Some SPAM filters use OCR to detect text in pictures like this one. This type of SPAM detection is very costly: all these nifty text recognition algorithms are running on your resources and are wasting your (CPU-)time. Additionally, due to spammer's creativity boost during the last two years even anti-spam techniques like fingerprinting with hash checksums are rendered useless when there's no image like the other one. Spammer's software just modifies the original picture, does some blur, some smear, puts abstract geometric patterns into the image, thus generating a unique piece of art, and will outsmart the fingerprint.
I'm getting these images every day. When they arrived for the first time I thought it was a joke, then I accepted the creative idea to fool my filters, then I got annoyed, as always, because it's just SPAM.
That message kept stuck in my filter, anyway, because it didn't like the gateway (just some contaminated Windows box, connected to the Internet via DSL or cable), and the mail's message ID was strange because it used a domain that doesn't exist.

When I tried to get some information about Xythos in conjunction with MS SQL Server 2005 (we are going to upgrade) my favourite search engine honestly presented its results. Plus the question:

Er, no :-) But let's see the myth behind SQL Server and click that link: no, just lame stuff, no myths here. My world view keeps its order.

George Ou writes about a possible exploit of MS Vista's speech recognition system, describing a scenario where spoken words played by an application (or web site) were able to engage the OS's voice recognition system. So if your computer reacts on spoken words - beware. Don't use these:
"Shutdown."
"Encrypt / Wipe / Format disk."
"Send mail to all known users."
"Quit, don't save."
"Activate Windows."
"Change Password to #%+&$#-§$)."
"Switch on web cam. Publish."
"Make share c$, allow for everyone."
"Run http://badwebsite.com/badsoftware.exe".
"Say: I must not let my Windows speak!"

Today's lesson: if you ever want to install a VPN on a Windows box make sure that you have a running TAPI service. Otherwise, you'll see disabled radio buttons on the various wizard dialogues. Took me quite some time to fix this.

StudiVZ, Germany's biggest dating portal for students student Web2.0 community site was obliged to do a password reset for all accounts, because hackers were able to read various account data. Offline for a couple of hours (To update their system? Watch the degree of this catastrophic event?), they're now online again. Since the site makers improved their security measurements I just hope that their improvements flow directly into the development of SchülerVZ.

Is it me or are these colourful web sites with their round edges spreading everywhere? Those sites with cute names ending with double-Os or at least with double-Os inmidst the name. Here's a new one: www.atoolo.de where you get a virtual desktop for free and some applications running on that quasi-OS. Nice idea, but I'd never store my letters or tax declarations on servers I don't own (*own*, not "0wn" :-)

That's the question asked by Australia's magazine The Age. Australia's and other English-spoken countries' universities experience a decreasing number of IT students. What's the reason for this phenomenon, especially if one considers our daily life that becomes more and more digital, filled with communication gadgets and computer power nobody would have imagined a few decades ago?
There's a multitude of reasons. First, computer science is subject to pig cycles like any other discipline. This wasn't the case 20 years ago when I started studying, where the future was bright and a lot of lies were told ("big money", "move the world") . Additionally, nobody in the job market is interested in computer science but IT jobs. Science doesn't make rich in money, so most people are interested in practical knowledge to get their job done.
A noticable cooling in Germany's job market happened during the mid-90s, after that the Internet was about to come and I got my first 'real' job. Before that almost every major company was off-shoring its developments and at a job fair I've been told that they hadn't any jobs because their developing staff was now residing in India.
Since then, ups and downs in that IT job market are a commonplace and job training changed significantly. IT's no longer rocket science but the supplier of everyday's technology and a lot of yesterday's old glamour is gone forever.
Society has changed, too. Though our administration launches success stories about our recovering and growing economy, the job market itself has changed drastically. Staying with the same company for the greatest part of your life becomes unlikely and once you're losing your job chances are high that you'll experience a real crash of your life style, especially if you're depending on unemployment benefits. Getting poor in Germany has become a real threat for employees, and the strength of one's company doesn't mean that her job is secure. There is no sign of resistance in the working class, so people seem to have accepted this situation.
And then there's the prestige. Especially in Germany, technical jobs never were that reputable. As long as they stayed with the academic community and were a matter of just a few people, computer and information science were considered as something sophisticated and boring. With every step of vulgarisation the reputation of IT was decreasing and to do IT wasn't special anymore. To become a lawyer or a consultant has always been more prestigious, these jobs are usually well-paid and promise high social status - as opposed to IT jobs, where employees often are service-oriented computer nerds and nothing else.
It's been a great loss of reputation that computer science experienced in the last twenty years. IT early left academia and became a business service. For the elite of a society becoming a servant isn't an option, so IT became the playground of social climbers. In a society that is affected by strong mechanisms of social distance service jobs don't promise enough reputation, thus less and less people want to do IT.

No, it's not the Doc having fun while waiting for spring but the European Commission found that the IT sector had a shortfall of 300,000 IT professionals across Europe by 2010. Dropping by 3% from 1998, there were only 22% female graduates in 2006 (USA: 28%). Possible reasons would be the "geek" stigma of IT jobs and women who left work for having children. Very interesting is the discovery of women preferring management courses over technical courses (that are mostly taken by men). I think they got it. IT is getting uncool. (Source)

John Backus, co-inventor of Fortran and the famous Backus-Naur Form died last saturday. Honoured by IEEE and ACM, he was one of my early computer science heros. Σ*, yeah!

The harder you declare a technology dead, the more obvious is its enduring existence. So no anger here, all these things are still needed. But it's obvious, that there are technologies and practices that are less important that they were a few years ago. Here's Computerworld's list. It includes: Cobol, nonrelational database management systems, non-IP networks, cc:Mail, ColdFusion, C programming,
PowerBuilder, certified NetWare engineers, PC network administrators, and OS/2.
Got any of these? Some more suggestions?

If you ever wondered if there are black holes in the Internet - you're right. Ethan Katz-Bassett, a computer science Ph.D. candidate from the University of Washington used his tool Hubble to detect that 10% of all sites on the Internet are unreachable, day by day. Problems reach from total unavailability up to various routing problems. 75% of these problems get fixed within an hour, some last for several days. If only the right sites would die everyday... (Source)

Ever failed reading these nasty distorted numbers and symbols while creating a new account or entering a comment on your favourite blog? Captchas are a quite efficient way to separate spamming bots from real users because the last one usually would be able to read the letters while the first one should be not. Unfortunately, times changed. An increasing number of bots acquired the ability to examine even twisted letters in a captcha image and continued spamming. As a consequence of this, captchas got more and more unreadable, even to the common human reader, thus rendering captcha protection impossible. New captchas will present pictures instead of text, and in the near future you'll be presented pictures with cats and dogs where you have to choose the right images. This kind of task is extremely difficult for software (i.e. bots) and so the next turn in this neverending fight between them and us is being heralded. Hope they provide a Perl module soon! (Source)

Paul Murphy's readable entry about the security context in the computer language discussion and why one language is not inherently more insecure than another. Except PHP, of course :-)

Hope they don't brag of their technology too much: ACM's current SIGGRAPH shows holographic, tactile, and virtual wizardry for the displays of tomorrow.

Upgrading to Movable Type V4. Took less than an hour, some little problems based on my server configuration. Older entries may look a bit garbled, but I plan to adapt the blog to the website's design anyway. Nice job, six apart!
Note: switch off mod_security before installing, get password for DB user close at hand, copy whole mt-static directory into target directory, refresh templates and republish entries.

Certainly you know the famous TV show CSI: Crime Scene Investigation. Here's a bunch of special reports to probe further.

IRTF is rethinking the Internet' routing architecture to be prepared for the billions of users to come and to remain routing tables maintainable. On debate is how backbone routers operate. Recently, they're exchanging routing information via BGP, the Border Gateway Protocol. BGP routing tables are growing fast what will result in growing costs for everyone in the end. To strengthen its efforts, IRTF has revived the Routing Research Group. One of its interesting problems deals with IP addresses.

Toni Li, a well-know designer of routers at various network manufacturers, says:

The IP address has both the identification of the node and the location of the node. The question becomes: Can we separate the identification from the locator semantics, and can we still run an Internet with that kind of architecture?

• The Locator/ID Separation Protocol (LISP) describes a method for splitting Internet addresses into endpoint identifiers and routing locators through the use of tunnel routers.


• The Six/One proposal proposes provider-dependent IP addresses given by service providers, while hosts would use addressing spaces from all providers on an interchangeable basis.

University of Wisconsin, Milwaukee assistant professor of information studies Thomas Haigh gives lectures in computer history and shows us that yesterday's problems are not too far away from ours today.

Another interesting repository is IEEE's History Center.

If you're alerted by your system about rejected mail hosts trying to throw spam at you, are experienced a little bit in Perl and if you have access to a database, this article might be interesting for you.

Recent developments in Germany make plain that protecting privacy is important and necessary. If you want to drop us a line, please use PGP: here's the public key. Get a key for yourself, learn how to protect your privacy! (Image reproduced with permission; for more, please see www.clap-club.de)

When I was studying computer science I tried to steer clear of functional languages like LISP, Haskell, APL and the like. I found them way too abstract to get the things done I had in mind. Instead, I concentrated on imperative languages like Delphi, C (in its various variants) and Perl.
Perhaps it's time to think this over. Microsoft research announced F# (pronounced: F sharp), especially for projects dealing with concurrency. Goals are high, as functional programming languages are meant to solve the "biggest challenges in the industry" (as usual, I think this is part of the hype to say that). Even if they are "in their own world", F# will run on the .net Common Language Runtime.

Reminds me of a small Perl.Net project I did several years ago: a small Perl assembly using LWP to crawl a website that was controlled by a GUI written in C#. Worked astonishingly well. Unfortunately, Perl.Net has been cancelled. Maybe ActiveState should have had it named 'P#'. (Source)

"Study shows Google favored over other search engines by webmasters" - you don't say! By crwaling more than 7,500 Web sites and examining their robots.txt files (if present), a Penn State study found out that 40 % of these web sites had a robots.txt (10 years before, it was a mere 10%) and that most of these mirrored search engines' market share. The study doesn't tell why. But IMO this is obvious: which principal ever heared of search engines other than Google? (Source)

Just want to contribute to this complex riddle Microsoft delievered with .Net 2.0. I got HTTP 403 status codes when my client tried to access the WebResource.axd handler. Fortunately, I was able to move the application (together with its web.config) into a different directory than the siteroot. That's it (at least in this case). Pathetic.

Yes, it's Fortran! After discovering F# last week, today I'm reading that Fortran is still alive and vital. It's not all Python and Ruby, folks. Learning on of the old languages still allows you to keep pace on recent trends and to be able to draw on lots of working code from the last decades. (Source)

A combinatorial approach to better detect bugs in software - please surprise me! If it works well, I'll definitely give it a try. Otherwise, I'll stick to lint.

Congratulations! The one and only language that has a "bless" operator had its first release 20 years ago. It taught me Regular Expressions and a relaxed way to address programming problems. I still enjoy its freedom and love its flexibility. This web is done with Mason, a Perl framework mostly used for web applications, and so is Lollop, our home-grown CMS. Perl-based is the blog software I'm using here and many other small tools I developed over the last ten years. While this in mind: happy birthday and kudos for a vital community!

In order to help people to make their virtual worlds or games more vivid, Stanford computer scientist Vladlen Koltun provides a neat software that creates masses of trees. Trees are well-measured objects and difficult to construct manually, so this tool is a real help for producing complex 3D data. Provided forms are ranging from naturalistic to fancyful. (Dryad homepage, Source)

(If you think that this entry is about life coaching: no, it's not. Just (very) technical stuff here.)

Quirky technical problems are my daily bread. This one was a toughie and I hope that its solution will help others with similar symptoms. I will first sketch the basic infrastructure, then the problem, our debugging steps and the solution(s).

The affected system consists of two Windows-based W2K3 (Windows 2003) Servers with SP2 installed, running in a NLB cluster with two nodes. Each cluster member hosts an IBM WebSphere 6.0 environment which is controlled by a third Network Deployment server. The Websphere Cell consists of a configured cluster of two nodes (our both servers).

Each server (we're just talking about the cluster here, forget that ND machine) has a Xeon processor at 3 GHz and is packed with 10 GBytes of RAM. The network interfaces (4 in total; each server has a frontend and a backend NIC) are HP NC373F PCIe Multifunc Gig Server Adapters with newest drivers. The load balancing happens at the frontend network interfaces, where the load of the web servers, whose IP addresses are bound to these interfaces, are balanced. The backend interfaces are unclustered and unbalanced, usually they're used for maintenance and administration, here they are also used to send the web server's answers back to the client.

This cluster is connected to a host machine (an IBM system) via Connect Direct (this realizes the connection to a DB2 database on the host) and MQ Series. Our deployed web application (running under WAS 6 and thus a Java app) gets its request from a client application (another Java app running at the user's deskop) and sends messages and also gets / sends data from and to the host.

The client application could be used for approximately one minute (or longer). It crashed with a "socket write error" exception and claimed that the TCP-connection to the web server was lost. In the web server's log we detected a "Socket Timeout Error". Obviously the TCP connection was broken! This behaviour was completely new to us, because previous tests in the test environment were running without problems. The architecture of the test system was identical to that of the productive environment. The only difference was the hardware: the productive servers are HP G5 machines, while the test servers are G3 servers.

After checking the whole application environment and every WebSphere setting available, we finally found the error in the NIC driver configuration. Obviously our very special software setting (Windows 2003 Server SP2, WAS 6) didn't harmonize with the NIC driver settings on this special hardware. It was the "Receive-Side Scaling" option that we found activated. After switching it off the problem disappeared and the TCP connection between client application and web server wasn't interrupted any longer. Since not every NIC does allow this setting to be set manually, there's a registry key where you can configure it on the affected server (after SP2 is installed): start regedit and look for the key HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters. Now set up a new DWORD value EnableRSS and set it to 0 (zero). If problems persist, you may set up another DWORD value DisableTaskOffload and set it to 1 (one). More detailed instructions.

This error was extremely difficult to detect, because all other network operations worked flawlessly. It was possible to connect and administrate via RDP, to copy huge amounts of data, and all WebSphere handling including deployment and node synchronisation via ND server worked like a charm. The problems occured only while running the application. We suspect some very strange side effects somewhere between driver settings and Java network operations.

Of course, problems didn't stop here. We learned, that the queue wasn't served correctly, because an older version of our software was still running on another (older) server. After stopping this old installation, everything went fine, finally!

If you have read one article here about social networks or another, you have learned that that the MadScientist is not necessarily a friend of social networks because I'm just too paranoid to see that many advantages in leaving my personal data and preferences on any servers just to make the providers happy and to simplify their advertizement strategies.

But - and this is the technical part beginning here - you have to know your enemy and the technology he is using. It's not bad to know about weaponry even if you're a pacifist, so let's have a look about recent techniques and trends to come.

What's it about? It is understood that users of social network sites (take anyone you know) want to migrate their data to other network sites for various reasons: because they want to leave one portal and enter another, or they might want to merge their data because MySpace, Flickr, and Facebook just isn't enough and the user is too lazy to type in once more her user data. So what does the industry need, as always when it comes to exchanging data? A standard, right!

According to this article (paid subscription req'd) by Karen Heyman, several techniques are already available that might be part of a future standard for exchanging user profile data between various providers: RSS, the Really Simple Syndication for exchanging syndicated content; OpenID, an identity system supported by some big players (Google, Yahoo, Microsoft), in order to enable users to register accounts without having to re-enter account data; OAuth, a protocol for secure API authentication from various sources; Microformats integrate meta information in HTML containers, thus adding semantic data to it; RDF (Resource Description Framework), another method of modeling information that adds metadata to content; APML, "an XML-based format for capturing a person's interests and dislikes" (Wikipedia); SIOC (Semantically-Interlinked Online Communities Project), a semantic technology for interconnecting web-based discussions (blogs, forums etc.); FOAF (Friend of a Friend), an RDF extension for "describing persons, their activities and their relations to other people and objects".

Several of these technologies are used by the DPWG who wants to give to the users control over their data. A lot of the logos you might have seen at the Wikipedia links above can also be found on DPWG's homepage. But, according to their web site, they don't prefer one technology and neglect the other but they promote and moderate in the desired standardization process. No wonder you find them on a lot of conferences and barcamps. Microsoft and Google are already members of DPWG.

Emerging standards, international enterprises with profit-oriented long-term goals, careless users and a vanishing awareness for privacy issues will set up the scene for the next following years. This is not only a 'digital' phenomenon: debit cards that allow customers to save a few bucks are also a perfect means to collect customer data and use their preferences for creating profiles. With powerful standards the exchange of all these data will be made simple, giving web users and customers a feeling of being well-known and, maybe, liked. Difficulties when trying to hide your identity will amplify until it will be completely uncommon to have a desire for (online) privacy, at least recent developments in politics suggest this. The data collector's club is still growing and there are several good reasons not to give your data away.

System.AppDomain.CurrentDomain.BaseDirectory is the current directory for a Windows service in .Net. Thank you!

Being an Apple Aperture user for months, the appearance of the main window suddenly got distorted: as if the window would be much larger than it really was, all tab controls and the contents of the browser/viewer area were displayed above the main window's top border. There were no scroll bars to get the thumnails or project folders back into the window's center, so the whole application was plain unusuable. Even my photos looked distorted under special circumstances, it was a whole mess. I've read about similar effects where the window extended to the right or left without giving the user a chance to make the content visible again.

After logging in as a different user and starting Aperture, there was no error. That proved that something in my local Aperture settings was wrong. Before this I re-installed the application, but to no avail. Seeing that it would run with different preferences was a crucial hint.

After fiddling with the various settings in com.apple.Aperture.plist and com.apple.Aperture.DodgeAndBurnPlugIn.plist in /Users/username/Library/Preferences/ with no effect I simply deleted them. Voilà: after the next program start the main window was working again. Problem solved.

Under certain cases an error message pops up when using Internet Explorer for downloading an Office document or a PDF file via HTTPS, telling that "Internet Explorer Cannot Download". This happens if the web server is issuing additional HTTP headers pragma:no-cache and cache-control:no-cache to prevent the client from caching the website contents locally. If you can exclude trivial problems like wrong access rights and permissions, things may get complicated soon. This issue is documented at Microsoft's Knowledge Base, but there's just a poor workaround that doesn't really help you, especially when you're the webmaster and you just have to use the HTTP headers. This problem occured with IE6 first, but after installing Windows XP SP3 it could be reproduced with Internet Explorer 7 as well.

Here's a trick that helped, provided by Microsoft support (though I cannot guarantee that it helps in every case): save the following lines:

Windows Registry Editor Version 5.00

"SSLPragmaNoCache"=dword:00000001
"BypassSSLNoCacheCheck"=dword:00000001
"BypassHTTPNoCacheCheck"=dword:00000001

into a .reg file and import this into your client's registry - or insert these key/value pairs manually at HKEY_CURRENT_USER\SOFTWARE\Microsoft\Windows\CurrentVersion\Internet Settings at every client. Since this is a client issue there's no way to handle this problem centrally.

Basically this fix is a mixture of the solutions given in article Q323308.

Windows Server 2003 SP1 (or higher): if your Scheduled Task is giving you back a result code of 0x80 (instead of 0x0 what would signal that everything's okay), then there's a plethora of possible causes. If you google a bit, you'll find many solutions ranging from exchanging a DLL up to notices that 0x80 wouldn't mean anything at all and that you don't have to care about.

Another possible reason for a 0x80 result code is lack of resources. When you encounter the 0x80 problem, read this fine discussion (I've seldom seen a problem description with so much detail like there) and have a look at the Task Manager of your server. You might be surprised like I was this morning. I erased all crashed tasks and presto! problem solved.

"Most data on enterprise networks rarely gets accessed after it is stored, largely because users are too busy writing new data to access old data." - every file server administrator knows that. Interesting!

Need to clean up directories that are clogged up with too many old files, and to do this automagically? If you're using a Windows box try my TidyUp script. Here's an example:

cscript tidyup.vbs -f c:\temp -d 20 -i "log|bak|tmp"

This will delete all .log, .bak, and .tmp files in c:\temp. If you want to delete all files in c:\temp, omit the -i parameter. If you aren't sure about the effects of your command, just append the -a parameter that does everything but delete the files. Just read the manual and copy the code.

They won't tell which Christmas Perl 6 will be released, but they know how to water Perl enthusiasts' mouths: Wall talked about that in his annual "State of the Onion" speech at the O'Reilly Open Source Conference (OSCON) (Source)

When you search for errors that might occur when logging into the Site Manager, a tool for maintaining the Microsoft Content Management Server (MCMS), you will get quite a decent list, but this time I have to offer a new one and, fortunately, its solution. After double-clicking the Site Manager icon the login dialog appeared, clicking the "OK" button resulted in an error message, stating a 500 server error. No further details.

Examining the web server log of the associated Site Manager website, revealed that the error occured when a file resstub.asp was called, obviously line 16 threw an error. Analyzing the HTTP traffic (you can easily do this with a tool like fiddler; just start it before your client application and get insight into the data flowing between the server and your client) confirmed this. Here I was able to read the error message, telling me that an object creation in the ASP code failed.

Line 16 in resstub.asp reads as follows:

Set pBinFiler = Server.CreateObject("AEBinFile.AEBinFile.1")

There was not much info I found about the object ("AEBinFile.AEBinFile.1"). I did a search on the local harddrive of the server where MCMS is installed for the actual DLL (AEBinFiler.dll) and I found it - twice. One location was the original place at \program files\microsoft content management server\server\iis_nr\system\marshalling\resstub.asp, the other place was the "bin" directory of another CMS website I deployed a few days ago. And that was obviously wrong! When I deployed that website the installer registered this other DLL in the new website's "bin" directory, thus causing the HTTP 500 error.

The problem was solved by issuing a

regsvr32 AEBinFiler.dll

in the original MCMS bin directory.

Microsoft's load balancing usually just works. But sometimes you experience these quirky little problems that maybe wouldn't give you a show stopper but that are simply annoying. Like this one: WLBS events (with event id 65, then 28 (or another id in the 20s range) in the "System" event log were telling me that one of my hosts was entering and leaving the cluster all the time, resulting in a lot of converging actions. NLB manager didn't know anything about configuration problems, so the reason for this had to be elsewhere.

And this hint by Russ Kaufmann (seems to be only available in the Google cache) pointed into the right direction: after comparing the various parameters of my NICs (such as TCP offload checksum) I found different settings on both servers. I adjusted the parameters on the machine that was permanently leaving and re-entering the cluster, and voilà, problem solved. Obviously, WLBS had problems with synchronizing the two nodes because of their different NIC configurations. So watch the details of your H/W configuration!

A very cool tool for simulating the effect of Tilt-Shift lenses: TiltShiftMaker. Recommended.

So Microsoft released its brand-new product 'Surface', a small desktop that runs a desktop, which means that a specially prepared Windows PC is used for providing a digital desktop experience. You might view videos, do funny games, have some user experience somewhere between iPhone and Windows etc.

I'm not that impressed. In principle it's the old desktop metaphor, only the desktop became a bit more colourful and it looks a bit more playful. It's a consumer machine that doesn't leave the well-known paths of common GUIs. I never thought I was working in such an innovative environment, when we developed a pioneer device in 1993: one of its features included a 3D display. It was meant to break the desktop metaphor.