Sunday, September 25, 2011

Poll: KDEPIM 4.7 - ready for you?

If you have read my previous post, you will already suspect that we may not be able to support the "old" KDE PIM applications (i.e. kdepim-4.4; the KDE PIM main applications are akregator, blogilo, kleopatra, kmail, knode, knotes, kontact, korganizer, ktimetracker) much longer. Problem is, most data on the usefulness and stability of new Akonadi-based KDE PIM (i.e. kdepim-4.7) is just based on anecdotes. So, to get at least some half-way representative numbers, here's a poll for you. As Gentoo has been offering the new software version all from the start, please tell me your opinion on the new KDE PIM (click on this link if the poll does not show up):

Thanks a lot in advance!

Who cares about users and distributions anyway?

As distribution developer, some of our most important tasks are
  • making packages work together nicely
  • and selecting "stable" package version sets for a broader audience
On the whole, this is appreciated also by "upstream", i.e. the software authors and the development communities. Every now and then, however, things just become slightly crazy, leading first to frustrated Gentoo devs, then to a lot of frustrated users.
  • Bad example number 1: Amarok.  Do you still remember the note that "you may lose your database on stable upgrade from KDE 4.4 to KDE 4.6" in the KDE 4.6 upgrade guide? Well, that note was the end result of quite some discussion, and (at least as far as I know) Debian stable users ended up with the same database mess. So what actually happened here? Simple, KDE 4.4 was for the Amarok developers so far in the past that they did not even consider an upgrade path. Every complaint that the upgrade was broken got a boilerplate response "The problem is fixed in the new version", although what upstream actually understood as fix (and may indeed have been an improvement) introduced the upgrade trouble in the first place...
  • Bad example number 2: KDEPIM. Right now there are very different opinions on whether Akonadi-based KDEPIM-4.7 is already ripe for mass consumption. Well, you'd say, if there are any doubts about that, how about at least fixing regressions and keeping KDEPIM-4.4 in a working state for a bit more time? Alas, the old version is now officially declared unmaintained, and the first bad regression has appeared with KDE-4.7.1. Discussion on IRC with one of the KDEPIM developers led me to this exchange...
    [12:42:12] <dilfridge> xxxx: re bug 279432 - please, please reconsider and have at least a look at it... we still have users with a lot of problems with kmail2, and it would be great if we could support kmail1 / 2  in parallel for a while
    [12:42:12] <bugbot> KDE bug 279432 in kontact (mail) "Error while uploading message (unexpected end of data) when applying filters" [Normal,Resolved: unmaintained] http://bugs.kde.org/279432
    [12:44:01] <dilfridge> I can help testing since I see the problem myself
    [13:30:29] <xxxx> dilfridge: you can help by writting a patch for it
    [13:30:58] <dilfridge> indeed, I can... problem is that I'm not familiar at all with the code
    [13:32:09] <dilfridge> I would not even know what parts are interacting there... kdepimlibs, kioslaves...
    [13:40:31] <xxxx> note that using online imap with kmail1 never worked correctly
    [16:22:02] <dilfridge> xxxx: maybe, but I've never used online imap and have the same problem :|
    [end of conversation]
    I talked to a colleague about that and got the response that in addition to focussing on KDEPIM-4.7, "the KDEPIM guys are now paid for doing the mobile version, desktop comes second". WHAT? Something's seriously warping priorities here, I mean, I like to sometimes check my e-mail with my mobile, but without a working desktop mail application I would probably not even have an e-mail account!
Summary of all this rant: If you're an upstream developer, please remember that what you consider stable is probably not the most conservative, end user-friendly estimation. People are using older versions, distributions are packaging older, "ripened" versions, and also these need some minimum support and upgrade paths! Otherwise, your users might just go away...

Thursday, September 1, 2011

PhD position available: Transport spectroscopy and theoretical analysis of few-carrier systems in carbon nanotubes

We're currently planning a research project in close collaboration with the theory group Prof. M. Grifoni, with working title "Transport spectroscopy and theoretical analysis of interacting few-carrier systems in semiconducting and small-bandgap carbon nanotubes". It combines equal parts of experimental work and theoretical data analysis and modelling. You've already done an excellent solid-state physics theory Diploma or MSc thesis and liked it, but would like to get your hands dirty as well? Then you're maybe the perfect candidate!

Interested? Please have a look at the PDF file with more details, at our web pages (group Prof. M. Grifoni, group Prof. C. Strunk, group Dr. A. K. Hüttel), and contact Andreas K. Hüttel (e-mail: andreas.huettel@physik.uni-r.de) for more information!

PRL accepted: Universality of the Kondo effect in quantum dots with ferromagnetic leads

I'm very glad to be able to report that our manuscript "Universality of the Kondo effect in quantum dots with ferromagnetic leads", describing results that we've been working on during the last months, was just accepted for publication in  Physical Review Letters.
So what is it about, in a few simple words?
In general, much of our work is about charges trapped inside carbon nanotubes at very low temperatures (0.05K). Such a trap for e.g. electrons is called a quantum dot; similar to the electron shell of an atom or molecule, the laws of quantum mechanics force the electrons to occupy specific discrete levels, or quantum states. By looking at a tiny tunnel current through a quantum dot we can characterize its quantum mechanical properties; this is called transport spectroscopy.
The Kondo effect is a special case, as it is caused by strong interaction between localized charges inside the quantum dot and charges in the leads that we attach to the quantum dot. Whenever the localized charge can assume either of two (or more) states with equal energy ("degenerate states") and these states all couple to the leads, the Kondo effect causes an extra electrical conductance through the system. This is one of the simplest many particle effects in quantum mechanics and has fascinated researchers for quite some time; its behaviour is called universal, as it is independent of many detailed properties of the system at hand. In a non-magnetic system, the degenerate states are usually given by different directions of the electron internal magnetic moment, its spin.
Now, we contact our nanotube quantum dot with magnetic contacts. In these contacts, 1) the different directions of spin can couple differently to the quantum dot, and 2) the number of charges with one spin direction differs from the other (that's just what makes them magnetic). Among other things, we've been able to show that all this modification just acts on the Kondo effect the same way as an (imagined) magnetic field, so by applying the reverse magnetic field with an external magnet coil, we can restore the universal behaviour as known from non-magnetic systems. This makes the system much easier to describe, and will, we hope, be useful for future work in spintronics, where the magnetic moments are to be used for information processing.

"Universality of the Kondo effect in quantum dots with ferromagnetic leads"
M. Gaass, A. K. Hüttel, K. Kang, I. Weymann, J. von Delft, and Ch. Strunk
accepted for publication by Physical Review Letters; arXiv:1104.5699 (PDF)

Note: the Wikipedia articles "quantum dot" and "Kondo effect" are not wrong, but describe special uses of these terms and not the most general case as known today. Unfortunately this makes them completely useless as references here...