Some weeks ago, I uploaded a 1.10 pre-release (approximately milestone 5) to Debian experimental. This is the soon-to-be-released new version, using a new level file format (with a much extended API to make level development even easier, ~50% less code per level now), new levels (of course), updated graphics (including support for new graphics modes), ...

Unstable still contains version 1.01; the reason is simple that I knew there would be another 1.01 maintainance release coming. However I believe it doesn't offer much against the current unstable version; it largely marks an upstream release containing patches already in the Debian package (since communication with upstream is really good).

So I have now two choices: refreshing the Debian unstable package to the "probably last" 1.01 release upstream, or going straight for the 1.10 milestones to give enigma some extra testing.

HP failed in a very subtle way: I had printed the covers, cut them, produced the CDs for them. Then I wanted to put the printed covers into the CD cases.

Despite the graphics being 12cm x 12cm in size, HP managed to print them in 12cm x 11.4cm. Without any notice (or giving me a choice) it had decided to scale them on the y axis. Which makes them completely unusable, since they don't fit the 12cm height of the CD case now.

After some more experiments, I decided to retry without duplex, and voila: 12cm x 12cm.

Duplex on HP OfficeJet Pro 8000 is only usable for draft printing, since it will distort your pages!

(See also this devidence in the HP forums, of people with the same issue, an attempt to investigate the margin messup happening, a report that the DJ990c driver can print duplex on this printer without messing with the margins, but is slower and offers less print quality. So it seems that this is an HP driver problem. And technically, it must be caused by the driver; at least it should be able to compensate for this!)

I also noticed another issue with the print. The bottom right corner of the graphic didn't get enough ink, it looks like the printer stopped printing a bit too early. I don't know if this also happens in non-duplex, since I worked around this by adding a header and footer to the page.

Seriously, we should send back the printer. On my first try to use it, I already encountered two bugs. I wonder how many bugs I would see if I'd use it every day?

My current choice:

• Single-shot playback: to view a random song or video I usually open them with Totem (the GNOME default) and that works okay
• Library: I use MPD as player because it just seems to be rock stable. As UI I currently use Sonata, but I don't use it for much more than choosing a song from the currentl playlist.
• Editing: ExFalso seems to have the best ID3v4 support, in particular it also allows multiple genre fields. (Note that Vorbis even suggests you should use multiple artist fields instead of the common "Arist A & Artist B" way of filling the fields)

However, there is one thing I'm really not satisfied with: when putting together a CD compilation for friends (say, as Christmas present), they are quite useless. A key issue here is the total playlist length. Guess what, I want to make sure it fits on a single CD. So I really need to know the total playlist length. Why do so many media players (e.g. totem, alsa-player-gtk, xfmedia4, vlc, mplayer, ...) not show you the total playlist length? They did read all the files to get artist and title. Many even have the individual song lengths, just not the total sum.

In the past I've been using old XMMS1 to check for the total length, or a CD burning application like K3B by repeatedly importing my current folder.

Right now, I'm using Quod Libet (since I like the tag-editing component exfalso a lot) to arrange the playlist. It also gives me the total length, albeit I belive I've had incorrect song lengths in it before (broken VBR files?), and it's not perfect, too: being database-driven it has really long startup times for occasional users (because of updating the database) and is much more heavyweight. I also believe I've lost some playlists because I had moved my files around once ... so I'm a bit sceptical.

Anyway, there are still hundreds of media players I havn't looked at. Don't bother me to send me an email about one I havn't mentioned!

But if you are developing a media player, please consider the use case of putting together a music CD for your friends. In particular, for users that do not use your player all day.

@-moz-document url-prefix(http://www.google.com/search)
{
a[href^='http://www.vitavonni.de/'] { background-color: yellow; }
}

@-moz-document url-prefix(http://www.google.de/search)
{
a[href^='http://www.vitavonni.de/'] { background-color: yellow; }
}

a[href*=vitavonni.de] { background-color: yellow !important; }

This modification obviously only applies to your browser; it's meant to help you finding links to your own site more easily.

The corresponding library - libstdc++-5, and old C++ library - is no longer available in Debian unstable, so you need to grab the package from lenny. It will install fine on unstable.

Things may or may not be different on other architectures.

[Update: But TPTP is far from stable for me. It freezes Eclipse pretty much all the time.]

I must admit that I havn't been very active on Pyroman (or Debian in general) the last years. This goes even so far as that "pyroman" was considered "abandoned" by Fedora or so. It is not; I use it on all my servers. It's still in use at the network I developed it for (after all there is not that much benefit for a workstation setup, where a 10 line iptables script will do the job just perfectly.).

Anyway, I'd like to get IPv6 support into pyroman, but there is one big issue here: I don't have any machine using IPv6, so I havn't used ip6tables myself yet, so I don't know about all the magic involved ...

So if you use IPv6, it would be very cool if someone would jump in to get full IPv6 support into pyroman. Madduck had already done some preliminary stuff, but I didn't get around to have a look at the integration or completeness yet.

The '--no-act' and '--print' modes of pyroman should even allow development without any IPv6 support or root permissions in the system.

Other things remaining on my pyroman wishlist:

• Fully automatic iptables firewall visualization
• Keeping traffic counters over firewall reloads
• Configuration UI
• A fancy 'arsonist' icon and a web page design

To use it, you need at least understand where to put it (call it in a try-catch in onLoad) and how to adjust the variable name of your page tracker (I'm not using the default).

function trackLinks(){
  var as=document.getElementsByTagName("a");
  var ig=["mydomain.tld","google-analytics.com"];
  for(var i=0; i<as.length; i++) {
    var ignore=false;
    var oc=as[i].getAttribute("onclick");
    if(oc!=null){
      oc=String(oc);
      if(oc.indexOf('urchinTracker')>=0
      || oc.indexOf('_trackPageview')>=0
      || oc.indexOf('javascript:')>=0)
        continue;
    }
    if(as[i].href.indexOf("mailto:")<0){
      for(var j=0;j<ig.length;j++){
        if (as[i].href.indexOf(ig[j])>=0)
          ignore=true;
      }
    }
    if(!ignore){
      as[i].onclick = function(){
        var o=this.href.replace(/:\/*/,"/");
        pt._trackPageview('/out/'+o)+";"
        + ((oc!=null)?oc+";":"");
      };
    }
  }
}

This code tries to attach an onload handler to any outgoing link, ignoring internal links or links that use JavaScript. If such a link is clicked, it generates a virtual page access with an "/out/" URL that can be analyzed in Google Analytics.

A side benefit (apart from knowing which links are interesting to your visitors) is that you should get more accurate "time on page" statistics for your pages.

Here's the code I use to fix this:

var r=document.referrer;
if(r.search(/images.google/)!=-1 && r.search(/prev/)!=-1){
 var e=new RegExp("images.google.([^\/]+).*&prev=([^&]+)");
 var m=e.exec(r);
 pt._addOrganic("images.google","q",true);
 pt._setReferrerOverride("http://images.google."+m[1]+unescape(m[2]));
};
pt._addOrganic("maps.google","q",true);
pt._addOrganic("forestle.org","q",true);
pt._trackPageview();

Note that image search is more complicated than the maps and forestle search engines I also add for keyword tracking. The original query is encoded in the "prev" parameter, and the easiest (or only?) way to get working tracking is to use the ReferrerOverride function of analytics.

Note: this is not a straight copy & paste, since I use this code in a compressed and encoded (for injection into the page via DOM ops) form. So no guarantee of syntax completeness. You'll need to adjust it to your variable naming anyway (I use "pt" instead of "pageTracker"). This is just to show you the use of unescape on the "prev" parameter for this purpose.

Say, a regular page should have less than 5% (and less than 20) nofollow links; a site that goes significantly above this value probably employs some spam bot.

The only really hard thing is how to avoid attacks on a site using this ... say, I write a bot that spams links to Microsoft on as many sites as it can find that DO use 'nofollow', in order to get that site above the limit, and have google penalize it.

So in general I don't think Google would automatically penalize such things, still it could be used to e.g. have a human check the destination site for useful content, and then only blacklist when it doesn't seem to be useful.

P.S. Which BTW is a reason why some of the SEO "do nots" are bullshit: it would be too easy to deliberately use these to blacken a competitor. So a 'link farm' will at most do nothing to raise your ranking; but Google must not allow you to actually lower a competitors ranking by setting up a link farm to him!)

P.P.S. On another side note: Who guarantees that Google actually ignores "nofollow" links? They could also just be assigned a lower weight or a penalty, so that a "nofollow" link from a strong site such as Wikipedia would still be worth a lot, while the average blog comments page link goes down to 0. Say a "nofollow" link from a PR 6 site is as much worth as a regular link from a PR 4 site, and PR 2 becomes PR 0. Would already do much of the trick in discouraging the use of blog spam bots. Because after all, ignoring the links on Wikipedia for page rank would be quite stupid. In German Wikipedia, the page contents are even "sighted" (aka: peer reviewed); this is a rather trustworthy source, especially when you take time effects into account. A link being constantly in Wikipedia on a popular page for more than a month very likely is good.

Short story what happened:

• Resumed the system from 'suspend'.
• I copied some files onto the first file system.
• I copied the same files to a second external disk (dual backup...)
• I copied some files from the first disk, which caused an access-beyond-end-of-disk, mounting the filesystem read only
• Unmounted the filesystem, started e2fsck
• Started copying the files from the secondary filesystem
• Got the same error on the second disk.
• Cancelled e2fsck doing more damage to the first disk.
• Shutdown and reboot
• Memcheck, three iterations. Nothing.
• Checked second disk, no errors in filesystem (!), copied the files I had issues accessing just fine.
• Filesystem on disk #1 seriously trashed.
• Had ext2fsck try to recover filesystem on disk #1
• Pretty much all data on disk #1 is now in lost+found, it seems as if all major folders were corrupted. Lots of corrupted file entries (character devices with random permissions and numbers) there, too.

• Reformat disk #1, and restore it from the other backup (Extra backup for teh win! I also have a 3rd copy of about 2 months ago off-site)

As you can see, something was wrong with the system, not with the file system.

I have a strong suspect to have caused this. In case you wondered why I included "resumed from suspend" above: I've been having system stability issues with resume ever since upgrading to the Intel driver 2.9.0 and KMS (Debian unstable+testing) with kernels up to 2.6.31. In about 1 out of 5 resumes, I get a Xorg or system lockup after anything from 1 to 60 minutes. Sometimes I also experience video corruption after a few minutes, trashing some terminal emulation until the next redraw. Just before writing this email I had a typical lockup: when scrolling the terminal emulator. This has been a typical trigger for lockups. On contrast I havn't seen any such crashes (or screen corruption) on a fresh boot.

Freedesktop bug reporting the same issue closed as "not our bug, blame it on the kernel".

Note that 2.6.32 release candidate Changelog contain many changes for the intel DRI kernel driver. So the bug might already be fixed in the RC kernels.

Same report in Kernel Bugzilla is still 'NEW' though.

Related bug report in Debian, blaming it on KMS.

[Update: I've disabled KMS and upgraded to 2.6.32-rc8 and not had such a crash since. But I can't pinpoint it to one or the other yet.]

[Update: just tried another external harddisk ...

[305032.148616] EXT3-fs: mounted filesystem with ordered data mode.
[305066.061708] usb 1-8.3.3: reset high speed USB device using ehci_hcd and address 27
[305081.132471] usb 1-8.3.3: device descriptor read/64, error -110
...
[305147.468857] sd 4:0:0:0: Device offlined - not ready after error recovery
[305147.468880] sd 4:0:0:0: [sdb] Unhandled error code
[305147.468886] sd 4:0:0:0: [sdb] Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK
...
[305147.473500] WARNING: at /build/buildd-linux-2.6_2.6.32~rc8-1~experimental.1-i386-g1b8iG/linux-2.6-2.6.32~rc8/debian/build/source_i386_none/fs/buffer.c:1159 mark_buffer_dirty+0x20/0x7a()

As you might have already guessed, I'm not convinced by Google Wave. It's technically interesting and well-done. The demos are all nice. It's just that the UI in the browser is a bit fragile and cumbersome, and the big question so far is:

What does Google Wave allow you to do that you couldn't do before?

• You can use it for mail, but it only works with other users of Wave and lacks good offline operation.
• It beats pretty much any instant messaging in functionality, but the UI isn't well for running in background. Most IM clients have a great UI for "background" operation.
• Collaborative editing - I prefer having a real editor and real files for that. Check out Gobby for that. I've heard Wave is good for remote brainstorming, though.
• Social networking, read "facebook". Wave doesn't have all the filtering stuff that Facebook is still trying hard to get useful. Just wait until someone releases "Mafia Wars" for Wave ...
• Blogs. Sure, I could do a 'Blog Wave' and invite my friends there. Makes sense for small-audience private blogs; not for blogs like mine where I mostly write to people that I do not know.
• Games. This probably is the current killer app on Wave: Sudoku. Although the (widespread) implementation sucks somewhat. Magnetic Poetry is a nice idea, but doesn't even work in Chrome for me properly ...
• All the web 2.0 stuff just gets on my nerves. I'm not going to use it for my blog; I by design do not have comments on my blog, either. Being able to web 2.0 everything doesn't make up for a lack in benefits.

Yes, I'm aware that you should differentiate between the protocol and the ui. Still pretty much everything is currently designed for the web browser with full JavaScript and Flash capabilities.

Of course this isn't the end yet, Google Wave will evolve. Maybe into something cool, maybe it will remain just a niche thing. Maybe some cool apps will just use Wave as protocol. But I figure, I'll mostly wait for these things to happen first before I become a frequent user of Wave.

The biggest thing I see is the "spam" (this especially includes 'Quiz', Mafia Wars and similar Scamville type of 'apps' that surely will show up in no time, once Wave is open to the public). What will Wave provide to me to handle this flood of worthless information that I'm getting more and more?

P.S. Please don't bother to ask for invitations to Wave.

P.P.S. here's how to replace the odd scrollbars with the regular OS scrollbars with a really simple user style (CSS).

In August 2009, however, the law was changed to exclude beer tents (Oktoberfest ...) and small bars. Many people belive that this was to get votes on the elections in september 2009 (which ended up in a minus of 6-7% compared to the previous election and a historical low for the biggest party).

This caused several organizations to call for a public vote on restoring the smoking ban to the 2008 state (without the 'smokers club' backdoor). In order to force a public vote on a law (without the governments support!), we need 10% of the voters to register as supporters for the vote. You have to register at your registered home town. For Bavaria, this means about 940.000 supporters.

If you are registered voter in Bavaria, please drop by your municipality and sign up. You need an ID and 5 Minutes, that's all. 940.000 supporters is an incredible lot of people to get to the offices, take along your friends!

When we get enough supporters, the Bavarian government has two options: accepting the changes as proposed (and thus making the initative obsolete), or conducting a public vote on it, offering an alternative (e.g. the current law, no change) and have the voters decide (which is quite expensive, so if many many people sign up, they might save that money and just pass the proposed change themselves).

For more information (german only), check the Nichtraucherschutz Bayern Website, including the sign up office locations.

P.S. In other European countries, the introduction of a strong smoking ban has led to a 10-15% decrease in heart attacks (20% for non-smokers). The german constitutional court has also already ruled that the protection of non-smokers and employees from passive smoke weights stronger than the individual's freedom to smoke in enclosed spaces.

However, this is a far from trivial challenge:

Rent in Munich, in particular for conference rooms, is far from cheap. In my opinion, unless we get some really big sponsor (and I'd still prefer spending sponsor money to fund developer trips to the DebConf instead!), the only chance we have is to get some rooms at the university.

However given the development of the recent years (budget etc.), it has become a lot more difficult to actually get rooms at the university for such events. Unless the event is considered to be fully a part of the universitys "work", we might have to pay rent to the university. Which again isn't that affordable.

Anyway, if you are in Munich, working at one of the universities, or in any way interested in supporting DebConf 2011 in Munich, please join the DebConf11 Germany mailing list. Also check our meetings scheduled on the DebianMuc Wiki page, currently every Monday, 18:00, at the new LiMux offices in Sonnenstr.

P.S. There will also be a Bug Squashing Party in Munich end of November: Munich BSP November 2009

The october layout change for example isn't too big in fact. As far as I can tell it's not much more than turning the "hot" items that were in the right sidebar into a special tab (and breaking the refresh for the live feed, but I guess they'll fix that soon). The "live" tab is basically all information (see below for getting rid of certain restrictions); the "News" tab tries to reduce this amount of information by only showing you certain posts Facebook magic considers to be "important". If you are a heavy user you will probably prefer the "Live" feed, if you are a casual Facebook user, go with the "News" feed to have less crap posts to read.

Still there are some things you should be aware of when you are a facebook user (not all of these are new):

• Privacy settings in Facebook. You really should have a look at them. Be aware that if you put them all to the maximum, it may be next to impossible for not-yet-friends to add you. So I recommend to leave the "search privacy setting" to "all" but only have it show your photo and the option to add you as friend. This is enough for people to be able to actually add you as friend. (Setting it to "Friend of a Friend" was not sufficient for me.)
• Friend lists. These are really useful when you use Facebook for promoting events. Make lists of regional and topic grouping, for example I have a "swing dancers in Munich" list. If you bother people with irrelevant invitations they'll just ignore or unfriend you!
• Friend lists can also be used as a privacy device. I have a friend list called "privacy" with only those I consider to be really close. Photo markers etc. are restricted to these friends.
• The "live" feed will not show all your friends. By default it is restricted to 250 friends. The "Options" button at the end of the live feed page will allow you to increase this setting to 5000 and allow you to check which friends activities facebook had been hiding from you. (And indeed some people were I wondered if they had stopped using facebook had just been hidden from the feed by facebook ...)
• I wrote a Greasemonkey script (Greasemonkey is a Firefox extension, but by now also available for many other browsers, apparently even IE) to move the filters from the left to the right column, expanding the main feed column. Much more senisble layout IMHO.
• Avoid using applications where possible, especially all those Quiz applications and games. Remember: you will give the application access to most of your data and most of your friends' data as well. So make sure you can trust that application! (See this Post by the American Civil Liberties Union for details)

Also you should never forget that all the data you put online is hard to get rid of again. Just don't put anything there you don't want everyone to know. Facebook can be really powerful when used right for example as promotion channel. But the way you should be using it is to first consider what you want people to have an impression of you, then try to present yourself this way. Don't just throw everything that comes to your mind there. (This even more applies to blogs and web sites, obviously, that don't have any privacy control)

First of all, Patrick F. Riley - I worked with him on some projects when I was visiting the UC Berkeley, one of which was a predecessor to his latest thing: LiveDash. It's really cool: it allows you to search almost in realtime in TV feeds. It also live-indexes Twitter, blogs, news sources etc.

Secondly, HoneyWish (currently only available in German) is a service for a "honeymoon travel gift list" thing. It works like the traditional gift lists, except that instead of putting all kind of household stuff on it, there are all the parts of the honeymoon trip on the gift list. This makes much more sense these days: people tend to get married later; they might even be sharing a house for some time before getting married. So they don't need much silverware anymore, but they for sure will enjoy their honeymoon trip - so what could be a better gift for them?

Third, Amiando a web-based ticketing and event management service. Founded already some years ago by some friends, it has been growing and coming along nicely. Every now and then, it won some award, many of them in the "top startup" category.

There are of course many more projects of friends I'd like to point out, but these three definitely are highlights.

This includes you, YouTube. In HD view, the user popup menu only has the top 3.5 entries out of 5 accessible for me.

The following XSLT stylesheet can be used to find such embeds in a bunch of XHTML files using the command line xsltproc findNoWmode.xslt $( find -iname '*.html' )

<?xml version="1.0"?>
<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
 xmlns:html="http://www.w3.org/1999/xhtml">
<xsl:output omit-xml-declaration="yes" indent="no"/>
<xsl:template match="/">
  <xsl:call-template name="t"/>
</xsl:template>
<xsl:template name="t">
  <xsl:copy-of select="//html:embed[not(@wmode) and (count(param[@name='wmode']) = 0)]"/>
</xsl:template>
</xsl:stylesheet>

You can of course also write a XSLT stylesheet to insert the wmode statements whenever there is none, to make transparent your default.

[Update: I've received comments that this comes at qutie a performance cost for Flash, and that this might be the reason why YouTube doesn't use it - in particular for the HD videos. Also it isn't supported by WebKit based Browsers so far (so Safari neither?) and nor does it seem to be working in Gnash, an opensource flash plugin. So you have to choose between multiple evils if you are using Flash...]

I'm aware that the red-yellow border on the left doesn't line up right. I'm waiting for fixed graphics from the designer for that. There is also a glitch with clicking the logo when scrolled just a little bit down. These are on my to-do. At some point I also want to increase the use of CSS spriting to further reduce page load times. Oh, and Internet Explorer sucks, btw.

The web site is about Swing dancing in Munich (so no tech today), and at this time only in German. At a later stage, we might add English, too.

During August we'll also be building our own studio, "Cats Corner", which will actually be somewhat similarly decorated. :-) Congratulations to Christine for doing all that for the Lindy Hop scene!

P.S. Bring Down IE6.com, IE6 No more.com

P.P.S. See this blog post on how it is impossible to use the CSS "clip" property in a way that both IE7 and IE8 will understand. While only one is W3C standard, Firefox just accepts both ... but at least IE8 goes with the official standard now.

Hans-Peter Kriegel, Peer Kröger, Erich Schubert and Arthur Zimek
LoOP: Local Outlier Probabilities

It's an outlier detection method based on LOF (Local Outlier Factor) but a bit more statistically robust and with an easier to interpret score. Given the statistical backing, it works reasonably well on samples such as data pages of an appropriate index structure, reducing complexity to linear for the approximative version.

This publication is a bit special to me: I suggested the approach to my colleagues and they gave me the abstract and title for my birthday. :-)

My demo was yesterday, titled:

Elke Achtert, Thomas Bernecker, Hans-Peter Kriegel, Erich Schubert, Arthur Zimek:
ELKI in Time: ELKI 0.2 for the Performance Evaluation of Distance Measures for Time Series.

While this release visible only adds a small piece - some distance functions for time series and some related visualization code - it still marks a major milestone in ELKI development. Large parts of ELKI were reorganized and rewritten (such as all the output handling code) and lots of stuff added, including a lot of visualization related code that is not yet completely used in this release.

For this, I'm trying out the Amazon "aStore" functionality. Basically you setup a few categories and add Amazon products to these categories for people to choose from. Of course Amazon will also show other products it considers relevant etc.

My Swing music on Amazon "store" (on Amazon.de, but I guess it will also somehow take you to other Amazon sites?)

The (editor) UI is not very convincing yet. For example, it lacks an obvious way of moving "products" from one category to another, and you can't see more than 9 entries on a page in the editor, reordering is via entering sequence numbers etc. - that definitely could use some improvements.

Anyway, some people might find this useful.

I wouldn't call it so. It's just a very minimalistic search frontend to a nice database with lots of numerical facts.

Yes, it can give you the height of the eiffel tower (because that's a fact in its databases). It can even compute for you what Pi times the height of the eiffel tower is. But that is about as far as you can go in combining. In my tests, I wasn't able to compare the temperature in Munich with the temperature in Berlin (both of which WA will visualize you with a pretty graph, so these are facts in WA) - their query parser just doesn't get my question.

The funniest reply so far however was to the question:

How many cars in Germany?

No

Seriously, I doubt that there are no cars in Germany.

At least it also offers an explanation why it comes to this conclusion:

Cars is a town in south-western France (which as you might guess currently is not a part of Germany. :-) ) - so for WA, there are at least cars somewhere in Europe, but not in Germany!

It looks like a classic to me: in order to improve rendering performance, you approximate arcs with straight lines at small resolutions (if it's just 2 pixels big, nobody will be able to tell the difference). Except of course, when you end up doing the same approximation at a large zoom value - of course a 100-pixel circle looks different from a 100-pixel diamond.

Reported in 2005, still not fixed in current Java (we're in 2009 now).

Sun is really slow at fixing Java bugs.

See also a related Apache Batik bug report. Fortunately, this only applies to Java rendered graphics - SVG export, PDF, Postscript are all fine.

The site already has alternate stylesheets for browsers such as the broken Internet Explorers, so if I could convince GoLive to use their stylesheet it might be looking a lot better in the editor, too ...

I am aware that GoLive CS2 has been abandoned in favor of DreamWeaver. Still it's going to be used in a project I help with the web templates.

(Other options would be Kompozer and Amaya, but none of them seem really fit for production use: Amaya was just removed from Debian because it had some security issues and the maintainers had the impression the code was such a mess that there will be much more such issues. And Kompozer seemed to be a mostly dead branch of a Gecko hack (although there has been a new alpha release this year) ... is there some reliable opensource non-source HTML editor that I'm missing?)

P.S. Sorry, no comments in this blog. Use Email: erich AT debian ORG

Arrays in Java are special: they can efficiently store primitive data types. The expected difference in efficiency between byte[] and Byte[] is pretty big (of course some good VM might optimize) for obvious reasons (think of: references, garbage collection, pointer sizes, ...).

This is probably why you need to know the type before creating an array (because an array of primitive types such as byte will be different from one that stores objects of some kind).

In particular, the following Java code

  String[] foo = (String[]) new Object[0];

public static <T> T[] test() {
  T[] te = (T[]) new Object[0];
  System.err.println(te.length);
  return te;
}
String[] foobar = test();

What happens here is that in the test() method, T actually is replaced with "Object" at compile time. Thus the array type works just fine, and so does the call to te.length. Upon returning, it is then cast into a String[] array and fails.

Now here comes a crazy Java hack:

public static <T> T[] test(T... ts) {
  T[] te = (T[]) java.lang.reflect.Array.
      newInstance(ts.getClass().getComponentType(), 0);
  System.err.println(te.length);
  return te;
}
String[] foobar = test();

The exception is gone, foobar is of the proper type now!

A result of discovering this hack are these two methods:

public static <T> T[] newArrayOfNull(int len, T... ts) {
  // Varargs hack!
  return (T[]) java.lang.reflect.Array.
      newInstance(ts.getClass().getComponentType(), len);
}
public static <T> T[] toArray(Collection<T> coll, T... ts) {
  // Varargs hack!
  return coll.toArray(ts);
}

Notice how elegant the last method looks - and it finally allows you to do toArray(collection) instead of collection.toArray(new WhateverClassTheCollectionHas[0]).

Note that this is still a hack, and may or may not work with all Java compilers, JREs and/or Java versions.

Update: Note that this 'hack' is also not transitive. The context calling toArray needs to know the object type at compile time. So it doesn't save you much more than writing "new KnownClass[0]" etc.

Update: So I'm actually not using this - it's just a hack, and often quite hackish. The problem is that when you call e.g. toArray in an Generics context, it will actually create an array of "Object", so it makes much more sense to verbosely specify the class you want to use for the arrays (and get some reliability in use back).

It seems to be an interesting web storage service, with 2 GB of free storage.

However, the Linux client seems to be closed source (which is understandable, it seems to have a lot of neat features) - so I intend to use the web interface only (at least for now).

Update #2: There is a RFP bug for Debian, some Source is on the download site. And while this sort (except the images) is GPL, it's just the nautilus integration part, not the daemon you also need.

Did you try Dropbox? Does it work well? I know some people (especially Windows users) who could benefit a lot from a service like that, so I wonder if I should recommend them Dropbox. Or is there some better alternative (it should allow sharing of files though - synchronization is not as essential, it is a lot about exchanging files too large for usual email in small user groups; still synchronization probably is a comfortable way of transferring the files without having to think about it yourself)?

No comments in this blog - email me via erich AT debian ORG.

P.S. I know there is some referral program to get more storage, feel free to send me your referral link - I'll remove this PS once I've signed up.

P.P.S. There also is Ubuntu one, but as far as I can tell Ubuntu only so far. Looks very similar.

P.P.P.S. So far, I've received a lot of praise for DropBox.

P^4.S. My own referral link, feel free to use this to sign up (+256 MB for you, too!) and "upgrade" my account.

As of Internet Explorer 8, the required syntax of the clip attribute is identical to that specified in the Cascading Style Sheets (CSS), Level 2 Revision 1 (CSS2.1) specification; that is, commas are now required between the parameters of the rect() value.
...
In Internet Explorer 7 and earlier (and in Internet Explorer 8 or later in IE7 mode, EmulateIE7 mode, or IE5 mode), the commas should be omitted.

I see a few options:

• Giving both rules, hoping that the browsers will just ignore one they can't parse (and not revert to no clipping) - untested
• Use Conditional Comments to use a "ie7fixes.css" file (works)
• Do not use clipping (the most commonly used solution)
• Ignore the existance of Internet Exploder 7 and just follow the CSS 2.1 standard (that's the way it should be)

I do not find that very surprising that this has happened, most of the captchas around are very naive, and I've seen multiple scientific articles detailing how to attack various captchas. Many use colors and thin lines to make them look hard, but after applying a naive energy function and doing some blurring to remove the thin lines, they break down.

ReCaptcha is quite interesting, because it doesn't bother with some useless colorification that doesn't change contrast. But I wonder if it can't be overrun by spammers and how long it will scale. Still I figure it is what I would pick right now, because they can upgrade it if it actually is attacked by solvers.

It doesn't help much for the proxy attack on Captchas though (offer users to view some pr0n in exchange for solving a Captcha that you actually were given to solve by another site) - at least not when combined with some XSS and/or bot net. (The 'obvious' proxy approach can be IP-filtered.)

I was wondering if some people could contribute me some summarized web access data; no URLs or IP addresses.

The data I'd like to get can best be explained by the preprocessing step:

... | perl -ne '/\[(\d+\/\w+\/\d{4}:\d\d):\d\d/
&& print $1."\n";' | sort | uniq -c

That should produce 24 lines per day (one per hour), looking like this:

  count day-of-month/month/year:hour

It would be cool if you could send me some series for a couple of sites, if you happen to be in the position to provide this data. The data should cover at least a few weeks, the longer the better even up to a few years.

Too small sites are however not very useful but might be too noisy (so probably not the personal home page of your mom). If you are providing a larger number of series, you are of course free to include them.

I don't care much about what the site actually contains, I'd just ask you to give a tiny amount of meta information:

• Server timezone if not UTC
• Typical user timezone (if applicable, that mostly applies to 'regional' sites)
• Coarse classification of site (e.g. "product website", "web service", "search engine", "company site", "OSS software site" or something like that ...)
• Redistribution permission (if possible; most likely I'll only use data series where redistribution is allowed, since some conferences ask you kindly to provide such material)
• Your email address will be withheld, to increase anonymity of the data

Data use:

The main project idea is to evaluate different distance metrics in their capability of separting the different data sources, assuming that there is some difference in the shape in these curves. A different problem can be constructed by breaking the series into chunks covering approximately a day and then trying do separate different days, starting hours of the series (or offset server timezone vs. user timezone) and/or weekdays from weekends.

In our experiments, we've come to the conclusion that the experimental results are most interesting when there is a sufficient number of classes; so I'd like to get like 20 different interesting data series. At the same time, the series should be long enough, so I can break them into multiple chunks to have a reasonable number of 'sub-series' per class. If I have really long series, or e.g. series covering the same site but from multiple servers, I could even experiment with taking sample of different length from these sets.

(Say I have series covering 2 years, that is ~17k samples, from 3 servers, then I can take 51 disjoint sub-series of length 1000, or 102 of length 500, ...)

But it's obviously not possible for me to collect this data myself - I don't operate two dozen of such sites myself ...

An extra project I've been considering some time is some peak prediction for web accesses. Say you're running some fast growing site, wouldn't it be useful to have a prediction when the number of accesses will likely hit some magical limit (and e.g. overload your server) so you can increase your capacity on time? Of course it would be more sensible to apply this prediction e.g. onto CPU usage, e.g. predicting when your system might hit 90% load average over a 5 minute window in regular operation. Network bandwidth and disk IO also come into mind. You get the idea.

Please send them via email to erich.schubert AT gmail com

Thank you.

[P.S. Already recieved the first series, thank you! I can take care of sorting myself, no need to worry about that. And yes, I'm aware that the series will probably all be quite similar - common computer usage patterns such as work hours - but that is common in real world data and part of the challenge. Separting apples from dinosaurs is not a challenge.]

Generating static HTML pages from that will be easily possible using an XSTL transformation (for example), and I got the feedback that I could just use Google AppEngine for blog comments, so my actual blog could remain static-only (and thus much more secure and reliable). Any attack/spammer/bot can then only kill the comment functionality, not my own site.

Which brings me to another design consideration: the editing widget. Either for the blog comment application, for writing my own blog entries (via a https protected script or whatever) or maybe for a small CMS I've been thinking about - having a reliable HTML in-browser editing widget is something I could use every now and then (well, I'm not doing much Web stuff anymore these days).

Geniisoft has a good overview over in-browser (aka: Through The Web, TTW) editors. The top candidates seem to be:

• FCKeditor
• TinyMCE
• Xinha

I've heard before of FCKeditor and TinyMCE; I think I've been on the Xinha page before, too. However, comments on them have not always been good.

To some extend they all seem to have (to some extend) feature creep which usually is a bad sign - most often this means that there are security issues in one or another module or plugin.

TinyMCE for example has been described as "a bit of a pain" and "a tad clumsy" on the GSoC mailing list. I have had less fights with it than with the Debian Wikis (MoinMoin) markup language though.

I'm not looking for anything as big as these - I just need an editor that allows for some basic formatting (bold etc., links) and that produces reasonable XHTML output. I'll be feeding the output through some custom cleanup script anyway, which will kill disallowed code. So I don't want any editor which allows the user to create code that will then be killed afterwards.

Any personal experiences with any of these, or an important alternative that I might have missed (no PHP involved, please!) - email me (no comments on blog) at erich AT debian org.

Update: I've received a couple of pointers. Please don't send me links to projects that are not actively maintained anymore - I don't want to care about having to fix bugs in the editor widget myself.

One link I've received twice is actually quite impressive: WYMeditor. It doesn't try to look like a word processor, but actually is more of a semantic editor. Much more what I'm looking for than any of the others. I've also received a link to the Yahoo! UI Library Editor, which is quite clutter-free, but in the default setup at least very text-formatting oriented, not very semantic (that doesn't mean you couldn't change it that way, I guess you can). I was also pointed to Dojo, but that framework is totally feature creep (which also explains why it loads so slowly I guess), and the last time I looked at it's source code, I had some WTF moments - code quality at least outside of core doesn't seem to be very high (Yes, that code implements the "mod 7" operation using list shifts instead of a simple arithmetic operation).

Looks like I'll give the first try to WYMeditor. Update #2: the code seems to be rather ... complex. I'm looking for something neat and clean; it doesn't need to bring along yet another XHTML schema and validator ... Maybe I should try one of the others first?

So I have this interface in Java called Pair<FIRST,SECOND> along with two implementations SimplePair<FIRST,SECOND> and ComparablePair<FIRST extends Comparable<FIRST>,SECOND extends Comparable<SECOND>>.

For performance reasons, SimplePair is declared 'final', and so is ComparablePair. It's written everywhere that making classes final can make a large difference in Java, and since these objects will be used in a lot of places, it seems reasonable to care about this here.

However, it would often be nice to have better readable code, that is assuming I'm using SimplePair<Monkey,Banana>, it would then be nice to make a derived class BananaPreference extends SimplePair<Monkey,Banana>, with added methods getMonkey() and getPreferredbanana() to make the resulting code more readable.

Having readable code is also often quite as important as having performant code, after all ...

If someone with solid experience in Java optimization has some ideas to share, please do so! Email: erich AT debian DOT org - no comments in blog.

Right now, I have one idea on how it could be possible to achieve both (seriously, I could use some feedback from Java Gurus on that): make SimplePair and ComparablePair abstract, all methods there final, then derive final classes as needed. Does that combine the benefits?

[Update: I received from Joachim Sauer the following helpful link: JavaOne presentation on performance tuning and various VMs. Basically this seems to indicate that in all these common situations, any modern Java VM should be able to figure out the inlining options automatically and optimize appropriately, so it won't benefit from any "final" hint by the developer. Note that a C++ compile doesn't do runtime optimization, but allows compile time optimization at a much lower level, so this rule doesn't apply to C++.]

So far, we have quite few applications. Deadline is April 3rd, 19:00 UTC. Usually applications arrive rather late, but still I have the impression that we have much less than the previous years. But less copy & paste, too.

If you are interested in doing a GSoC project at Debian:

• Check the Debian Wiki which has all kind of relevant information.
• Talk to Debian people
• Make sure it's related to Debian (and not just "runs on Linux")
• Talk to Debian people
• Make sure your application shows your genuine interest and has some original ideas, copy & paste will not be sufficient
• Talk to Debian people

P.S. as far as I can tell, current Debian Developers can be eligible as well, although it has also always been a goal of the project to get new contributors involved.

Anyway, what I'm more interested in is the technical background. The last months, I saw lot's of instability in Facebook, and it often gave the impression of being next to breaking down because of load. With the redesign, it feels more stable to me. Given that the 'mobile' page still seems to be pretty much the same, they still seem to be aggregating the same information, and the difference is just in the front page.

Guess their servers just weren't able to handle all the 'live feed' access for all their users. Maybe they'll get some database experts to redesign that feature and bring it back; after all it seems one of the things people most miss in the new layout.

P.S. Add this to your user stylesheet to hide some ads:

@-moz-document domain(facebook.com) {
.UIHotStory_Ad { display: none; }
#profile_sidebar_ads, #sidebar_ads { display: none; }
}

After deleting a leftover playlist, suddenly I noticed with another playlist that had worked the day before that it was now empty. Ouch. So I investigated a bit more...

At some point, I ran across a post on a similar Sansa device titled

Firmware Bug - Playlists numbers 43, 46, and 49 are always empty

This is also what I'm seeing on my device. Certain 'slots' in the playlist list don't work.

No, I'm not kidding.

The way I 'fixed' my non-working playlist was this: I duplicated the file. Now the duplicate is on the broken slot, and the old copy is working again. Unfortunately, this also has shifted the later playlists into other slots, so I'll have to repeat this for the other affected playlists. I guess I'll just write a script to keep duplicate copies of all my playlists, since apparently there are no two consequent slots affected.

Broken slots on my Sansa Clip v2: 22, 25, 28, 34, 37, 40 (I have 59 lists currently on the player, so there might be more broken 'blocks' somewhere)

[Update: Sansa Firmware 02.01.32 says in the 'bugs fixed': Playlists redirect to GoList when there are many playlists on device, so this might be fixed now]

On my Core 1 Duo system (32 bit), official Adobe Flash crashes my browser when I close a tab which had a Flash plugin running. Going to a blank page then closing the tab usually helps, but it seems that sometimes Flash continues to run in the background (sound doesn't stop either) and then will still crash.

On my AMD64 system, the official Adobe plugin crashes my browser. There are reports at Adobe that link it to Gmail. So here, the Adobe flash is unusable. The 32 bit version via nspluginwrapper did not have sound for me, probably some issue with Pulseaudio.

So I'm now trying out Gnash. First thing I noticed: it has FlashBlock built in, all those stupid Flash things won't auto-run but I'll always have the nice play button to enable them when I want them to run. And while Gnash is working pretty well on most sites, every now and then something just does not work. Like some YouTube movies not playing (usually very short ones - maybe it will only start playing a video when the buffer was completely filled, and a video which is smaller than the buffer will thus not play?) etc.

Some wishlist items:

• Add a 'auto run whitelist' to Gnash, when I go to YouTube I usually want to actually run the video.
• Provide some 'fall through' option, so if the Flash doesn't work right in Gnash I can pass it on to the Adobe plugin if I really need to.