most familiar; GOGS ...; 
SourceHub tooling good for developers, not necessaliry for others (very 
mailing-list oriented). GitLab enables drop-by contributions. It would be 
interesting to see what GNU ends up doing. Might be worth pinkng Ian Calling 
(?). Sturm ; We're not tied to GNU tools specifically, as long as there's no 
proprietary JS.

sturm: what's the best way to choose?  ask the ML?

cwebber: Step 1: ping Ian; Step 2: once we have a better idea, see how long 
it's gonna be; Step 3: decide whether we want to go with that. Ask Ben/Sturm if 
he want to volunteer to do this.

sturm: yes I'm happy to ping Ian

simonft: so I think that's probably worth figuring out if the FSF is going to 
host Gitlab or etc... I think one question worth asking with this is whether or 
not we have the resources to host one ourselves.  I would not like to commit to 
personally hosting it just because gitlab in particular is complicated... if 
it's not gitlab hosted by the fsf our options are somewhat limited.  I'm not 
sure what our definition of free software is in terms of JS.  I know gitlab 
itself is not fully free software in terms of the backend.  I think we've 
discussed this before and there aren't a lot of good answers.

sturm: amelia rose mentions bug squashing/triaging parties; it would be a great 
thing to do; could be done after

shtrom: I guess if we're going to triage bugs... maybe if we manage to get an 
instance running in time, we could look at the bugs that are there, see if we 
can manually get them into the new instance and if ready to go

sturm: sounds good and we can probably pick out the high priority ones

cwebber:  we have an order of operation problem; suggested resolution: Ben to 
send an outline to the ML; no issue in merging bug triage and install day; bug 
triage is not necessarily dependent on the bug tracker, but the install day 
isn't; new bug tracker not necessarily needed to be able to triage

sturm: we could use the bug triaging to capture the high priority things like 
what ayleph is experiencing , even if just a list

rose: are we 100% we are moving from trac

sturm: :looks like it, not sure of the options
moving on, CI is also something that Michael mentioned on the ML; it's a good 
idea to be able to run tests all the time, and the test suite is slightly 
unloved, seems to be a further down the list of priorities at the moment

ayleph: if we're evaluating different issue trackers, maybe consider that, 
since some deal with CI

simonft: I think that makes a lot of sense, gitlab has one built in... there 
are some CI servers attached to it.  Other CI services are tied in very closely 
to the issue tracker and git hosting.  I think it can be worth keeping in mind 
when choosing the issue tracker.

sturm: that's good, I've also done a bit of work on docker scripts that 
essentially give you such an environment.  We are moving int this area 
hopefully we can make a wonderful beautfiul integrated thing

sturm: next item is blocking issues, maybe this isn't the right format, but one 
is definitely the audiolab stuff, the issue being that the audio processing 
stuff that does the spectrographs is for python 2, there isn't a python3 
version.  at the moment I've patched in a hack that doesn't generate 
spectrographs for py3.  At the moment there are definitely things that throw up 
issues that don't feel so robust.

shtrom: I have a small prioritized list of issues so I'll probably drop that 
in... my main issue is the installation on sqlite, and one user, I've had 
trouble with (??)... I have celery consuming about 7% CPU all the time and I 
don't know why... I also run into issues with database locking if you add too 
many uploads at the same time.  It can lead to needing to restart mediagoblin.  
I don't know how to debug that... I've tried updating the process, it's a 
heisenbug which is frustrating.  That's the thing, that's my main issue, I'll 
drop the rest of it in there about locking issues and debuggability

sturm: and maybe these are things that should go on the ML... maybe write up 
about that and send to the mailing list?

shtrom: sure I'll give it a few more days to think about it then send

ayleph: I'll mostly echo ben and shtrom, I was never able to fully integrate 
from py2 to py3... it's gotten harder.  I used to use fastcgi with flup, which 
doesn't work there... I can't get the right debug issue with wsgi, it's hard to 
get backtraces and errors.  Something about the changes in python2 -> python3 
has made it hard.  Maybe if I set up a new CI deployment I could better 
document on bugtracker?

sturm: sounds like some interesting challenges, obviously some challenges with 
moving rom py2 to py3, some serious stuff to deal with how we deal with from 
here

shtrom dropping off

sturm: how to both get mediagoblin moving and also deal with the py2->3 stuff

cwebber: just wanted to say thanks to Ben for running this meeting, and it's 
nice to see famliar faces again and get people on here.

sturm: echo that

ayleph: i want to echo that, thank people for continuing it moving along

simonft: yes, thanks for pushing things along

sturm: yes, sounds good, thank you everyone!  We'll post some things to the 
mailing list including this stuff here today.  We'll try to schedule another 
one!

signature.asc
Description: PGP signature