A lot has been said and written about Ubuntu Unity, the new ‘shell’ that’s replaced the ‘classic’ default GNOME desktop in Ubuntu 11.04. Despised by many that interpreted Canonical’s break from the ‘open-source’ norm of restricting modifications to upstream platforms to a bare minimum, as a threat to the upstream projects’ existence (a valid point to an extent), that found it to be half-baked and offering little more (if anything) over the classic desktop and a couple of additional programs (e.g. a Dock, a launcher etc.) while much slower and kludgey (a totally valid point, but it’s a 1.0), Unity is here to stay.

It is true that, despite Shuttleworth’s ramblings on his blog, most of Unity is hardly innovative. Most useful things in there can already be found in most modern desktop environments (including some linux desktops) while Unity’s implementation of those very features is hardly the best. But there are also some unique offerings that are different, such as lenses and the proposed (but not yet included, thankfully) windicators. The question there is: are those features really useful? Are they well thought-out?

I think not. Take for example desktop search, a hot subject in mid 2000s desktops that’s been largely solved in an exemplary way in OS X by Apple’s Spotlight and a number of third party tools on that platform (LaunchBar and then Quicksilver are prime examples of early game changers), and even Windows 7 to some extent through the built-in search field in the start menu. Then, with five years of hindsight, Canonical decides to make things somewhat harder for users by exposing the search context to the user in the form of completely separate ‘lenses’ as opposed to keeping the distinction internal (in the same way OS X does) and presenting filtering options in an innovative way. Put it simply: I’d much rather have a single search field, ala Mac OS X’s Spotlight that searches for my input text across ‘data domains’ and contexts and returns useful, filterable lists of data, than the frustratingly badly designed ‘lens’ concept that forces a clear separation of searches while taking up screen real estate and wasting the users’ time with additional clicks and keystrokes.

Which begs the question: why on earth did the fine people at Canonical make such a bad design decision, when the stated mission of Unity was to streamline the desktop while taking less space etc. and at the same time there are numerous implementations of search/launch applications (even in linux) that work significantly better than Unity? Were they afraid of being labelled copycats? Is that worse than been called bad designers?

The same can be said about the new ‘global menu’ and AppIndicators that replace Gnome panel in Unity. Having few replacements for the staple Gnome Panel widgets of yesteryear is fine, given it’s a 1.0. Having botched the whole concept of a global menu through inconsistencies when windows are maximised and in multi-display scenarios betrays a badly designed (viz. not just incompletely implemented) system that shouldn’t have been out in the first place.

Unity has divided the GNOME community by introducing a new shell on the world’s most popular linux distribution. While it’s true that the state of linux desktop has been moving frustratingly slow for a number of years and that a quasi-open project, funded by a commercial entity with a focus on usabilty and æsthetics — exactly like Unity is on paper — could help accelerate its development and help reach parity with the two main desktops in some of the more difficult areas where linux has been falling back over the years. Still, Unity is largely incomplete, it’s missing many of the configuration options and functionality that linux users are used to — nay, demand — and, sadly, what’s there betrays a rushed, badly designed feature set that should never have gone past alpha inside Canonical, let alone be part of the world’s most popular distribution.

There is a saddening shortage of proper MySQL administration/query browser tools on linux-based systems. MySQL Workbench is a free tool that consolidates what used to be MySQL Query Browser and the MySQL Administrator and introduces a data modelling editor that promises round-trip design and generation of DBs.

It’s great news that Workbench is being implemented and made available to the public for free, yet I can’t avoid comparing it to some of the existing solutions for other platforms.

While it’s easy to justify why Query Browser of old sucked — it was an old application that was being marginally maintained in the last several years — it’s hard to do so for a new piece of software like Workbench. Its query browser sucks, not just because of its implementation (slow, occasionally crashing, often providing inconsistent/misleading data), but also because it betrays bad design at every corner; design of the sort that disregards usability and tries to shoehorn user interaction to a flawed model chosen because it suited some developer during implementation.

The Workbench developers seem open to suggestions, and in this light I can only provide a concise piece of advice for their query browser development effort: find a Mac, download Sequel Pro, use it and then copy the damn thing: It works, it’s many times faster, more usable, covers more use-cases and is much more painless, plus it’s free and open source. Sadly, it’s not available on linux, for if it did, I wouldn’t touch Workbench with a three point oh-five metre pole.

If anything the sudden appearance of Moblin 2.0 Beta and its excellent User Interface has proven, beyond any doubt, that corporate support is essential if linux — and the open source community — is going to survive beyond a very very small niche.

Linux on the server has been doing well despite Microsoft’s pretty good record with Windows Server in the past few years (and contrary to its failings with Vista), its dominant position on the desktop and its proven marketing muscle and the reason for this has been that Linux on the server has had the support of many large corporations living off it.

This has not been the case on the linux desktop, and it is probably the reason why so little has been achieved in the past seven or so years in that field. On one hand, the stagnant Gnome 2 platform barely kept alive primarily by Redhat and Novell that depend on it and on the other the interesting and fresh KDE4 platform that’s extremely immature and incomplete and leaves thousands of everyday use cases unsupported.
»

“…And what can we do about it”. Linux usability (and the sorry state of desktop linux) has become a staple of this blog, but bear with me for a bit. Here’s a video by Bryan Lunduke from the Linux Action Show with reasons why the linux desktop still sucks for many (most) users. This comes from someone that like linux and wants to see it succeed; most of the stuff mentioned is pretty valid criticism that touches upon the lack of cohesion, regressions, QA and many other aspects of modern linux distributions.

Moblin 2.0 is a netbook-optimized linux distribution/environment originally created by Intel and largely based on the work by OpenedHand, a startup bought by Intel in 2007. This is the introduction video to Moblin. And, as far as linux goes this is by far the most advanced, well-thought, usable environment I’ve ever seen.

Sure, it may be tailored for netbooks, which means the task of creating it was much simpler than creating a beautiful, usable and functional full-fledged desktop environment; and of course, from a developer perspective Hildon and GTK are not exactly ‘modern’ or well documented, in contrast to — say — modern Cocoa. Still, it’s a fantastic first step in the right direction and shows the promise of what focused work can do to bring FLOSS closer to the state of the art in those areas where it sorely needs improvement.