From Fedora Project Wiki

Revision as of 02:45, 30 October 2017 by Ncoghlan (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

This page is a draft only
It is still under construction and content may change. Do not rely on the information on this page.

Based on my experience in the Python community, I've largely split up the target audiences for open source development tools into two key groups: software developers, and data analysts.

The key difference I see between the two is that software developers are interested in software for its own sake, while data analysts are interested in understanding their data, and the software is just a means to that end. I describe this split in a bit more detail at https://fedoraproject.org/wiki/Env_and_Stacks/Projects/UserLevelPackageManagement#Audiences_to_be_considered

At the moment, Fedora Workstation targets the "software developer" group. I think that's a sensible idea, but it means we don't have a full desktop distribution specifically aimed at the "data analyst" group, even though the latter represents a *much* larger target audience than folks that would class themselves as software developers (think every engineering discipline, research scientists across all domains, business analysts, financial analysts, policy developers, etc, etc).

The previous attempt at addressing this gap, "Fedora Plasma" suffered from the fact that it was formulated as a technology focused push from KDE developers (vs the GNOME focus in Workstation), rather than as a focused solution aimed at address the needs of a particular class of users.

The Fedora Datascreen concept would instead focus on taking the existing Fedora Scientific Lab (https://labs.fedoraproject.org/en/scientific/) and expanding its target audience to cover the broader data analysis community, rather than focusing specifically on scientists (I've discussed this concept with Amit Saha, the lead maintainer of Fedora Scientific, and he's enthusiastic about the idea).

Fedora Datascreen's target users would be found at events like Strata, SciPy, PyData, Mining Software Repositories, and various academic conferences, moreso than at open source programming language conferences.

Some key technology stacks to consider would be:

  • Project Jupyter
  • Scientific Python
  • RStudio
  • GNU Octave (in its own right and for ex-MATLAB users)
  • KDE Cantor
  • Apache Spark (probably via OpenShift Origin and https://radanalytics.io/)

If this concept is pursued, then a concrete goal of Fedora Datascreen should be to become the upstream integrator for the components of Scientific Linux that are not provided by "The Upstream Vendor" (see https://www.scientificlinux.org/about/ for details)

In terms of potential alignment with Fedora's edition branding, I even have a logo in mind: a 3x3 grid of rounded boxes in the style of the logos on https://getfedora.org/

I'm less sure on a good thematic colour, but making it yellow in reference to the blue+yellow Python logo comes to mind as one possibility (that may just be my Python bias showing itself, though).

As a side effect of Fedora Scientific being based on KDE, this concept would *also* have the effect of elevating a KDE based system to full Fedora Edition status. For folks coming to Linux from other operating systems, this would make it relatively straightforward to direct Mac OS X users to the GNOME based Fedora Workstation, and Windows users to the KDE based Fedora Datascreen.