I usually only use English Software, but sometimes have to create documents in German.

The English Office 2010 Pro has proofing tools for English, French and Spanish, but not for German.

The German Office supports German, English, French and Italian. But after installing the German version, I get the UI in German which I can't stand. You can buy a language Pack at office.com for 25 US dollars, but if you have a second license and both the English and German versions of Office do the following:

Install Office, I would use with the secondary language first, then start the install process for the Office in your primary language.

It recognizes that Offices is already installed in a different language. If you click "Repair" you get the option to add a second display language.

You can change the languages in options or in the separate “Microsoft Office 2010 Language Preferences” tool.

This only works if you have both language versions and you may need two full Office licenses which I have, but I am not sure about that.

My use case for this requirement was that users can add text on a web site and could include links (a tags). I want all these links
to open a new window when clicked on. So I needed to add a target="_blank" to all the a tags in the text.

My first idea was to use pure Regular Expressions, but I couldn't figure out how to use the negative lookahead feature, I needed
this because I can't simple add a target="_blank" to each a tag, what if it already has one?

So instead I am using a simple Regex to find the appropriate tags and then use a MatchEvaluator, which is a very cool
feature of the .Net regular expression class. It allows you to compute the replacement string in a method for each
match.

          public
          class HtmlHelper
{ string _attrib = string.Empty; publicstring AddAttribute(string source, string tagName, string attrib)
{ _attrib = attrib; string term = "<" +
tagName + " [^>]+>"; Regex r = new Regex(term,
RegexOptions.IgnoreCase); MatchEvaluator myEvaluator = new MatchEvaluator(this.ProcessMatch); return r.Replace(source,
myEvaluator); } privatestring ProcessMatch(Match
m) { string tag = m.Value; if (tag.IndexOf(_attrib)
== -1) { tag = tag.Replace(">", "
" + _attrib + ">"); } return tag;
} }

HtmlHelper helper = new HtmlHelper();
htmlText = helper.AddAttribute(htmlText, "a", "target=\"_blank\"");

Recently I saw some people complaining about the Gmail user interface: Jeff @ Coding Horror, there was also a reference to this in one of the Mix 10 talks but I can't remember which one, because I watched too many of them.

So all the super smart guys now work for Google, but even they don't always get it right.

There are many factors that make up good software and one of them is to follow the standards of the Operating system it runs on.

Here are two examples of why I don't like some Google Software.

There is a process named "GoogleToolbarNotifier.exe" on the computer I am sitting at. I guess this has something to do with the Google Toolbar in web browsers. I do not run a browser though. So lets open Task Manager to kill the process: "Access Denied". What, why can't I kill it. It runs under my own user account and I should have all the rights to stop it.

The only way to kill it is to use a really low level StopProcess API that is usually not available to the end-user.

The other example is the Chrome browser. So far I had only used the portable version of it, which is not installed on the computer. Yesterday I installed the normal version and in the setup wizard chose to not create entries in the Start menu entries or on the desktop. After the wizard was finished I tried to start the browser and had a hard time finding it. Nothing in "Program Files", it turned out it was installed under my user profile directory which is according to Microsoft for application data and settings but not for executables.

Also because as I usually installed Software as an administrator, my normal user account who is not an admin, did not have access to the Software. I had to run setup again under my normal user account, download the whole package again from Google and then ended up with a second copy of the application on my hard drive.

There is actually a standalone installation package for Chrome, but it is not mentioned on the Chrome pages and you have to google with Bing for it.

The fact that the setup wizard doesn't allow me to choose a destination for the software I install, pisses me off. What if I have my user profile on a remote network drive? This is now filled up with Google junk.

Don't get me wrong, Chrome seems to be a nice browser but until Google gets its act together I will not install it on my computers. At least I can use the portable version.

Some third party Windows Services think they are very important.
They do not give you the option to stop them in the Services MMC console.
You can kill the process using Process Explorer but you get an "Access Denied" message or the services are coming back right away after you killed them. That is because they have their Recovery options set to "Restart Service". When you try to changes this or even disable the service, you get an another "Access Denied".

One way around this is, to open Autoruns.exe, go to the Services tab and double-click the entry in question. This will open the registry editor with the key for the service selected. Delete the "FailureActions" value, this will reset the recovery options to the default which is 'do nothing', now you can kill the services and it wont come back.

If you want to disable it, set the Start value from 2 (Automatic) to 4 (Disabled).

However they are services that think they are super important, one of these is "Service for G-Buster Browser Defense". Access to its configuration in Windows tools is denied. When you change the start value in the registry to 'disabled', within seconds it is changed back to 'Automatic'.
This is done by the winlogon.exe process, one of the core Windows services. A dll of the G-Buster software 'gbiehAbn.dll' is loaded into the winlogon.exe process. (I think this happens because of the entry
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Winlogon\Notify\GbPluginAbn\DllName). The Software also comes with a kernel mode driver that starts even earlier in the startup process. However it does not come with an un-installer.

The Software comes from a Brazilian bank and does not seem to do any harm.

Deleting the FailureActions registry value and killing the process with RootkitUnhooker.exe works, but so far I have not figured out how to get rid of it permanently.

• FixRegistry.cmd fixes some things in the registry, which makes it easier to execute the other batch files.
• saj.cmd (stop-all-junk): stops unneeded services and kills many common unwanted applications
• sas.cmd (stop-all-services): stops additional services that are not really needed.
• sarj.cmd (stop-all-root-junk): Stops some processes that I can not stop as a normal admin, I need to run it with this special command:
  psexec.exe -s C:\bin\sarj.cmd
  

I use a whole bunch of the Sysinternals tools on my USB stick when checking computers for viruses and fixing other problems. On every new computer, each of the tools first pops up the Microsoft License Agreement box which you have to agree to. It is just one click but it gets annoying over time.

So I use a small batch file "FixReg.cmd" which among other things has the following lines:

reg.exe ADD HKCU\Software\Sysinternals\PsKill /v EulaAccepted /t REG_DWORD /d 1 /f
reg.exe ADD HKCU\Software\Sysinternals\PsList /v EulaAccepted /t REG_DWORD /d 1 /f
reg.exe ADD HKCU\Software\Sysinternals\PsInfo /v EulaAccepted /t REG_DWORD /d 1 /f
reg.exe ADD HKCU\Software\Sysinternals\PSexec /v EulaAccepted /t REG_DWORD /d 1 /f
reg.exe ADD "HKCU\Software\Sysinternals\Process Explorer" /v EulaAccepted /t REG_DWORD /d 1 /f
reg.exe ADD "HKCU\Software\Sysinternals\Process Monitor"  /v EulaAccepted /t REG_DWORD /d 1 /f
reg.exe ADD "HKCU\Software\Sysinternals\Autoruns"         /v EulaAccepted /t REG_DWORD /d 1 /f
reg.exe ADD "HKCU\Software\Sysinternals\TCPView"          /v EulaAccepted /t REG_DWORD /d 1 /f
reg.exe ADD "HKCU\Software\Sysinternals\RootkitRevealer"  /v EulaAccepted /t REG_DWORD /d 1 /f 

I run this on every new computer before I do anything else. After all I have already read and agreed to the license terms more than once. As you can see from the HKCU registry hive, each user has to agree separately.

When using the psexec.exe tool with the -s switch to run a command prompt under the almighty System account, you first have to agree to the License Agreement, which is fine. But then when you want to work with additional Sysinternals tools within the System console, you have a problem. When starting pskill.exe for example, the program tries to pop up the License Agreement dialog box. This does not work because the system account session does not have a desktop and this point and the program hangs because nobody can ever agree to the License Agreement.

To fix this, add a few more lines to the FixReg.cmd batch:

reg.exe ADD HKU\.DEFAULT\Software\Sysinternals\PSexec /v EulaAccepted /t REG_DWORD /d 1 /f
reg.exe ADD HKU\.DEFAULT\Software\Sysinternals\PsKill /v EulaAccepted /t REG_DWORD /d 1 /f
reg.exe ADD HKU\.DEFAULT\Software\Sysinternals\PsList /v EulaAccepted /t REG_DWORD /d 1 /f
reg.exe ADD HKU\.DEFAULT\Software\Sysinternals\PsInfo /v EulaAccepted /t REG_DWORD /d 1 /f

The System account looks in the HKEY_USERS\.DEFAULT section of the registry for any settings it should use.

Now you can kill processes that you couldn't kill as normal administrator:

psexec.exe -s cmd.exe 
pskill.exe badprogram.exe 
exit

First I looked into a problem I had since the last new version of my travelog website. When I add a new entry for the day, the date and time is send along with the other data to the server with an http post. But every time I save a entry the time is reduced by 7 hours. So after a few edits even the day itself is wrong.
I can explain the 7 hours, that the time difference between where I am (Peru) and the server (Germany), but where does it come from?

I looked at the database server and it has the time as specified in the web form. So the issue is getting the data back from the server. I checked my JavaScript which pulls the data via AJAX from the server and puts it onto the form, that looked fine too. Then I used Firebug to look at the JSON data that comes from the web server. The time there is seven hours less than whats in the database! I use ASP.NET web methods to return an object serialized as JSON.

The JSON serializer figures out, the client is 7 hours behind the server and takes those seven hours off each DateTime value. How does the server even know about those seven hours. I first thought about the http request headers but there is no time zone information in there. Well I still don´t know how it knows.

To fix the problem I send the timezoneOffset in the browser available through the JavaScript getTimezoneOffset() method when making a request. On the server I find out the time zone offset of the server and then fix all DateTime fields before sending them down. Not nice but it works.

// get a local time zone info TimeZoneInfo
tz = TimeZoneInfo.Local; // get it in hoursint offset
= tz.BaseUtcOffset.Hours; // add one hour if we are in daylight
savingsif (tz.IsDaylightSavingTime(DateTime.Now))
{ offset++; }

The second problem I had was with a very simple site, I added an aspx page to download zip files from the site and log the downloads. It worked fine locally but on the live server the line:

if (System.IO.File.Exists(fileName)

When travelling around the world I plug in my USB devices in countless public computers.

Many viruses use the autorun.inf file to spread themselves around. When plugging in a USB device into an infected computer the virus copies itself into a hidden location, often the Recycler directory and adds a autorun.inf file in the root of the device to automatically start itself on the next computer.

An easy way to notice whether you may have a virus on your device is to create your own autorun.inf and assign an icon to the device.

Create a text file with the name autorun.inf and place it into the root of your device. The content should look like this:

[autorun]
icon=panama.ico

Now place the icon file "panama.ico" into the root of the device as well. The next time you plug in your device the Windows dialog will look like this:

Notice the icon in the top left corner. If you do not see your icon, you know something has overwritten your autorun.inf, most likely a virus.

What to do in this case? Check my post Manually finding and removing malware

I prepared some icon files in case you do not have any:

On TweeNet to access a band page, the old url is

www.twee.net/bands/band.asp?key=smiths

You have to know the internal name of the band as well.

Now you can get to the same page using

www.twee.net/band/The Smiths

The site runs under ASP.Net 3.5 SP1 on IIS6, so I have all the routing goodness of that ASP.Net version.
To get this working, I first had to add the routing assemblies and modules to my web.config

          <
          assemblies
          >
          <
          add
          assembly
          ="System.Web.Abstractions,
Version=3.5.0.0, Culture=neutral, PublicKeyToken=31BF3856AD364E35"
          />
          <
          add
          assembly
          ="System.Web.Routing,
Version=3.5.0.0, Culture=neutral, PublicKeyToken=31BF3856AD364E35"
          />
          </
          assemblies
          >
          <
          httpmodules
          >
          <
          add
          type
          ="System.Web.Routing.UrlRoutingModule,
System.Web.Routing, Version=3.5.0.0, Culture=neutral, PublicKeyToken=31BF3856AD364E35"
          name
          ="UrlRoutingModule"
          />
          </
          httpmodules
          >
        

These are additions to the already existing assemblies and modules which are not listed here.

Next I implemented a routing handler:

          using System; using System.Web; using System.Web.Compilation; using System.Web.Routing; using System.Web.UI; namespace TweeNet.PublicSite.RouteHandlers
{ publicclass BandRouteHandler
: IRouteHandler { public BandRouteHandler(string virtualPath)
{ this.VirtualPath = virtualPath; } publicstring VirtualPath
{ get; private set; } public IHttpHandler
GetHttpHandler(RequestContext requestContext) { foreach (var
urlParm in requestContext.RouteData.Values) { requestContext.HttpContext.Items[urlParm.Key]
= urlParm.Value; } var page = BuildManager.CreateInstanceFromVirtualPath (VirtualPath, typeof(Page)) as IHttpHandler; return page;
} } }

protectedvoid Application_Start(Object
sender, EventArgs e) { RegisterRoutes(RouteTable.Routes); } publicstaticvoid RegisterRoutes(RouteCollection
routes) { routes.Add(new System.Web.Routing.Route ( "band/{bandname}" , new TweeNet.PublicSite.RouteHandlers.BandRouteHandler("~/bands/band.aspx")
)); }

if (HttpContext.Current.Items["bandname"]
!= null) { string userTerm = HttpContext.Current.Items["bandname"].ToString(); //
now use that term to find the band data in the database and display the page content }

The last problem was that IIS6 doesn't pass something like "The Smiths" to the ASP.net pipeline. Steve Sanderson's blog post describes some options how to deal with this 

But in my situation the best option was the following:
As a /band directory did not existed yet, I created that empty folder in the filesystem.

Then in IIS I selected it and created an application for it. I now added a wildcard application mapping to the asp.net runtime. So all requests under /band/ will go to asp.net.

All the other static files on the site are not affected. I then removed the application from '/band/'. IIS still keeps the wildcard mapping even if the folder is no longer an application and you can not see that mapping in the IIS Manager GUI.

I listen to most of the popcasts on the Twit network since 2006 and really like them, they are not super technical like most others I listen to, but they are very entertaining. However the hosts keep repeating the same incorrect statements about Microsoft Windows over and over again which is a bit annoying. The statements below are not quotes but I paraphrased them from memory.

1. Microsoft Beta Software has active debug code and runs slower than the RTM version.
The versions Microsoft releases as public betas are retail code, meaning there is no debug code active and it works just as the final version, otherwise it would be a bad test version. Windows debug versions are called "Checked builds" and are usually for driver developers, these may run slower than the retail version. You can still download and install the Windows debug symbol files for any normal Windows version to troubleshoot certain problems.

2. Windows is an old OS which carries a lot of baggage with it. OS X is shiny and modern.
OSX is based on BSD Unix, first release in 1977 and NextStep, first released in 1989. Windows NT has ideas from DEC’s VMS but was designed and built in the early 90s and released in 1993. Of course OS X was only released in 2000 and incorporates some modern technologies (as do the current versions of Windows NT).

3. Windows is a single CPU platform OS limited to Intel’s 386 platform
Windows NT was especially designed and built as a portable OS. So far it ran on Intel 386, Mips, Dec Alpha, IBM Power PC and Inter Itanium. OS-X on the other hand ran on Power PC and 386 only. Using NT’s architecture it is relatively easy to port it to other platforms. However Microsoft stopped support for less popular CPUs over time.

4. Security and networking in Windows was an afterthought.
The first NT based Windows (3.1) was designed as a multiuser, multiprocessor and multiplatform network operating system and had all the standard security and networking features from the start: Access Control List, user account management, a secure filesystem, TCP/IP stack etc.

5. Microsoft merged the Windows 9x and Windows 2000 code bases to form a common new Windows version with the best of the two previous families.
AFAIK, no code from Windows 9x made it into XP, it was just a tuned Windows 2000 with some extra features. Some of these were first introduced in Windows ME but that doesn’t mean they used the code. Microsoft also increased compatibility with older Win16 applications and games in XP.

6. "New is bad", it's untested and insecure
This claim by Steve Gibson especially against Windows Vista is laughable, Vista is soo much more secure than XP on any level. Leo would say Windows 2000 was more secure when comparing it to XP. Are you kidding me? XP had tons of problems but only a few of them were specific to it, most of the problems were also in 2000. And 2000 has many problems that were fixed when XP came out. So does Mr. Gibson want us to wait until version 3 of his new Cryptolink product before buying it? Surely version 1 and 2 are new and therefore bad and insecure!

7. It is crazy to put the graphics subsystem and its drivers into the kernel.
That was also the opinion of the NT architects in the early 90s, so they put the graphics subsystem into user mode in NT 3.x It turned out that the hardware at the time was not fast enough for a complex OS like NT. So for NT 4.0 they moved the graphics into the kernel, a move not uncontroversial at the time. But it improved graphics performance even though it meant a less stable kernel, problems with drivers and even Blue Screens of Death. In Windows 2000 and XP the architecture was largely unchanged but for Vista they made some big changes. With current hardware fast enough, they moved the graphics subsystem back out of the kernel into user mode. This is one reason why a new driver model was needed. Of course, "new is bad" (see 6) so Mr. Gibson is still using a ten year old OS with the graphics in the kernel.

8. The registry is the root of all evil in Windows
In over 17 years of working with the registry I never had a corrupt one and any other major problems with it, and I hack around inside it quite a bit. Compare it with working with the old ini files or any flat file configuration system. Putting all of the LocalMachine hive into files would be a big mess. How would you protect a single value from being changed, while other values remain writeable? The registry has ACLs on each value. How do you remotely access the configuration of a computer without granting file access? For application level settings, Microsoft promotes the .net framework since 2001 which applications do not use the registry at all, but store it's configuration in XML files. Even non-dot.net Application use or could use this approach.

9. Never use Windows in ATMs or other devices
So they have been bluescreens on huge public screens during the Olympics and malware has been found in ATMs in Eastern Europe. But to blame that on Windows makes it too easy. BSDs usually indicate a hardware problem. If you use cheap or non compatible hardware or bad drivers it is the owner's fault. If the hardware fails, no OS keeps running. How did the viruses get onto those machines? It is likely they were put on the machine during setup, another human security problem. Of course it is easier to use existing Windows malware than to write custom malware for a specific embedded OS, but it is not impossible. I have seen Windows boxes in system critical situations with an uptime of years and no problems. I don't want to say to use Windows in all possible situations but sometimes the use of other OSes in cost prohibitive

10. Open Source is more secure
So how many people are actually looking at all the open source source code? And what are the chances that many of them are doing it to find exploits they can use rather than fix. There are tons of Linux flaws found all the time by the small Unix community alone. What if all the professional crackers would spent more time on Linux code because they could actually made money from their exploits like in the Windows world. What if Microsoft would make the Windows source code public? For years this would give crackers the opportunity to find exploits more easily. Security experts reviewed every single line of active code within Microsoft and found tons of problems but still not everything and exploits still come up. So I doubt just because something is open source, it is more secure. It all depends. Many open source projects are done in a distributed fashion with developers only knowing each other by email. Some code checked in by person A and review by person B can cause a major security problem for code written by person C. At Microsoft and other traditional software houses, the team usually works closer together and such problem can be avoided by communication.

Interestingly enough, none of these myths are promoted by Paul Thurrott, the host of Windows Weekly, the Twit show covering all things Microsoft. However he also does little to debunk them.

Please note, that I am not saying anything like "Windows is better than OS X or Linux", I just want to point out some facts

As I use more and more JavaScript and Ajax on my sites the number and the size of the JavaScript include files grow. What do to about this? jQuery comes in a uncompressed and minimized version, ideally I want to use the first on my Dev-box and the second on the production server. Asp.net 3.5 SP1 comes with the feature CompositeScript which allows us to combine several script into one download. Well, I am trying to get away from the asp.net ScriptManager, so I came up with my own solution.

The first version supports the following:

During development the fullsize script files are used

In Production mode, a single minimized file is used that includes all the JavaScript code needed for the page.

This file is not generated on the fly, instead it is stored on the file system. This makes it more scalable.

Here’s how in works:

All my common JavaScript include files are in one location /css/. Instead of referencing them with

          <
          script
          src
          ="../css/jquery.js"
          type
          ="text/javascript"
          ></
          script
          >
          <
          script
          src
          ="../css/topas.js"
          type
          ="text/javascript"
          ></
          script
          >
        

On the page, I just have a single literal control:

          <
          asp:Literal
          runat
          ="server"
          ID
          ="litSciptIncludes"
          ></
          asp:Literal
          >
        

Which results in the following html:

          <
          script
          src
          ="../css/lib/3.js"
          type
          ="text/javascript"
          ></
          script
          >
        

3.js has the compressed content of both jquery.js and topas.js

In the Page_Load method in code behind I have some code to add scripts:

ScriptIncluder inc = new ScriptIncluder();
inc.AddScript(ScriptIncluder.Scripts.jQuery); inc.AddScript(ScriptIncluder.Scripts.Topas);
litSciptIncludes.Text = inc.GetIncludes();

The ScriptIncluder class is responsible for building the <script src…> tags. We add all the scripts we need on the page with the AddScript method, a fixed number of scripts are defined in a enumeration.

[Flags]
publicenum Scripts
{ jQuery = 1, Topas = 2, TopasDate = 16, TopasEdit = 32, jQueryUI = 256, jQueryMaskedEdit
= 512, } publicvoid AddScript(Scripts
scriptID) { _scripts.Add(scriptID); _scriptCode += (int)scriptID;
}

In my application framework I have a setting in the web.config which defines the Debug-Mode, when in development I just loop through all the added scripts and return one <script> tag for each added script. However when in production, I return a single script tag that points to a single file.

          public
          string GetIncludes()
{ string html = string.Empty; string start
= "<script type=\"text/javascript\" src=\""
+ Core.Config.VirtualRoot + "css/"; string end
= "\"></script>" + Environment.NewLine; if (Core.Config.CurrentDebugLevel
>= Core.Config.DebugLevels.SharedDev) { // include plain
text filesforeach (Scripts script in _scripts)
{ html += start + _scriptNames[script] + end; } } else { //
include compressed file html = start + "lib/" + _scriptCode
+ ".js" + end; } return html; }

_scriptNames[] is a generic list of all the script file names, _scriptCode is the sum of all the scriptIDs as defined in the Scripts enumeration.

So where do all the /css/lib/nn.js files come from that are needed for the single compressed file?

I use a batch file to create them:

pushd \websites\foo\wwwroot\css
jsmin.exe lib/769.js jquerymin.js jqueryuimin.js jquery.maskedinput.js
jsmin.exe lib/819.js jquerymin.js jqueryuimin.js jquery.maskedinput.js topas.js topas.date.js topas.edit.js
jsmin.exe lib/49.js topas.js topas.date.js topas.edit.js
jsmin.exe lib/3.js topas.js jquerymin.js
jsmin.exe lib/1.js jquery.js
popd

In theory I need hundreds of these /lib/nnn.js files, but in practice there are only a few combinations that I use. What is jsmin.exe? It is a small tool based on Douglas Crawford’s jsmin (http://www.crockford.com/javascript/jsmin.html). He has C# code to minimize a single JavaScript file. I added some code to process any number of files into a single file. I can just run the batch file every time I have changed any of my JavaScript include files and can also make it part of my build process.

Any drawbacks? If every page uses a different combination of include files, all these combinations have to be downloaded once and then cached, so the actual jQuery code may be downloaded several times.