• Subscribe

    Subscribe to This Week In Panospace by eMail.
    Subscribe in a reader
  • License

    Creative Commons License
    This work is © 2008-2012
    by Yuval Levy
    and licensed under a
    Creative Commons License.
  • Entries

    January 2019
    M T W T F S S
    « Dec    
  • Archives

Not all ‘buntus are Born Equal

Precise Pangolin, or 12.04 is a long term support (LTS) release.  One might think that particular care has been put into it.  11.10 (Oneiric Ocelot) has served me well during my first year in law school, but I thought that going on an LTS version will serve me better for the next two.  Wrong.

Contrast and compare Ubuntu 12.04 and Kubuntu 12.04.  When I boot my backup workstation attached to a 47″ FullHD TV with the installer CD, Ubuntu installs just fine, it recognizes the 1920×1080 resolution of the display and gives me a usable desktop.  Kubuntu in contrast set it up with tiny fonts that are not even readable with a magnifying glass.  Moreover, I don’t even get to update the settings: the second or third window or dialog no longer open / displays, and the old venerable 3GHz Athlon X2 is slower than a snail.  The same machine that runs smoothly with Ubuntu.  Both on the same 12.04 LTS basis.  What is wrong?

One of my keyboards is a Logitech diNovo Edge.  Bluetooth.  A long standing bug has not been fixed yet, so I must install with another keyboard and then pair the diNovo.  Up until 10.04 it used to work perfectly without the need for such workarounds.  The bug seems to be fixed in the upcoming Quantal Quetzal release, but hold your breath if it will be fixed in the long term supported Precise Pangolin, released less than six months ago.

My laptop’s wireless used to work.  Until a bug broke it.  I found a workaround: blacklist the kernel module (i.e. disable the driver in Linux-speak) that was too greedy and grabbed for himself the wireless card even if it was not able to drive it.  The bug is fixed in the upcoming Quantal Quetzal release, but the developer asked me if I needed a backport (i.e. a fix in previous versions of ‘buntu) and to mark it as invalid if I don’t.  Since I have the workaround with the blacklisted module, I would lie if I would say that I need it.  The point is not whether I need it or not.  The point is: is Ubuntu 12.04 really supported until 2017 and what exactly does “support” mean, if obvious bugs that have been fixed in the development version are not fixed in the “supported” versions?

Under such circumstances it is difficult to recommend any ‘buntu version.  Those are situations that the ordinary user should not be confronted with.  Upgrading?  What was I thinking…

Tipping Point

Embarrassing:  Monday morning I came to class, opened my notebook’s lid, and loud music filled the silent room.  I was caught by surprise.  It took me three long password entry attempts until I unlocked the screen and could hit the mute button – “Fn F8” did not work in locked screen mode.  In the meantime Tim Simenon and Mark Saunders remix of Strangelove, from the last Depeche Mode release, digital download, pumped up the beat.

Ex-post analysis: the previous evening I had my earbuds on while working on my exam summaries.  My last actions have been to close the notebook’s lid and unplug the earbuds, in that order.  Kubuntu remembers different audio settings when earbuds are plugged.  Unfortunately the sleeping notebook did not register the unplugging and on waking up from sleep did not bother to check if something has changed.  This makes the difference between a useful feature or a useless one; and between a smooth experience and an embarrassing one.

This was also the moment I decided that my next personal computer will be a Mac.  I will make the switch at the next opportunity and will relegate Kubuntu where it belongs: in the toys box.  I am thankful to Kubuntu for the great playtime, but I no longer have time to play.   I need a machine that delivers quickly and predictably.  I need to be productive.

Initially I thought I could get through law school using Kubuntu.  Indeed the only thing that I could not make work was the exam writing software.  Everything else I got working, including access to the library’s printing infrastructure, the university’s email system and knowledge base, and document interoperability with my Mac and Windows using friends.

However, this came at a cost I can no longer afford: time.  Three month into it, I am calling the experiment off.  There is no major problem with Kubuntu, just a lot of papercuts – little issues that can mostly be worked around but cost me more time than I can afford.

When I started school, in September, I had set up both my workstation and my notebook from scratch.  I bought a new hard disk for the desktop and a new solid state drive for the notebook.  I installed a clean Kubuntu 11.10 from scratch when it was still beta but quite promising.  I needed a dependable infrastructure.

My conclusion after three months (and after years of general ‘buntu usage) is that Open Source desktops do not qualify for the “dependable” qualifier yet.  And I think that a purely free organization based solely on the principles of Open Source is not well equipped to make a dependable desktop.

A few examples:

I have a Brother all-in-one inkjet and a Brother laser printer.  On my first day at law school they were both fully functional.  During the past three months, I have lost twice use of the printing functionality following updates of the cups package.  Once only the desktop was affected.  Once I also lost the ability to print to a PDF file.  I was able to re-install and re-configure things.  Not a wise use of my time when approaching a deadline for a paper that had to be submitted in print.  I also lost the scanning functionality, on the notebook but not on the desktop.

As a law student, I mostly deal with text.  LibreOffice stopped scrolling horizontally.  I can move the cursor inside the window, but it is time-consuming.  In October LibreOffice Writer stopped displaying formatting characters.  The button in the toolbar would not toggle them. Finding the workaround was time consuming, and the issue reappears spuriously.  I can work around it going to the menu Tools -> Options -> LibreOffice Writer -> Formatting Aids and checking the checkboxes.  Every time I have to do it I am tempted to save my time and buy a Mac.

At some point, gimp did not start.  I found the workaround online.  It only fixed it for my notebook, not for my workstation.

And of course I do email.  Kmail has been updated and now it only saves email addresses and not the names associated with them.  It is a major time waster to manually connect semi-random sequences of letters and numbers to actual names.  I even took the time to file a bug report.  I did not have time to migrate to a different mail client, but I am seriously considering it, especially since I get a whole bunch of cryptic errors messages from Akonadi and Nepomuk, usually just when I do not need it and they cover the area of the screen where I am typing class notes.

This duo of underlying technologies managed to fill my new 400 GB Kubuntu partition on the desktop and crash it with a disk full error.  Now I only use webmail on that machine.  Akonadi is also terrible at dealing with mobility – when I move between home and campus, or even between different campus locations, I will not see new emails until I kill Akonadi and its spawned processes and restart Kmail.  This alone is so complex that I had to write a script for it.  Why can’t Akonadi kill its spawned processes?

Another papercut I bump into constantly is the interference of the touchpad while typing.  Ubuntu has a setting for it but I have not found anything equivalent in Kubuntu.  Disabling the touchpad while typing (i.e. for about a second since the last key stroke) should be default on notebooks in 2011!

I could go on listing plenty of such small issues.  Individually they may be small and insignificant, but together they consume a lot of my time.  And I am learning that listing them is not helping me, which brings me to my conclusion that a purely free organization based solely on the principles of Open Source is not well equipped to make a dependable desktop.

Unless there is somebody who thinks of the user’s experience, end-to-end, integrating all the different bits and pieces of Open Source code into a cohesive unit that makes sense, Open Source desktops will be a waste of time, like this bug report.

In the coming weeks I need to focus on exams, and I will work around the papercuts until then.  But it is most likely that 11.10 will be relegated to be second (or third) in the boot up pecking order of my desktops soon.


This blog has been on hiatus for most of the summer.  And it is likely to stay so for an indefinite time.  Or maybe just change its frequency and focus.

Not that there was nothing to say – there was simply no time to write.  The summer went by extremely fast.  On the personal side loads of joys and satisfactions from the next generation growing up fast and discovering the world one day at a time.  One sad but inevitable moment was the passing away of my last grandmother.

The Hugin release process has stalled.  I released release candidate 4 three weeks ago but don’t consider it of good enough quality to be declared final.  I announced at the beginning of the cycle that September 5 is my last day of contribution for the foreseeable future as I am embarking on a new and exciting path: law school.  Hugin will do fine without me.  Maybe somebody will pick up where I left; or maybe 2011.2 will be one of those branches that dry out without blooming into a full release.  Trunk development has already moved on.

In August I suffered two hard disk crashes within less than two weeks, losing (temporarily) my netbook and my workstation.  In the recovery process I did not reinstall a Hugin development environment – a convenient way to focus myself on my new commitments.

Also in the recovery process, I learned a lot about setting up and optimizing a Solid State Drive (maybe I’ll write about it later).

I reinstalled Windows.  The exam software at my school works only on Windows or Mac OS X.  Since I already own an unused Windows license that came bundled with my netbook, I might as well give it a try.  If it is not good enough I may buy a Mac.  Student’s life.

Unintended consequence:  I also reconsidered my desktop environment.  Before the crashes my workstation worked well with Kubuntu Lucid 10.04 and my netbook worked well with the newer Kubuntu Natty 11.04.  Since I won’t have time to deal with the unexpected I decided that it is too early for me to jump on the Oneiric cycle and settled for Kubuntu Natty 11.04 on both the workstation and the netbook.  Or so I thought.

The netbook was no surprise.  It ran Natty already before the hard disk crash and on the optimized SSD it runs smoother than before.  Ready for school.

Kubuntu 11.04 on the workstation is an unexpected disappointment.  When more than a few windows are open, the next one will open black and display its content only after downsizing.  As the number of windows grows, the maximum size at which the next window displays shrinks.  I never encountered this phenomenon with 10.4 – same hardware (dual display and nVidia 6150 integrated graphic – this is a very old setup) and same drivers (nVidia proprietary).  I quickly replaced KDE with Unity and the problem is gone.  Extra bonus:  Unity is much faster and responsive than KDE.

KDE has served me well for the past two years.  I find KDE’s functionality works better for me than Gnome’s intentionally designed limitations.  Dolphin is a true help to access my files, while Nautilus feels more like an obstacle than a help to me.

I am aware that fully fledged desktops such as Gnome or KDE are more  taxing on computing resources than lightweight ones.  The trade off is acceptable to me as long as the system is functional and responsive and the overall user experience is fluid.  I was not aware how bad KDE compares to other Linux desktops in terms of resource waste; and how bad they all compare to to Windows.

So why not try to use Windows as my main desktop?  Honestly I find the Windows 7 experience to be better than its reputation.  Smother than my four years old memories of Windows XP.  Smoother than Ubuntu/Gnome 2.x that I used when transitioning.  But also smoother than the KDE 4.x that I am using these days.  Windows Explorer is as useless as Gnome’s Nautilus.  On the positive side bluetooth devices work seamlessly and drivers installation is no longer a painful disk-shuffling.  Microsoft has done its homework.  And yet the bare Windows 7 still feels hollow and devoid of functionality.

The two things that I am missing most in Windows 7 are

  1. An equivalent to the Debian package manager to enable easy access/addition of apps/functionalities.  I guess they will call it an “app store”.
  2. A complete toolchain to empower writing/modification of software and harness the power under the hood of the PC.

There are ways around those limitations.  A credit card helps.  Using a more advanced system helps too.  Apple delivers a complete toolchain free of charge and with the newest Mac OS X 10.7 Lion iteration there is an app store (although I am not sure about the choice inside the walled garden).

App stores are the future,  unfortunately.  Not because they are bad (one could argue that the Ubuntu repository is a big app store) but because they are being used to create walled gardens and restrict consumer choice.  That’s a battle for another day.

For now, my bottom line is that I am back to dual-booting Windows and Linux; and on the Linux side I need to find a usable desktop environment for my workstation because Kubuntu 11.04 fails miserably where Kubuntu 10.04 worked fine.

Hugin 2010.4.0beta1 released.

Hugin 2010 Logo

Yesterday I released the first tarball in the new Hugin release cycle.  The goal is to release 2010.4.0 before the end of the year.

It’s only a couple of months since the last release, but a lot has changed, in the code, in the process, and in the infrastructure.

I wrote about the infrastructure change three days ago.  The activity in the new bug tracker is massive.

In the code, the most important news is that with its own brand new control points detector, Hugin can be considered feature-complete.

To underscore this achievement, the project has given itself a new look, contributed by Cristian Marchi, that has given an evolutive face lift to the original design by Dr. Luca Vascon whose source files have been lost in time.

In terms of process, this time around we have more contributors than ever, on multiple disparate platforms.  The project will still stick to its policy of releasing source code as soon as it is good to go and leaving it up to the user communities on the different platforms to produce and distribute binaries because it does not make sense to delay the release of source code only because there are no binaries; and it does not make sense to delay the release of binaries for a platform with faster builders only because there are no binaries yet on other platforms.  However the natural and inevitable time lag between a released source package and a working binary package (which is what most users are looking forward to) is likely to be reduced for most platforms.

First to respond to my call was Matthew Petroff.  He made Windows binaries in four variations (32bit/64bit, installer and standalone zipped) available within a few hours and before anybody else.  Matthew has joined the team recently and he has done some excellent polishing work on the Windows side of thing.

Then the indefatigable Harry van der Wolf followed up.  Building for OSX is always a little bit different/special and require more effort than most other platforms.  He reported this morning that everything works and will produce the coveted bundle installer tomorrow.  What would Hugin Mac users do without Harry?

Andreas Metzler reported a “work for me” update to the Debian experimental source package.  Based on his work I will try to produce my first Ubuntu packages of Hugin for Ubuntu Lucid (my main system) and Jaunty (chrooted), and Gerry Patterson will tag-team for Maverick.

On the Fedora front things are quiet but not less up to date.  Between Bruno Postle and Terry Duell recent versions of Fedora should be covered soon.

Lukáš Jirkovský will try to use OpenSuSE Build Service, but he’s very busy and there is no guarantee.

No promises.  There is always an inevitable lag between the release of a source tarball and that of a usable binary – at minimum the time it takes for the builder to download the tarball, build it, run a minimal test, and publish it.  But we are doing our best to make this the Hugin release with the shortest delay from source to binary.

This weekend is a test run.  The really interesting run will be when we approach the final release.  Keep your champagne cold for now.

And when will somebody report success building Hugin on Android or iOS?

Update that dcraw in Ubuntu, please.

It has been three months since I tried to process RAW files on Kubuntu. The out of camera JPEGs of my SONY Alpha 850 are so good that I don’t feel the need to process them any further. I can concentrate on the photography and forget the computing. Bliss!

Last week however a friend asked me for some “magic” (his word for tonemapping), and for this I need the RAW. Before rebooting into Windows I looked again at my problem with the washed out  RAWs. Since I did not have much time I installed a prebuilt “daily” snapshot of RAWstudio, courtesy of Anders Kvist:

$ sudo add-apt-repository ppa:rawstudio/ppa
$ sudo apt-get update
$ sudo apt-get install rawstudio-daily

The RAW files are shown correctly! What’s the difference from my self-built binaries from SVN? Since most RAW converter use the same dcraw code, maybe…? maybe!

So I tried with LuminanceHDR and at first it still looked off-color:

A quick check revealed that Ubuntu 9.10 ships dcraw v8.86 which fails my SONY ARW files. Current version as of writing is v8.99. With brute force (and in five minutes):

$ sudo apt-get install gcc libjpeg62-dev liblcms1-dev
$ wget http://www.cybercom.net/~dcoffin/dcraw/dcraw.c
$ gcc -o dcraw -O4 dcraw.c -lm -ljpeg -llcms
$ sudo cp ./dcraw /usr/bin/dcraw

And here we go!

Bonus: LuminanceHDR recently introduced  Mantiuk08 TMO (and new dependency on the GSL).

Why do Ubuntu 9.10 (October 2009) and the upcoming Ubuntu 10.4 (April 2010 and meant to be a long term supported version) ship with an outdated v8.86 (April 2008)?

Most Linux software that open RAW images depend on dcraw. LuminanceHDR and Krita are just two examples I care for. 18 months are eons in digital photography. Not updating means not supporting many of the new cameras/features that are released continuously. Dave Coffin does a great job at keeping up to date his code to read their files. Where is the bottleneck?

Between v8.86 and v8.99, support was added for many new and popular cameras that appeared on the market during that period, including SONY Alpha 850, Canon EOS 7D and 5D MkII, Pentax K-7, Nikon D700 and many, many, many more.

I don’t understand Ubuntu. On the one side they are totally bleeding edge. Sometimes even too much for my taste: they unleash on the average user software that is IMHO not ready for prime time such as grub2 in 9.10 that can barely deal with dual booting Windows but fails in many situations where the venerable legacy grub works just fine and does not need to be replaced any time soon (if it ain’t broken, don’t fix it; and if you want to promote the new kid on the block, at least give users an easy way back to the tools that are known to work). On the other side, they are totally outdated. Like with a two years old version of such critically important software as dcraw. At least they made some progress with Hugin: 10.4 will ship with Hugin-2009.2.0 while the default for 9.10 was still 0.8.0.

Next, I would need an intro on how to make packages from the binaries I made; and information how to submit them to Ubuntu for inclusion in the next release. Too late for 10.4, but things can only get better for 10.10 and later. Anybody care to help?

And here is the “magic” that my friend wanted to see:

out of camera JPEG (resized)

dcraw / ImageMagick B&W

LuminanceHDR / Mantiuk08

LuminanceHDR / Fattal, desaturated

Which one do you prefer?

Digital Photo Professional Linux

It’s been a while since I installed Canon Digital Photo Professional on Ubuntu / Wine. In the meantime, there have been upgrades to Ubuntu, Wine, and Digital Photo Professional. The good news is that the installed v.3.2 no longer shows the background quirks.

I’ve tried to upgrade to the most recent DPP update (v.3.7.2) as well as the last one supporting Windows 2000 (v.3.4.1) on Ubuntu 9.10 / Wine 1.1.26.dpp3.7.2_install_errorYou can see the error message above. Given the popularity of my previous post on this topic, I expected it to be known, and I found the bug report in Wine’s bug tracker. Go to the bug report and vote for it to be fixed!

7: the Eighth Capital Sin?

windowsMicrosoft launched Windows 7. They must be relieved, they got it out a few days ahead of Ubuntu 9.10. Apple comfortably released OSX 10.6 Snow Leopard at the end of the summer. The line up of mainstream desktop operating system for the autumn 2009 is complete (Ubuntu being a representative for the desktop Linux/BSD).

And how is vintage year 2009? Stagnating, mostly. Sure, it’s nice to see Ubuntu making it easier for users to encrypt their valuable data; but all contenders are still stuck with the windowing metaphor, a 1975 legacy of PARC. Boring.

And when the capital P (as in Product) no longer differentiate the competition, marketing kicks in: Product, price, place, promotion, … and public opinion.

Nobody can beat Ubuntu’s price. Not only it is free (as in: zero price tag), it is Free with a capital F (as in: no vendor lock-in).

Apple and Microsoft played opposed pricing strategies, with Apple offering Snow Leopard at a corrupting low price (probably the cheapest Mac OS ever), and Microsoft going the opposite way, making 7 probably the most expensive Windows desktop OS in human memory and attracting a lot of critique from market analysts.

Place? the Internet is ubiquitous.

So we’re left with promotion. And indeed the marketing budgets seem to have swollen. Users drive Ubuntu for themselves, so marketing is just the measure that the users really need.

Apple’s formidable propaganda machine has a history of making its followers believe all sorts of things. Such as “the PowerPC is more powerful than Intel’s CPUs”. And a few weeks later that “Intel powered Macs are twice as fast as the previously PowerPC powered generation”. There are rumors their next campaign will convert cats to vegetarians. If only they used this skill to promote peace on Earth!

Microsoft’s marketing was traditionally more clumsy. It’s performance is a refreshing surprise. Besides the revamped websites, they copied a few tricks from Apple’s storybook (e.g. retail shops). And like Apple, Microsoft is now aggressively pursuing university students. Get them hooked early. During a limited time U.S. students can buy Windows 7 for only 30 USD. In the U.K. it’s 35 GBP. In France it’s 35 EUR. The offer runs in a dozen countries, Canada included (40 CAD).

I had decided that XP was my last Windows. I’d extend its life until Ubuntu was good enough to replace it for every application. With a few exceptions (video editing!) it works well for me and since four months I’m spending most of my computing time in Ubuntu.

But my wife is a student eligible for the discount. Have I committed the eighth capital sin? The download is still running (used wine to run Digital River’s download manager).

Warning. If you use this offer you are downloading from Digital River. Beware. They try to add an extra charge with an extra “service” that bear resemblance with the bad value for money of the extended warranties in electronic shops. Avoid wasting money on that “backup service”. Buy a few empty DVDs for a few cents and make as many backup copies as you want.

And what has public opinion (or politics) to do with all this? Most western countries are running scary budget deficits because of the stimuli introduced to counter the global recession. Money is tight. How about saving some (and creating local jobs) by replacing current systems, most of them based on Microsoft products, with Free systems?