[Anubad] L10n Dashboard

[Sankarshan Mukhopadhyay]

On Thu, Dec 5, 2013 at 2:07 PM, Sayak Sarkar <sayak.bugsmith at gmail.com> wrote:
> Since the past few weeks I have been thinking about a problem that has been
> a consistent pain for me ever since I started contributing to various l10n
> projects here and there. The problem was that there was no single central
> entity where I could go to, to find a list of all the projects that I
> contribute to, and at the click of a button get directed to whichever
> project's whichever locale I wanted to work on.

I would like to consider a different aspect. Localization,
irrespective of whether it is completely done by the community or done
in-house for a product, has the need for project management.

In other words, while a language team might be interested in checking
where they stand per project, a project team would like to view the
coverage and status of language/locales. A quick example, while bn-IN
might be interested in how it stands across the products within the
Mozilla family, Mozilla might be interested in checking how a specific
release of Firefox is covered by languages. These are various ways of
looking at the collated data - views. The underlying data-model that
powers the store ensures that such flexibility of reporting can be
in-built.

> Too many URLs to remember, too many translation systems to remember, and
> most importantly keeping track of all what's happening where, and what is
> the statistics of the various projects. Having all these in a single place
> would allow a contributor to make a better decision about which projects to
> prioritize while working.

So, the unification that a dashboard provides is not only the view. It
provides a ready-to-go framework. Consider the functionality you
expect from feed-readers - you need to add an URL (or, a feed) and,
you want to keep polling the target system for content refresh.

> I looked up a bit and couldn't find any such dashboard, so as a prototype I
> have started working on creating such an l10n-dashboard, which I think
> might also be a helpful tool for other contributors well.

An approach to take would be to separate the back-end store services
from the view model. Right now you are handing over the view to the
actual system (your dashboard -> Transifex). The ideal way would be to
figure out how you can systematically poll the content and, keep it
refreshed.

> The project repository can be found at:
> https://github.com/sayak-sarkar/l10n-dashboard with a detailed summary
> about where I'm currently working on this project as well as the
> prospective future roadmap of the project.

Additionally, while you can consider some means of running a globally
available service that provides the dashboard, any other project (say
a company doing non-open-source translations) should also be able to
install it on-premises and plug-in their content repositories and, get
the dashboards set up

> I'm quite interested to get some feedback on this project from the members
> of this group to better understand how this tool can be better implemented.
> Also, as it's already on GitHub, anyone can simply fork it and help me with
> a pull request based upon the project roadmap.

In short, what your prototype now does is a view. But what would give
you maximum impact is if you add reporting capabilities into the view
- quick example, tie in project freezes/timelines and, thereafter
visually depict which languages are falling behind.

> A demo of the current prototype can also be found at:
> http://sayak.in/l10n-dashboard. Not much at the moment, but it's atleast to
> something to start with. :-)

This is a good start. And, you have identified a large gap that is a
pain point for a lot of projects. I am hoping that you can encourage
more contributors and, set very aggressive goals on quick iterations
to get results.


-- 
sankarshan mukhopadhyay
<https://twitter.com/#!/sankarshan>