Wolfgang Riedmann Blog

In WPF, you can set the SelectedDate property to any DateTime value, but this value is not shown as initial date in the calendar control.

If you wish to set the initial date, you need to set the DisplayDate property.

If you use databinding, you can bind both values to the same property on the ViewModel, or to two different, what you like more. Of course in the latter case the initial value needs to be set to both properties.

Another suggestion: if you work with the Calendar control directly, you should embed it in a ViewBox control to have it adjusted when you resize your window.

In this post I will try to list some things that are needed for registration-free use of Vulcan.NET or X# components in Visual Objects applications. Some of these issues have costed me several days of tries.

First, a background article from MSDN: How to: Configure .NET Framework-Based COM Components for Registration-Free Activation

Then: a big “Thank You” to Meinhard Schnoor who has helped me a lot not only with this issue, but often also with other issues.

How SxS (how I will call it to shorten) works basically? I’ll try to simplify at maximum what is needed and where the potential problems stay.

Normally, COM works only with “registered” components. This has some issues:

With SxS things are different:

Unfortunately, SxS is very bad documented, and there are not many helpful articles and tools available. And if something goes wrong, it may be very difficult to find out the “why”. sxstrace.exe normally is a help, but not every time. Manifest caching can also disturb.

Manifests

Manifests are the basic mechanism for SxS, they are needed for both the component and the executable. The manifest for the component needs to specify the available classes and GUIDs, and the manifest for the executable needs to specify the name of the component. At loading time, the loader first reads the manifest for the executable and then the manifest for the component. If something goes wrong, you exe will not start. Look first in the event viewer and then use sxstrace to see the errors. Because of manifest caching it can be necessary to change the time stamp of your executable (best: regenerate it). Manifests can be standalone (name of the exe or dll with added .manifest extension), or built into the executable. My recommendation would be to use standalone manifests for development and embedded manifests for deployment.

GUIDs

Every class and also the interfaces need their own GUID – generate a new one with the guidgen tool (You’ll need the Visual Studio Developer Command Prompt for this). Please pay attention to use the right GUIDs!

Versions

The entire system is very sensible to versioning! You need to use the correct version in every place: manifests and component itself (an AssemblyInfo.prg with the correct AssemblyVersionAttribute and the AssemblyFileVersionAttribute entries is needed – otherwise SxS will give no errors, but don’t load the component! This alone has costed me several days of tries.

Naming

Correct naming is the next potential issue: this is not only important for the name of the DLL and for the name of the classes, but also for the namespace used in the component. Name the component the same as the used namespace (and use a namespace!) to save you troubles (with naming errors the registered assembly may work, the SxS loading will fail silently at runtime).

Any error in one of the components will cause the failure of the entire loading mechanism.

Actually, I have a project where I have to retrieve Google AdWords statistic data and store it locally to generate reports.

Unfortunately, I lost a lot of time to find out how to start. Therefore you can find here the most important things from the view of a Windows programmer.

1) request an API token. Google ask something from you, like the use you make, if your software will be distributed, if you plan only read access or also to write. You have also to write a short project description. And if you don’t use your API token for more as 3 months, they will ask you the reason. More informations are here:
https://developers.google.com/adwords/api/docs/signingup

2) decide how you will authenticate. OAuth2 is not very simple to understand. Unfortunately, I had opted first for a service authentication. This one is very complicated and you need a Google Apps domain for it, i.e. one of your domains must be registered for a Google cloud service like GMail or Apps. So my recommendation is to use the APPLICATION authmode.

3) on the developers console https://console.developers.google.com you need then to request a new client ID – as type I opted for “OTHER”, as my project is neither a web project, nor a mobile app.

4) for the authentication you need basically 4 things: your API token, a client ID, a client secret and a refresh token. The API token you should have from step 1, client ID and client secret come from step 3, and the refresh token will be described later.

5) generate an AdWords account that has read-only access to the Adwords accounts you need – if you are a developer like me, ask the Adwords guys of your company.

6) download the client libraries here https://github.com/googleads/googleads-dotnet-lib. I would recommend to download the entire project as ZIP and build the client libraries yourself – open the Visual Studio solution file and build all. In this solution you will find a lot of useful things: not only the complete source code of the client libraries, but also samples in C# and VB.NET

7) the OAuth2 flow is not so easy to implement – I have used the OAuthTokenGenerator application from the client libraries from step 6. This application (beware: you have to start this application with elevated rights as it uses internally the httpListener class!) gives you the last pieces needed for authentication: the refresh token

8) use the samples from the client libraries to understand how the entire API works. It is very well documented, and the fact that you have the sources is very helpful. For the authentication best you use the app.config file (yourApplication.exe.config), copy it from the sample in the client library and replace the authentication data

9) last error I made: the AdWords account I used had no campaigns in it since it was the account of the company itself, I had to use one of the customer accounts and set it not in the .config file, but in my program. How this can be accomplished, is documented in the app.config file.

I have to make a compliment to Google: they are very helpful, the API is well documented and you have the full sources to it.

This is code for a Vulcan.NET or X# Listener class (ported over from C#):


// Application : HttpListener
// httpListener.prg , Created : 16.02.2016 08:37
// User : Wolfgang
// C# code see here:
// http://mikehadlow.blogspot.it/2006/07/playing-with-httpsys.html
//
// to make it run without admin rights you need something like this one:
// netsh http add urlacl url="http://*:8080/" user=[username] listen=yes

#using System
#using System.Net
#using System.Text
#using System.IO
#using VulcanHttpListener.riedmann.it

begin namespace VulcanHttpListener.riedmann.it

function Start() as void
 local oProgram as MainProgram

 oProgram := MainProgram{}
 oProgram:Start()

 return

class MainProgram
 protect _oListener as HttpListener

 constructor()

  _oListener := HttpListener{}

 return

 method Start() as void

  _oListener:Prefixes:Add( "http://*:8080/" )
  _oListener:Start()
  Console.WriteLine( "Listening on port 8080, hit enter to stop" )
  _oListener:BeginGetContext( AsyncCallback{ self:GetContextCallback }, null )
  Console.ReadLine()
  _oListener:Stop()

 return

 method GetContextCallback( oResult as IAsyncResult ) as void
  local oContext as HttpListenerContext
  local oRequest as HttpListenerRequest
  local oResponse as HttpListenerResponse
  local oSB as StringBuilder
  local oString as string
  local oBuffer as byte[]
  local oOutput as Stream

  oContext := _oListener:EndGetContext( oResult )
  oRequest := oContext:Request
  oResponse := oContext:Response

  oSB := StringBuilder{}
  oSB:Append( e"\n" )
  oSB:AppendFormat( e"HttpMethod: {0}\n", oRequest:HttpMethod )
  oSB:AppendFormat( e"URI: {0}\n", oRequest:Url:AbsoluteUri )
  oSB:AppendFormat( e"Local path: {0}\n", oRequest:Url:LocalPath )
  oSB:Append( e"\n" )
  foreach cKey as string in oRequest:QueryString:Keys
   oSB:AppendFormat( e"Query: {0} = {1}\n", cKey, oRequest:QueryString[cKey] )
  next
  oSB:Append( e"\n" )

  oString := oSB:ToString()
  oBuffer := System.Text.Encoding.UTF8:GetBytes( oString )
  oResponse:ContentLength64 := oBuffer:Length
  oOutput := oResponse:OutputStream
  oOutput:Write( oBuffer, 0, oBuffer:Length )

  _oListener:BeginGetContext( AsyncCallback{ self, @GetContextCallback() }, null )

  return

end class

end namespace


These days, I was working on a webservice that worked directly with http.sys.
This service was planned to wait for requests, and launch programs depending on the request.
Firstly, I planned to use the local system account, and launch the external programs as specific user. This program should be started from a network share, using UNC paths.
Unfortunately, that does not work, Process.Start cannot do that:
http://blogs.msdn.com/b/winsdk/archive/2013/11/12/runas-verb-process-start-doesn-t-work-from-a-localsystem-net-service.aspx

My next decision was to add a user for this purpose (to run the Windows service). Unfortunately, Process.Start ignores the essential settings for a Windows service

oInfo:UseShellExecute := false
oInfo:CreateNoWindow := true
oInfo:ErrorDialog := false
oInfo:WindowStyle := ProcessWindowStyle.Hidden

The page from MSDN (https://msdn.microsoft.com/library/0w4h05yb%28v=vs.100%29.aspx) states:
“If the UserName and Password properties of the StartInfo instance are set, the unmanaged CreateProcessWithLogonW function is called, which starts the process in a new window even if the CreateNoWindow property value is true or the WindowStyle property value is Hidden.”
At the moment I’m not changing the user from the service, but executing the external program under the service account.

I had also another small issue: since I was reading the password from an ini file, I thought the StartInfo:PasswordInClearText property was enough. On my development machine it worked, but crashed on the production server, an SBS 2011 (based on Windows Server 2008 R2) because with the .NET Framework 4 this property simply does not exists.

Another issue: if you plan to start a process as another user from a service: this user needs local logon rights (on a server OS this is a manual change)

Today, trying to understand also events in Vulcan.NET/X#, I wrote the following sample that should give an idea how events work in Vulcan.NET and X#.

#using EventSample

begin namespace EventSample

function Start( ) as void
  local oClass2 as Sample2Class

  System.Console.WriteLine(“Start of the SampleEvent application”)

  oClass2 := Sample2Class{}
  oClass2:RegisterEvent()

return

public delegate SampleHandle( cParameter as string ) as void

class SampleClass
  public event oSampleEvent as SampleHandle

constructor()

  return

method CallSampleEvent() as void

  oSampleEvent( “Hi there” )

  return

end class

class Sample2Class

constructor()

  return

method RegisterEvent() as void
  local oClass as SampleClass

  oClass := SampleClass{}
  oClass:oSampleEvent += self:method2
  oClass:CallSampleEvent()

  return

public method method2( cParameter as string ) as void

  System.Console.WriteLine( “method2:” + cParameter )

  return

end class

end namespace

Hi,

when working with X# and Vulcan.NET, I have created this particular sample. It should help understand how delegates work (and that they are immutable):

#using DelegateSample

begin namespace DelegateSample

public delegate SampleDelegate( cParameter as string ) as void

function Start( ) as void
  local oSample as SampleClass
  local oSample2 as Sample2Class
  local oDelegate as SampleDelegate
  local oDelegate2 as SampleDelegate

  System.Console.WriteLine(“Start SampleDelegate”)

  oSample := SampleClass{}
  oSample2 := Sample2Class{}

  oDelegate := SampleDelegate{ oSample:Method1 }
  oDelegate += oSample:Method2
  oDelegate += oSample2:Method3

  oDelegate( “Hi” )
  oDelegate2 := oDelegate

  oDelegate -= oSample:Method2
  oDelegate -= oSample2:Method3

  oDelegate( “Hott” )
  oDelegate2( “Hups” )

  return

class SampleClass

  constructor()

  return

  public method Method1( cParameter as string ) as void

    System.Console.WriteLine( “method1, parameter ” + cParameter )

    return

  public method Method2( cParameter as string ) as void

    System.Console.WriteLine( “method2, parameter ” + cParameter )

    return

end class

class Sample2Class

  public method Method3( cParameter as string ) as void

    System.Console.WriteLine( “method3, Sample2Class, parameter ” + cParameter )

    return

end class

end namespace

In some of my software programs, I have integrated a thing like document archiving, and there customers are putting several files, mostly PDF.

Since Adobe Reader is out and installed on the machines, often PDF files that are on a shared network folder, cannot be opened anymore – Adobe Reader gives an “access denied” error.

It is the “Protected Mode” of Adobe Reader that gives this problem.

Fortunately you can disable this mode, and Adobe Reader works again as expected. Please don’t forget to enable the protected mode afterwards!

This is my VO code for this functionality:

// disable Adobe Reader protected mode
aVersions := { “11.0”, “DC” }
aReset := {}
nLen := ALen( aVersions )
for nI := 1 upto nLen
 cVersion := aVersions[nI]
 nMode := RegistryDWord( HKEY_CURRENT_USER, “Software\Adobe\Acrobat Reader\” + cVersion + “\Privileged”, “bProtectedMode” )
 if nMode == 1
  SetRegistryDWord( HKEY_CURRENT_USER, “Software\Adobe\Acrobat Reader\” + cVersion + “\Privileged”, “bProtectedMode”, 0 )
  AAdd( aReset, cVersion )
 endif
next
oProcess := SpawnProcess{ cExe, cFileName, SW_NORMAL }
for nI := 1 upto 20
 ApplicationExec( EXECWHILEEVENT )
 Sleep( 500 ) // let Acrobat reader start
next
// re-enable Adobe Reader protected mode
nLen := ALen( aReset )
for nI := 1 upto nLen
 cVersion := aReset[nI]
 nMode := RegistryDWord( HKEY_CURRENT_USER, “Software\Adobe\Acrobat Reader\” + cVersion + “\Privileged”, “bProtectedMode” )
 if nMode == 0
  SetRegistryDWord( HKEY_CURRENT_USER, “Software\Adobe\Acrobat Reader\” + cVersion + “\Privileged”, “bProtectedMode”, 1 )
  AAdd( aReset, cVersion )
 endif
next

Trying to optimize building SQL insert and update strings, the following VO code:

cTable + ‘”‘ + cFieldName + ‘”‘

in Vulcan.NET works as follows

cTable + “.” + chr( 34 ) + cFieldName + chr( 34 )

But of course is not very performant. Better is the following code:

string.Format( e”{0}.\”{1}\””, cTable, aValues[nI,1] )

Please note the “e” character before the string – it enables escaping the double quotes with the backslash.

The entire method for building an append statement ist the following:

static method CreateInsertString( cTable as string, oFieldValues as List<BaseField> ) as string
 local oField as BaseField
 local oStatement as StringBuilder
 local oValues as StringBuilder
 local cStatement as string
 local cValues as string
 local nLen as int
 local nI as int
 local cField as string
 local uValue as usual
 local cValue as string

 cStatement := string.Format( “insert into {0} ( “, cTable )
 oStatement := StringBuilder{ cStatement, 512 }
 oStatement:Append( ” values ( ” )
 oValues := StringBuilder{ 512 }
 nLen := oFieldValues:Count – 1
  for nI := 0 upto nLen
  oField := oFieldValues:Item[nI]
  cValue := FirebirdHelper.ConvertValueToStatement( oField:Value )
  if nI < nLen
   oStatement:AppendFormat( “{0}, “, oField:Name )
   oValues:AppendFormat( “{0}, “, cValue )
  else
   oStatement:AppendFormat( “{0}) “, oField:Name )
   oValues:AppendFormat( “{0}) “, cValue )
  endif
 next
 oStatement:Append( oValues )
 cStatement := oStatement:ToString()

 return cStatement