I tend to be the type of person that likes to separate out the log4net configuration into it’s own file, usually log4net.config.  Setting up this new project; however, I started running into a problem I’d never seen before.  When trying to use the XmlConfiguratior to read my log4net.config, I would see this exception:

[SecurityException: Request for the permission of type 'System.Security.Permissions.FileIOPermission, mscorlib, Version=2.0.0.0, Culture=neutral, PublicKeyToken=b77a5c561934e089' failed.]

Searching for others with this problem has not yielded much success.  I’ve seen articles stating that it’s an issue with medium trust security (on m local workstation trust was set to full) to problems with some sort of breaking code within log4net itself.  Most of these articles I suspect were not using IIS7, but it was hard to tell.

I ended up digging into the log4net source code to find out what was happening.  The problem would occur on a line of code that tried to access the FullName property of a System.IO.FileInfo object.  That property threw the exception worked fine it I tried to access it from my web project, but once it got down into the log4net guts, it would not.

After a lot of frustration, I finally started looking elsewhere.  I ran across a comment on Stack Overflow that stated they used .xml instead of .config files.  I had dismissed it due since others claimed .config files where fine, but recalling what I’d seen there I decided to try it.  *Poof* it worked!

I didn’t really want to leave the configuration in a file with the .xml extension since that could easily be downloaded from a server.  The discovery told me that it was likely due to something IIS was doing with ASP.NET to hide files, but the odd thing was that it could read from the web.config, so I was a little perplexed.  I started digging into how IIS7 handles these types of protected files. 

It turns out that there’s a section in the hosting configuration that lists protected files under the name “requestFiltering”.  Hmm…that was an interesting sounding name.  Unfortunately all the entries were only by file extension, not by file name directly so there had to be something else.

I ended up in the IIS7 Manager application and found the Request Filtering area.  I began to poke around in there and discovered another tab called Hidden Segments which had an entry for the web.config!  I clicked the Add Hidden Segment link under Actions and added a new entry for log4net.config and viola, my application worked!

I know that most of the IIS7 configuration settings are stored in various config files, including the web.config, so I started looking in there and found this new element in the <system.webServer> section.

   1: <security>

   2:     <requestFiltering>

   3:         <hiddenSegments>

   4:             <add segment="log4net.config" />

   5:         </hiddenSegments>

   6:     </requestFiltering>

   7: </security>

I believe that including this section will allow the files to work.  I even changed my trust level in my dev environment to Medium and it still ran just fine.

There may be other things that can cause this sort of problem, but this fixed my example.  Hopefully this helps someone else out there as it was a bear to track down!

After much searching I was able to determine that the PowerCommands for Visual Studio 2008 extension was somehow involved in the problem.  I tried quite a few different solutions including adding a dependantAssembly reference in my Visual Studio configuration, doing a clean solution then rebuilding the project.  Up to that point the only thing that worked was uninstalling the extension.  Reinstalling afterwards would cause the problem to return.

Working without PowerCommands was an option I did not want to consider.  Many times each day I would use functions such as “Open Command Prompt” which fires up cmd.exe starting in the directory of the file or folder selected in the solution explorer or “Open Containing Folder” on a file, just to mention a couple.  It was looking like that was going to be the only solution, I was bummed.

I decided to check with the source a bit further.  I looked through their issue list and the top voted issue was VS 2008 crashes when adjusting toolbox items.  Reading the description did not give me a lot of hope as it dealt with a problem when the toolbox items were adjusted.  It also included an actual error which was lacking in my case.

Among the comments to the above issue, I found the suggestion to run the command “ngen /delete *” (without the quotes) from a Visual Studio 2008 command prompt.  Ngen is the native image generator which creates native images from managed assemblies and installs a native image cache on a computer, which is a reserved area of the global assembly cache.

Wow, that sounds important, and I’m suppose to delete * (star)?  Ok….I was desperate and I did it before investigating what it meant.  Dangerous, I know, I know, but as I said I was desperate!

It worked.   Nothing seemed to break…and it worked.  Ok, off we go!  Wait, time to investigate….just in case problems start cropping up later.

Running ngen /? from the command prompt yielded no information, not a mention of the /delete switch.  In my experience, when a command line utility fails to mention an argument that clearly did something, this is an argument that should only be used with full knowledge of what it does.  Now I started to get nervous!

I did a quick Google Search and was lead to the MSDN documentation for ngen.  It seems that my little command deleted all native images in the native image cache.  Hmm….that sounds bad.  Fortunately, a little further down the page the following information was also present.

This made me feel a little better.  The assemblies were still there, simply not the native images for them.  Native images simply cause the applications or assemblies to start up a bit faster.  With today’s hardware, I don’t know that I’ll even notice a difference, or if I do, it will be worth it to keep PowerCommands and get rid of the problem.

If this behavior resurfaces, I will do a little more investigation by using the /show argument to find out what native images are installed.  Then I will started deleting them one at a time to find the culprit.  My guess is that one of the native images somehow became corrupted, or the configuration for it got messed up or something like that.  Perhaps finding the culprit will aid in the resolution of the issue that this wonderful extension is experiencing by, if the Google results are any indication, a fair number of developers.

On the up side, installing from source has never been easier.  In the past I have went through many hours of trying to get the right versions of different libraries that were needed.  With the official release of 2.0 it seems much better.  I thought I would share the steps that I went through.

Disclaimer: This worked on a fairly fresh install of CentOS 4.7.  I have not tried it on 5.x, nor on any other flavor of linux (SUSE, Ubuntu, etc..) so your mileage may vary.

At the time of the install (and this writing) the current mono stable version is 2.0.1 so all references will be to that version.  Here are the steps that I went through.

Preparation

Always be prepared – Boy Scouts motto…

In rooting around the web I did find a few helpful pointers.  First, make sure you have gcc installed.  Now this is one of those duh pieces of information, but in the fairness of completeness I thought I would mention it. (Note: If you do not have gcc or bison, install them! Credit The_Assimilator’s comment)

# yum install gcc-c++
# yum install bison

yum install httpd-devel

You may also require the glib-2.0 libraries (thanks to Michael Walsh for that bit).  If you receive the error “Package glib-2.0 was not found in the pkg-config search path” you can install it via yum as well.

yum install glib2-devel

The Main Dance

Next comes the meat of the installation.  First, I downloaded the necessary source packages.  I simply used wget to snag the core mono package, xsp (mono web server) and mod_mono (apache integration).

wget http://ftp.novell.com/pub/mono/sources/mono/mono-2.0.1.tar.bz2
wget http://ftp.novell.com/pub/mono/sources/xsp/xsp-2.0.tar.bz2
wget http://ftp.novell.com/pub/mono/sources/mod_mono/mod_mono-2.0.tar.bz2

Next we install the mono core

tar -vxjf mono-2.0.1.tar.bz2
cd mono-2.0.1
./configure
make
make install
cd ..

Next comes xsp

tar -vxjf xsp-2.0.tar.bz
cd xsp-2.0
./configure
make
make install
cd ..

At this point I recevied an error (I believe it was in the make process) that the compiler could not find the file dotnet.pc.  I found that it was indeed on my system so I simply had to export the path and then finsih the compile.

export PKG_CONFIG_PATH=/usr/local/lib/pkgconfig
make
make install
cd..

Note: Make sure the file dotnet.pc is in that location.  If not, adjust the path above.

Finally we install mod_mono

tar -vxjf mod_mono-2.0.tar.bz2
cd mod_mono-2.0
./configure
make
make install

There, easy huh?

Configuration

You may want to verify a few thigns to make sure the configuration is ready to rock.  In my case, I am keeping the mono configuration in a separate file for sanity sake.  You can do that or put it all in your httpd.conf, it’s up to you.

<IfModule !mod_mono.c>
    LoadModule mono_module /usr/lib/httpd/modules/mod_mono.so
    AddType application/x-asp-net .aspx
    AddType application/x-asp-net .asmx
    AddType application/x-asp-net .ashx
    AddType application/x-asp-net .asax
    AddType application/x-asp-net .ascx
    AddType application/x-asp-net .soap
    AddType application/x-asp-net .rem
    AddType application/x-asp-net .axd
    AddType application/x-asp-net .cs
    AddType application/x-asp-net .config
    AddType application/x-asp-net .Config
    AddType application/x-asp-net .dll
    DirectoryIndex index.aspx
    DirectoryIndex Default.aspx
    DirectoryIndex default.aspx
</IfModule>

Update: I finally was able to start downloading.  Apparently MS decided to use a different distribution method (Akami?) to get over the initial rush, or at least that is my guess on the matter and it is purely my own speculation.  The download location was not the normal MSDN subscriber downloads area.  In addition, I had to install a new download manager and of course allow popups from msdn2.microsoft.com before I could get it started, but now it’s on is own merry way to my computer!  It’s coming down at a good clip too, ,about 570 KB/sec so if things remain as they are I should have it within a couple of hours. 

Good luck to anyone else, hopefully grabbing this one

Today I noticed that page 4, one with a large viewstate, is showing up #1 at Google.  To my knowledge, this is the first time one of the pages has received the #1 spot.

I’m not sure why the other pages are not showing up.  Perhaps they are being marked as duplicate content or something, but it is interesting that the page with Viewstate is the one ranking right now.

1. Initially the index page with a link to each of his tests ranked #1.  This went on for most of last week.
2. One of the tests ranked under the index page when you click on the “View similar results” link on the initial Google search.
3. My original post was ranking #2 for most of last week.  It only had the word “arkliode” once in one of the comments.
4. This morning, my original post shows up #1.  Yes, that is the post with the word only displayed once in a comment.
5. Google gives more weight to blog posts.  This has been generally accepted for quite a while.  That explains the change in #1 ranking.
6. Incoming links play a large part in the role.  I am attributing that reasoning to the drop of the one page that was returning in the results.  The incoming links simply go to the index page.
7. Duplicate content has been assumed to have a negative impact.  Since most of the test content pages have the same or very similar content, I am theorizing that Google is recognizing them as duplicate and since the only links going to them are from the index page.

The next question is, how do we begin to make the test valid?  I suppose we would have to post the different pages on separate sites to try to get a better idea, yet the popularity and ranking of the sites would undoubtedly play into the ranking.

Dave has changed things up a bit, which might account for the changes over the weekend. Should people begin linking to the pages on their sites to see if indexing begins to happen?  That’s what Dave did.  Here are the links to the pages.

• test
• test
• test
• test
• test
• test

I will check back in later this week or early next week.  I’m sure Dave will have at least one update in that time-frame as well.

I have had my first real look at Microsoft Team Foundation Server and I must say I am not impressed.  It looks like it has some nice features, however I would not say that it is worth the price of admission.  Licensing seems a bit high and the preferred method of using it via Visual Studio is a bit clunky.  The add-in (Team Explorer) works well enough, but it is plain and seems to bloat VS as many add-ins do.  If I am going to settle for reduced performance in my development IDE I would expect more out of it.   At this time I am looking into SharpForge, a somewhat new and growing open source project designed to perform some of the functionality of TFS or SourceForge.  The web interface from TeamPlain, recently acquired by Microsoft, is not bad.  I will admit that I am a bit predisposed against the entire system since it is based on Microsoft SharePonit Portal Server.  I have yet to be convinced that the platform is desirable enough to move to.

In the mean time, I also passed another test on my way to MCSD and then MCPD.  70-315, Developing Web Applications with Visual C#.NET is complete.  On to the next one!  Target date is set for 3 weeks, although budget may push it to next month.  I hate tax time. 

Cheers!

One of my favorites is “Unexpected Error”.  Well, of course it is unexpected, otherwise I would expect the program to handle more elegantly, provide more information or perhaps fix it!  A friend of mine that is a former Microsoft employee says that this message is one of his pet peeves, not just in Visual Studio mind, but in any application.  I met him ten years ago when we were both on contract to the same company and I can remember back then his disdain for a programmer who let one of these slip through.  Laziness, nothing more, is responsible.

You might ask, how can any developer anticipate every single exception that may occur?  They can’t of course, however even for those “unexpected” errors they should have something better to tell the user.  Unexpected error is almost a redundancy for goodness sake!

Thanks for the interested read Dave.

David points out the way some of the features that have been rolled into the .NET Framework or into the Visual Studio Team System lately have had roots in the open source community.  Examples of the unit testing, MS Build, etc..  Most of these are welcome additions to developers that have lived without them, however those of us that were early adopters of NUnit, NAnt and other projects realize that these pale in comparison to the current releases.

Microsoft has done a lot for the development community, don’t get me wrong.  I love the .NET Framework, the productivity and power it brings.  I also believe that Microsoft has had some successes with their new features.  The ASP.NET AJAX extensions exceeded my expectations.  After using other open source libraries such as Michael Schwarz’s Ajax.NET, and then seeing some of the earlier ATLAS releases I though that Microsoft was on a path to “do it again.”  I decided to give the RTM a try and was pleasantly surprised with the jump they made from some of the CTP’s to release.  Way to go guys!

I have two friends that are former Microsoft employees.  One was with MS for over a decade and as such has what I (and others) like to refer to as “Microsoft Blinders”.  His enthusiasm is great, however he fails to give most non-Microsoft technologies much of a chance.  This is too bad and I feel that, although he is brilliant, this limits him slightly.  Don’t get me wrong, I highly respect his skills, his architecture and he is a great person.  It does lead me to believe that many inside of Microsoft share this viewpoint and that may be the reason they bring some of these features on the way that they do.

Scott and crew, please take Davids letter seriously.  There is an army of .NET developers out here who are genuinely happy with the framework.  Please don’t dishearten those of us that have used these open source libraries with less-than implementations of your own and do not make us wait for the fabled “version 3″ which we all feel to be the first “usable” release of most Microsoft products.  When there is already something in the community that fits the bill, giving us less hurts.  You have the resources and the talent, live up to the vision!

One thing though David, what’s with the rabbit ears?  I suppose it may be antenae however such as those worn by Arthur from The Tick.

A product we are currently working on has a process that, well, processes a lot.  It goes through several different data gathering, manipulation, saving and printing operations.  The end result of this process is a print job that takes about an hour and produces about 1000 printed pages.

For the development we normally sent the jobs to a PDF printer or simply had it stop after printing 20 or 30 pages.  Finally the time came to give this a real test, a complete dry run!

*poof*

I know what you’re thinking, “you should have done that in Dev at least once!”  You are right of course, however sometimes we let things slip due to schedules and pressure.  Lesson learned, I hope!

It appeared as though there was a memory leak causing the application to crash.  Monitoring the memory usage with Process Explorer confirmed this to be the case.  Now to the task of tracking it down.

I must admit that I have never had a leak like this one.  After some initial code reviews there were a few places where we were able to determine the potential for problems.   Implementing code to fix these “phantom menaces” were not successful.  Now it was time to really dig in.  The downside was, I did not know how to dig, and I didn’t have a shovel.

After some searching around on Google, I ran across Finding .NET Memory Leaks by Phil Write.  It was not the easiest thing to find, but it was well worth the time.  Phil goes step by step through using sos.exe (Son of Strike) debugging extensions.   He explained the basis of how to track down what you think the problem is.  Unfortunately the problem was not that easy to find.  I ended up doing comparisons of the output of !dumpheap -stat from very early in the process and another dump from much later on down the line.  It was a tedious exercise, but a necessary one.  Finally I happened upon an object that had a large jump in it’s count between the two samples.  Now I had a place to start!  Using Phil’s instructions again I was able to find out what was holding on to a reference and implement a fix.  It also lead me to a second leak that we did not know existed and had been around for quite a while.  It turned out that the first leak that we fixed would not have been a problem if the other one had been behaving properly.

This is a good example of why bugs can be good.  The second memory leak will be taken care of within the next day or two and the product will be that much better for it.

Success!

Thanks Phil for such a wonderful and simple to understand article!

EDIT: 8/18/2010 – updated link to Phil’s article.  Thanks Aaron D for pointing it out!