So, somewhat miraculously, I've been doing a little KDE hacking again this week for the first time since, oh, 2006 or so (aside from TagLib, which recently moved to GitHub).

Being as I'm all a startupy web weenie these days, some of the design sins of my youth have been haunting me. I wanted to give JuK a little bit of a fixer-upper. Almost all of the 20 commits that cia.vc says that I've done this week stem from either polishing the interface some or improving the initial experience on startup.

The major bits are:

• Made the selected playlist thing nice and antialiased and gradiented and shiney. That had been bugging me for ages.
• Ganked the slider and volume control code from Amarok and and popped them into JuK.
• Cleaned up the tag editor (still not shown by default, but in the screenies below for the before-and-after).

We'll probably be sacrificing a few of the more broken features on the alter of polishedness in the next bit trying to get a nice spiffied up version into KDE 4.7.

Here's the before and after:

I'd love to someday have time to go all OCD and just spend like a couple weeks fixing things like text alignment and margins all over KDE. I tend to look at apps like I would web pages at this point, and such things pop right out at me at this point. Alas, I reckon I'll probably be back to the grind of work here pretty soon, but it's been a fun nostalgic interlude.

I thought this might be interesting for some folks in the KDE world -- for work stuff we needed a fast implementation of Ruby's ActiveResource, so I wrote a Qt / C++ ActiveResource consumer. The performance relative to the default Rails backend is somewhat telling:

It gives a nice QVariant-filled API for working with Rails web services APIs. Full story is here and the code up on GitHub.

Lukáš, who's taken over TagLib maintainership these days, has just released the latest bug fix release for TagLib, also posted in his blog:

Changes from 1.6.1 are:

• Read Vorbis Comments from the first FLAC metadata block, if there are
  multipe ones.
• Fixed a memory leak in FileRef's OGA format detection.
• Fixed compilation with the Sun Studio compiler.
• Handle WM/TrackNumber attributes with DWORD content in WMA files.
• More strict check if something is a valid MP4 file.
• Correctly save MP4 int-pair atoms with flags set to 0.
• Fixed compilation of the test runner on Windows.
• Store ASF attributes larger than 64k in the metadata library object.
• Ignore trailing non-data atoms when parsing MP4 covr atoms.
• Don't upgrade ID3v2.2 frame TDA to TDRC.

As a side-note, some of my buddies from RethinkDB have been looking hard for good C++ folks out in Silicon Valley doing a MySQL backend for solid state devices. If you're a systems-y C++ wonk in search of a job at a hacker-friendly company, they're good folks. If you're interested and we know each other drop me a line and I'll do a little intro show-and-dance.

So, after far too long, TagLib 1.6 is out. I finally asked Lukáš Lalinský, who's been the largest TagLib contributor other than myself and veteran of the MusicBrainz project, to step in and take over maintainership as I've been off doing the whole interwebs startup thing for the last year and change and time is exceedingly scarce of late.

The real highlights of this release are a whole bunch of new formats supported:

• MP4 [compile time option]
• ASF (WMA) [compile time option]
• WAV
• AIFF

The first two are compile time options so that distributions can easily decide if they thing just dealing with the containers / tags of encumbered formats is murky territory. Tarball has been up for a couple of days. Enjoy, report bugs and give a big round of thanks to Lukáš!

I didn't get to be one of the cool-kids at Akademy this year, but it's still a pretty exciting week for me. I won't drone on about it too much, but since I mentioned here a while back that I'd just founded a new company I thought I'd drop in a link now that we're actually talking about what we're doing.

Directed Edge is doing a graph-based recommender system for web sites that we partner with. Basically, we take a page or user as a starting point and find related or interesting stuff. We built a prototype based on Wikipedia's content to show what the system can do. There's still a lot of room to make the system better, but we're pretty excited to get something out there for folks to start messing with.

The announcement, with more details is here.

There are a few scattered updates in the world-o-wheels of late. The biggest of which, as a number KDE folks are already aware is that I'll be leaving Native Instruments, where I've been for the last couple of years and starting my own company with a friend or two rather soon. I'll post a link once we're to the point of launching a public beta. It's not desktop software, and it's not a consulting service, but this will mean that my primary (professional) development platform will be Linux once again.

I've been considering founding a company for a long while, and having recently been granted permanent residence in Germany it's now legally possible. I've been going crazy the last couple of months trying to sort out all of the technical, financial and administrative details that will go into getting that off of the ground. Above is the table next to my bed.

In other news, coinciding with a meet-up for startup founders on the same weekend in Prague, I'll be around the upcoming Ubuntu Developer Sprint on Thursday and Friday. I'll also be around some of the time at LinuxTag in Berlin the following week. There's a reasonable chance that I'll make it out to Akademy this year too. My specialty at conferences seems to be taking embarassing photos, so I'll try to do my worst.

My current employer is also in the process of switching over to using TagLib and so last week as one of the tasks that I wanted to finish up before I'm away from there I implemented, per request, tagging and audio properties for AIFF files. More or less for free along with that came a generic parser for RIFF formats. There's been a lot of traffic on the TagLib development list of late, and once life slows down a little bit it looks like it might be time to do a quick 1.5.1 or 1.6 release and then go for a 2.0 push.

TagLib 1.5 is out.

As always, file any bug reports that you happen to run into in the bug tracker. As there are specifically a couple things that I intend to implement (wav / aiff support as well as support for ID3v2 tags in RIFF chunks) I expect a 1.5.1 (or 1.6) to be much faster in coming around this time.

I'd like to give a special thanks to Lukáš Lalinský for the numerous bug fixes, testing and new features, Urs Fleisch, Aaron VonderHaar for their ID3v2 frame implementations and of course Michael Pyne for helping keep the bug list in check.

The TagLib 1.5 RC is up. There have been a huge number of changes since 1.4 (two years ago) and even a number of changes since last week's beta.

I've also updated the documentation on the web server, put the new sources up and also put up a Mac OS Framework. The real release (or shortly thereafter) will also contain a Windows build as this is the first TagLib release to officially support Windows as well.

Major (file corruption, crashes) or trivial bugs may still be fixed fixed before the release. If you're using TagLib in a project please consider taking some time in the next week to try out the new RC, valgrind your app with it, etc. If no major issues are discovered within the next week in approximately one week this will be renamed to TagLib 1.5 and released.

I'll prepare a change log some time before the real release (which I'll do another annoucement for).

Doc suggestions should go to the mailing list (or comments), bugs to the bug base.

• Source
• Docs
• Mac OS X Framework

So there's a lot of speculation floating around about the recent Nokia acquisition of Trolltech. There will be a lot more information to unfold in the coming months. The first thing I noticed was the price tag. Around €105 million. (I'm going to convert all Norwegian Kroner values to Euro since that's easier for me and most readers to think in.) That seemed low, based on some nebulous not-grounded-in-anything, notion of what I supposed Trolltech was worth, so I did a little digging.

Trolltech has been publicly traded since the middle of 2006, which means that they've been issuing performance reports. The first thing that hit me when I started sifting through them was to note that Trolltech has been losing money for the last three years. Looking over to the stock market, when they went public there was a fixed share price of €2. A year and a half later they're worth about half that. Nokia, interestingly, agreed to pay double the market rate for the shares (incidentally, exactly the same price as the IPO). Trolltech has around 50 million shares, with about 17m held by board members (mostly the two cofounders), 14.5m by investment firms and 8.5m on the market. (I didn't find documentation for the other 10. This may be related to the 50 million being a current figure and the last report of share holdings being from the end of 2006.) Trolltech's total revenue is around €25 million. Total losses were around €6 million for the last couple of years. Total cash on hand at this point is around €13 million. They've got around 250 employees worldwide.

So, those are the facts, now I'll get into a mix of extrapolation and wild guessing (and please note, that I'm definitely not an expert in this stuff):

There are reasons that a company will budget to run at a loss for a period of time to try to grow the business. Some of this was probably intentional. However, given the consistently decreasing stock prices, it seems that at least investors were not wholly convinced that it would bounce back to profitability in a timely manner. Revenue has been growing at a pretty consistent 40%, but in 2005 and 2006 costs doubled. Things leveled out a bit as of the third quarter of 2007, they'd only increased by 30% relative to the previous year. In a nutshell, Trolltech was hardly in the toilet, but performance wasn't great either.

Eirik Chambe-Eng (14% ownership) resigned this year and Haavard Nord (14% ownership) is based in California, and I would imagine for both of them, at this point in their lives, Trolltech fits differently than it might have a decade ago. Another 29% of the company is more or less purely interested in profit (investment groups), and you see, Nokia gave them a pretty sweet deal (again, double the market price). Being a public company and all, one would assume that at some point the founders would sell their stakes, and this way they got a much better deal than cashing them in on the market by converting them to common shares. If they had both sold their shares on the open market, with then the majority of the shares then outstanding, Trolltech could still have been bought out, via hostile takeover or otherwise.

So, Nokia. One thing to put out there, data-wise is scale. The total cost of the acquisition of Trolltech is about 0.1% of Nokia's value. At current levels Trolltech would represent 0.05% of their revenue and 0.2% of their workforce. (Compared to, say, Novell / SUSE, where value and number of employees were about 10%.) One thing to take from that is that this probably isn't that huge of a deal for Nokia. Their profits for this week will cover the Trolltech acquisition.

What that says to me is that it's really hard to speculate on what this means for Nokia. Again, comparing to Novell, where the acquisition of Ximian and SUSE signified a major shift in corporate strategy, I don't suspect this will be a major upheaval for Nokia. Qtopia could become their answer to Android, Trolltech might become their WebKit gurus, who knows. Nokia may even have a few ideas they're planning to try out to see how to integrate Trolltech's technology into the company and keep the ones that work, and write off the others.

One of the most interesting things to me is that Nokia will not have the need to run Trolltech as a profit center (though they may). Nokia makes enough profit every 2.5 hours to cover Trolltech's current annual losses. A really interesting thing to watch will be if and how over the next year or so Nokia attempts to market and place Qt (as opposed to Qtopia and WebKit). In some purchases, especially with this big of a ratio in scale, the resulting division is left more or less intact, in others, there's an attempt to bring them closer to streamline them to fit the company's overall goals. I could imagine, at least with the Android scenario of Qtopia, Qt potentially being licensed under a free-as-in-beer license as a counter to Android.

As far as ginormus organizations go, Nokia's been pretty good about working with the OSS community. We'll have to wait a few months for some more cards to be on the table before it's really clear if this ends up being a good or a bad thing. Like most changes, it'll probably have elements of both.

In the world of things completely unrelated to KDE...

My life has been rather concentrated on music the last few months (including the various small-ish will-someday-be-released OSS things that I've been hacking on of late). One of the things that came up yesterday in a jam session was that based on the group's name, which contains a reference to Spanish, is that it would be fun to have a collection of samples for use in our set, with various voices in various languages saying:

"Do you speak Spanish?"

So, here's the call, dear blogosphere, send me a wav, mp3 or ogg to wheeler@kde.org with you asking that question in your local language / accent, and let me know where you're from / which language it is. Bonus points for additonal samples from girlfriends, boyfriends, housemates. If it pans out and there's enough of a selection to do something musically interesting with, there's a fair chance that your voice will be ringing out over the PA system at some dirty techno club in Berlin within a few months. :-)

Laptop Battle v2.0

In other music related news, Laptop Battle Mannheim and Laptop Battle Stuttgart are both on the horizen and still taking demos for contestants. (I was helped out a bit with the Mannheim one last year and will be on the jury again this year.) If you happen to live in those areas and are produce electronic music, consider sending a demo in!