What's in a MediaGoblin version 1.0?

[Ben Sturmfels]

Hi Folks,

Now that MediaGoblin 0.11.0 is out, I've been pondering what should be
in a version 1.0. I'm interested in your thoughts too.

So where are we?
----------------

We need to be honest - the project's momentum is low. That's a key
challenge for me as maintainer. I'm hoping we can regain at least some
of this with regular small releases and attending promptly to any new
contributions.

We already have a solid and useful Python 3 codebase with tests 100%
passing. Yay!

We're not currently running our test suite under continuous integration.

Documentation is in pretty good shape. Yay!

We have a few annoying bugs with celery and media processing that need
work.

A few small user-experience enhancements would help - things like an
upload progress bar.

The issue tracker is unstable and not currently an effective tool for
existing or new contributors.

The MediaGoblin main website is currently unstable and intermittently
offline.


What should be in (and out) of a 1.0?
-------------------------------------

I think version 1.0 is mostly about house-keeping, and that any extra
features we get along the way are a bonus.

My hunch right now is that distro packaging is *critically* important
and should probably be our main hurdle to 1.0. This will help make
MediaGoblin available to a much wider range of people. I think GNU Guix
is the first place we should package for. It's a niche operating system,
but Chris and I have already done significant work on this. Getting it
into Guix will also force us to solve a bunch of problems that will
hopefully make it much easier for subsequent Debian or Fedora packaging.

I think that federation should probably be out for 1.0, sadly. It's
always been part of MediaGoblin mission, but for now I think there's a
lot of value we can offer without it. We also don't have anyone actively
working on this so, in practise, I don't think we have a choice.

Celery has been implicated in a lot of MediaGoblin bugs, and RabbitMQ is
overkill for a small task queue. I'm starting to wonder if we would be
better to switch to something simpler like RQ (Redis Queue) or similar.
Either celery needs to be bulletproof or gone in 1.0.

I'm also like to get past major contributors involved in some kind of
low-time-commitment advisory role to help us make decisions about things
like the above possible Celery replacement.

I think version 1.0 needs a better bug tracker with support for merge
requests and ideally with integrated CI. I'm actively working on this.

We need a stable website too, so that involves setting up a new host.
Not super-complicated, just another job to doe.


What might the time-line look like?
-----------------------------------

I'm thinking either in the next one or two releases; ideally within 6
months.


Again, your thoughts are very welcome! I'm planning to organise another
community meeting soon, but wanted to get some ideas written down first.

Regards,
Ben

signature.asc
Description: PGP signature