On closer investigation, it turned out I was hitting the default maxReceivedMessageSize (set to 65536 bytes) when POSTing the data on the basicHttpBinding binding I was using, as shown on the Fiddler trace above (highlighted in red) — but only just.

How to get around it.

The WCF service was defined by an endpoint using the basicHttpBinding binding, so it was a case of adding the following:

<system.serviceModel>
    <bindings>
      <basicHttpBinding>
        <binding name="" maxReceivedMessageSize="262144"></binding>
      </basicHttpBinding>
    </bindings>
    ...
</system.serviceModel>

Here I am setting the default basicHttpBinding binding to accept messages up to 256KB in size. This immediately fixed the problem. This comes with a caveat though as setting this value too high on production code could increase your exposure to possible DoS attacks from large payloads.

Bookmark

What this gave me was the following MSBuild arguments that I put in Process -> Advanced -> MSBuild Arguments, within a TFS build definition I had set up:

/p:DeployOnBuild=True /p:DeployTarget=MsDeployPublish /p:Configuration="Staging" /p:CreatePackageOnPublish=True /p:MsDeployServiceUrl=https://<server name>:8172/msdeploy.axd /p:AllowUntrustedCertificate=True /p:username="<user name>" /p:password="<password>" /p:MSDeployPublishMethod=WMSVC /p:SkipExtraFilesOnServer=True

I was looking for a way to emulate the option you get when you import an application package in IIS (with the web deploy extension installed), which lets you chose whether you want to overwrite existing files in the destination, or not, as I wanted.

Unfortunately, the documentation for MSDeploy seems a little thin on the ground, but it seems argument I was lacking was /p:SkipExtraFilesOnServer=True. Adding this to the end of the above MSBuild arguments gave me the required behaviour, resulting in the following MSBuild arguments:

/p:DeployOnBuild=True /p:DeployTarget=MsDeployPublish /p:Configuration="Staging" /p:CreatePackageOnPublish=True /p:MsDeployServiceUrl=https://<server name>:8172/msdeploy.axd /p:AllowUntrustedCertificate=True /p:username="<user name>" /p:password="<password>" /p:MSDeployPublishMethod=WMSVC /p:SkipExtraFilesOnServer=True

Bookmark

I had declared the DataGrid control as follows in Window.xaml:

<DataGrid AutoGenerateColumns="False" EnableRowVirtualization="True"
ItemsSource="{Binding Source={StaticResource AlbumsSource}}" Name="AlbumsDataGrid" Margin="5"
RowDetailsVisibilityMode="VisibleWhenSelected" Grid.Row="1">
            <DataGrid.Columns>
                <DataGridTextColumn x:Name="IdColumn" Binding="{Binding Path=Id}"
Header="Id" Visibility="Hidden" />
                <DataGridTextColumn x:Name="NameColumn" Binding="{Binding Path=Name}"
Header="Album Name" Width="*" />
            </DataGrid.Columns>
</DataGrid>

in which

... ItemsSource="{Binding Source={StaticResource AlbumsSource}}"

is responsible for binding the items to the data grid. I had set a collection of resources on Window that I was below the DataGrid as follows:

<Window.Resources>
        <CollectionViewSource x:Key="AlbumsSource"></CollectionViewSource>
</Window.Resources>

Then I set the data source of the CollectionViewSource I’d defined in Window.Resource in the code-behind as follows, in the Window_Loaded event handler:

private void Window_Loaded(object sender, RoutedEventArgs e)
        {
            //look up AlbumsSource in resource dictionary
            CollectionViewSource src = ((CollectionViewSource)Resources["AlbumsSource"]);

            using (var ent = new MusicTrackerContainer()) {
                src.Source = ent.Albums;
            }
        }

Where MusicTrackerContainer is simply a Entity Framework model. This did not work as expected, instead throwing an XmlParseException stating ‘Provide value on ‘System.Windows.StaticResourceExtension’ threw an exception.’.

To fix this, the Window.Resources element should be declared in the XAML before it is referenced in the Binding. I moved the Window.Resource element to be the first child of the Window element and it worked. XAML follows:

<Window x:Class="MusicTracker.MainWindow"
        xmlns="http://schemas.microsoft.com/winfx/2006/xaml/presentation"
        xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml"
        Title="MainWindow" Height="350" Width="525" Loaded="Window_Loaded">
    <Window.Resources>
        <CollectionViewSource x:Key="AlbumsSource"></CollectionViewSource>
    </Window.Resources>
    <Grid>
        <Grid.RowDefinitions>
            <RowDefinition Height="*" />
            <RowDefinition Height="*" />
        </Grid.RowDefinitions>
        <DataGrid AutoGenerateColumns="False" EnableRowVirtualization="True"
ItemsSource="{Binding Source={StaticResource AlbumsSource}}" Name="AlbumsDataGrid" Margin="5"
RowDetailsVisibilityMode="VisibleWhenSelected" Grid.Row="1">
            <DataGrid.Columns>
                <DataGridTextColumn x:Name="IdColumn" Binding="{Binding Path=Id}"
Header="Id" Visibility="Hidden" />
                <DataGridTextColumn x:Name="NameColumn" Binding="{Binding Path=Name}"
Header="Album Name" Width="*" />
            </DataGrid.Columns>
        </DataGrid>
    </Grid>
</Window>

Bookmark

ASP.net impersonation

Though WCF has been designed to be transport-independant, you can utilise ASP.net compatibility mode, which gives you the opportunity to get your services to impersonate as a given user through standard ASP.net means, by adding the following to your web.config:

<system.web>
    <identity impersonate="true"/>
</system.web>

This will cause your services to impersonate the current user. See http://msdn.microsoft.com/en-us/library/ff647404.aspx under the section ‘To impersonate the original caller’, which does a great job of explaining this.

Then, if you want ALL your WCF service requests to run under the identity of a pre-defined user, the identity element further supports the username and password attributes, e.g.

<system.web>
    <identity impersonate="true" username="user" password="pass" />
</system.web>

Where user is a local or domain user. The above will give you access to the resources that user is capable of accessing.

If you want to take advantage of this for your WCF services, you should configure your services as follows:

1. In your web.config, set the aspNetCompatibilityEnabled attribute to true on the system.serviceModel element:

<system.serviceModel>
    ...
    <serviceHostingEnvironment ... aspNetCompatibilityEnabled="true"/>
</system.serviceModel>

2. On your service implementation (the class that implements your service contract), set the AspNetCompatibilityRequirementsMode attribute at the class level.

[AspNetCompatibilityRequirements(RequirementsMode=AspNetCompatibilityRequirementsMode.Required]
public class Service : IService {
    public bool DoSomeWorkThatRequiresElevatedPermissions(int id)
}

The AspNetCompatibilityRequirementsMode attribute will ensure that the service instantiation fails if the aspNetCompatibilityEnabled attribute has not been added to the web.config.

This should be enough to get the service to run as the impersonated user you set in the web.config, and on the surface it seemed to give me the appropriate behaviour, with the WCF services able to access folders I had secured for access only by a given user – until I began using the Task Parallel Library (TPL) from within the service operations to spin off some long running operations in seperate threads. TPL provides a nice clean API to spin off threads as Task‘s, but it seemed the running of the service in the context of the impersonated user in the web.config did not extend to any threads spun off by the service.

Run WCF service as….

In order to enable WCF services to run as a specific user, I thought of extracting the services that required elevated permissions and have them run as a windows service. Or instead, with the services being hosted under IIS, run the services under a separate application pool and set this application pool’s identity to the user with the permissions to access the required resources. This solved my problem – any threads spun out as part of the service ran under the context of the new application pool identity, meaning you can use all the loveliness of the TPL.

To take advantage of this, head to the Internet Information Services (IIS) Manager. I’m using Windows 7 64-bit, which comes with IIS 7 by default, but most of these instructions should apply to earlier versions also, though the location of the settings panels will be different.

Select ‘Application Pools’ over in the left hand bar of the IIS Manager and create an app pool relevant for your WCF services (.NET 4.0/Integrated pipeline should do). Here, I’m going to call mine ElevatedAppPool.

Once you’ve created it, select it and click ‘Advanced Settings’ in the Actions pane on the right hand side of the IIS Manager. Under Process Model you should see an ‘Identity’ element, which if you click … you can assign a custom account to it. This can be either a domain or local machine account.

Finally, go back to the application that your WCF services are located by expanding the Sites tab on the left hand side of the IIS Manager. Click ‘Advanced Settings’ and select the application pool you just set up. You should now find your WCF services are running under this user with elevated permissions.

You could of course, have given the ApplicationPoolIdentity the extra permissions (such as reading or writing to a particular folder outside wwwroot) but having a seperate identity/application pool makes the changes specific to this set of WCF services, and not for other application pools that may be using the ApplicationPoolIdentity user.

Bookmark

Making good use of the command window in Matlab I wanted to recover this history – so armed with the hard drive and not much else I went looking for where the history of the command window is kept.

The file you’re looking for is history.m, which if you are using Vista/Windows 7 you’ll find in the following folder:

C:\Users\user_name_here\AppData\Roaming\MathWorks\MATLAB\R2008b\history.m

* of course you’ll need to replace R2008b with the version you are using also!

Bookmark

4. Click the ‘Specify IP addresses’ and type an IP range that is in line with your current router setup (e.g. if your LAN is on a 192.168.11.x network, then you might want to set it as follows):

You should now find, like me, that the VPN works once more! I’ll post up a tutorial on how I got it working and accessible to the public Internet soon.

Bookmark

Removing shading

The following article also applies to Excel 2010. Unfortunately there still is no option to remove the shading from the user interface, so you have to resort to a bit of VB scripting.

Remove shading from a surface chart in Excel 2007 - http://social.answers.microsoft.com/Forums/en-US/excelchart/thread/2b147908-636b-4445-9262-46463406accb

Colouring the plots

Here, the essence is that to colour the plot yourself, you have to colour each legend item individually. Do this by selecting the legend (make sure it’s not the entire legend!), then right click and select ‘Format Band’. Original article: Formatting Surface Charts part in Surface and Contour Charts in Excel 2007 - http://pubs.logicalexpressions.com/pub0009/LPMArticle.asp?ID=447

Removing filter labels

The filter labels serve a very good purpose when you want to filter individual rows or columns but get in the way a little if you want to embed the graph into a document. If you want to remove the filter labels, the toggle button to do so is tucked away in PivotChart Tools, which pops up if you select the surface plot derived from Pivot Table data. Then click the Analyse tab and click ‘Field Buttons’ to toggle between showing and hiding the filter labels.

Bookmark

My problem was to set PHP directives on a per site basis in order to enable large file uploads on a single web application running on the web server. I also wanted to modify the temporary upload folder PHP uses for file uploads, and enable file upload progress monitoring through the APC extension (preferrably only on the file uploader web app). The directives corresponding to these are in a number of different “modes”, so I was looking for a way that provided complete freedom over specifying the directives.

Depending on your PHP install and operating system there are several ways to customise your site’s PHP configuration. PHP itself can run in CGI mode, or as an Apache module. I’ll leave it to http://docs.joomla.org/Should_PHP_run_as_a_CGI_script_or_as_an_Apache_module%3F#What_is_the_difference_between_CGI_and_apache_Module_Mode.3F as it does a great job at explaining security and performance differences between PHP running in CGI or Apache module mode.

Determining your PHP runtime environment.

To determine the best way for configuring PHP on a per-site basis you’ll need to determine your PHP runtime environment (i.e. your operating system and whether PHP runs as a CGI or an Apache module). So, create a script that outputs your PHP runtime environment information:

You should then look at the “Server API” field, which will say something like “CGI/FastCGI” for PHP running as a CGI:

or “Apache Handler” for PHP running as an Apache module:

Now we’ve identified the mode of operation, I’ll explain some general methods for customising PHP directives on a per-site basis:

General per-site configuration methods

This should work on PHP running in any environment (Linux/Windows/Mac/etc) as either an Apache module or CGI.

Use ini_set() (limited number of options available)

You can customise a limited number of PHP directives on a per-script basis, by adding ini_set() to the top of your scripts. You can adjust PHP_INI_USER directives within scripts. See http://uk.php.net/ini_set for more information on this.

Allow user defined INI files (greater number of options available)

Edit your php.ini to permit user defined .ini files, by looking for a field user_ini.filename (which might be currently commented out). Uncomment it and state the filename you want to use for the user defined .ini files (which by default is .user.ini).

Then, within your web application (e.g. C:\phpwebs\helloworld), create a new file called .user.ini and add in the directives to suit:

upload_max_filesize = 1024M
post_max_size = 1500M

You can adjust PHP_INI_PERDIR and PHP_INI_USER directives in user defined INI files, which may or may not be enough. In my case, it was not, as I wanted to also update upload_tmp_dir, a PHP_INI_SYSTEM directive.

Define folder and directives in php.ini (most, if not all options available)

In this approach, you edit php.ini wherever it is located (see output from phpinfo() above), and at the end add your site-specific configuration, by specifying the folder that the web application resides in. Lets say our web app is located in C:\phpwebs\helloworld\. To configure the PHP runtime on this web app, add

[PATH=C:/phpwebs/helloworld/]
upload_max_filesize = 1024M
post_max_size = 1500M

At the end of php.ini.

Note virtually any directive can be customised on a per-site basis using this method, and this was the approach I went for when hosting a PHP web app on IIS to in order to adjust upload_tmp_dir for the uploader app. What I didn’t seem to be able to do was to enable some extension based directives on a per site basis. The progress meter failed when apc.rfc1867=on was specified under the per-site declaration, so I made this part of the global php.ini.

Per-site configuration options with PHP as a CGI in Windows IIS

Any of the general per-site configuration options mentioned above should work. You can also edit the registry to manage a subset of PHP directives, illustrated at http://www.php.net/manual/en/configuration.changes.php, in the section entitled ‘Changing PHP configuration via the Windows registry’. The process is as follows:

• Open up regedit
• Navigate to HKLM\SOFTWARE\PHP\Per Directory Values
• Create keys corresponding to the directory hierarchy of your web app and at the child node specify the customised directives via New > String Value. Using our C:\phpwebs\helloworld\ example see the screenshot below for how you modify PHP directives:

Note that you can customise only PHP_INI_USER directives in this manner.

Per-site configuration options with PHP as an Apache module

All-in-one *AMP stacks (such as XAMPP) mostly have PHP running as an Apache module. Any of the general per-site configuration options mentioned above should also work for PHP configured as an Apache module.

If your PHP install in running this way, you can also edit httpd.conf file to specify PHP directives for virtual hosts and additionally directories and subdirectories exposed by the virtual host.

To do this:

1. Ensure AllowOverride All in httpd.conf. This should reside in the conf folder in the Apache folder.

2. In XAMPP, virtual host declarations are containerised and are specified in <apache_dir>/conf/extra/http-vhosts.conf. In this file you would type the following:

DocumentRoot "C:/phpwebs/helloworld"
ServerName www.helloworld.com
php_admin_value upload_tmp_dir "C:\tmp"
php_value post_max_size 1500M
php_value upload_max_filesize 1024M
php_admin_value apc.rfc1867 on

With this method you have complete freedom over which PHP directives you want to set. Valid apache directive syntax can be found at http://www.php.net/manual/en/configuration.changes.php.

Bookmark

1. Support for citing as parenthesised and non-parenthesized references like in LaTeX. Unfortunately Word only supports the former, which is fine when citing references numerically, but it does not provide the flexibility which is important for Harvard author, date referencing. There is no automatic workaround that I am aware of to assist in this.
2. Search and import citations into the Source Manager via common Internet repositories such as ieeexplore, citeseer, Google Scholar etc, or provide a mechanism by which this functionality can be plugged in. I’m currently using JabRef as my reference manager which provides a search of these libraries.
3. A larger selection of bibliographic styles provided out of the box. In the meantime, BibWord, available on Codeplex, has a number of alternative bibliographic styles, including variations of Harvard style referencing.
4. Make it easier to cite multiple authors within the text. Currently, the process is a little clunky.
5. Import other bibliographic formats natively. I use JabRef as a middleman to store citations as it is able to deal with citations in a number of formats (Endnote, .ris, .bib etc). I then export the JabRef library into a Word 2007 XML bibliography. See my article on how to export a bib using Jabref into XML format in order to read it into Office 2007/2010.
6. Offer a search when inserting citations so if an author/author tag or some other information about the reference  is known, you can filter the list without having to scroll through a (potentially long) list of citations.
7. Sort the performance issues when rendering extremely large reference lists. To mitigate any performance issues you can export just the citations you need to Office XML format (see 5.) from within Jabref, rather than your entire library.

Bookmark

I originally thought that APC was statically compiled into Zend as it pops up in the list of extensions in the Zend dashboard:

But, on closer inspection of phpinfo(), there was no reference to any apc directives at all.

So here is how I did it:

Enabling Alternative PHP Cache (APC) extension in Zend Server CE

1. Download non-thread safe VC9 APC DLL (I went for php_apc-5.3-nts-svn20100226-vc9-x86.zip) from http://downloads.php.net/pierre/
2. Unzip it and put the extension dll php_apc.dll into C:\Program Files\Zend\ZendServer\lib\phpext (assuming you left the Zend Server install location as its default)
3. Add the following to your php.ini (in C:\Program Files\Zend\ZendServer\etc):

   ;Enable APC
   extension=php_apc.dll
   ;Enable upload progress bar
   apc.rfc1867=on

4. Give Apache/PHP a restart

Now check out phpinfo() on the Zend dashboard and you should see something like the following, with apc and the various directives being output in the configuration part of the output.

Any of the APC file upload progress bar plugins (e.g. http://www.ibm.com/developerworks/library/os-php-v525/index.html) should now work.

Bookmark