Registration-Free COM for standalone VB.NET application - complex case - vb.net

I need your help.
I read whole internet about Registration-Free COM/DLLs but my problem is more complex.
I'm preparing an application in VB.NET which will be used in an environment in which users don't have admin rights, so I can't simpy install it or register COM. This COM is a LogParser library designed by microsoft.
DLL also doesn't have to be embeded - would be nice, but it may be also extracted from exe during startup - i'm ok with this approach
Generally in a main form i've got a button which invokes another form by:
LogParser_Form.Show()
This another Form 'Imports MSUtil', which is a Interop.MSUtil.dll and which is embeeded to exe by Fody Costura add-on.
Form contains also a class which has multiple declarations of variables defined in COM, eg:
Dim IISW3CLOG As New COMIISW3CInputContextClass
(there is more than one)
But this dll refers somewhere to bigger: LogParser.dll which is acutally a COM component which requires registration, so my LogParser_Form doesn't appear when button is clicked, but it throws an exception that COM component is not found...
Unfortunately Fody Costura or Ilmerge don't work for the COM...
I tried multiple tricks wich manifest files, etc, but no luck...
You are my last hope - please help me... How to embed this COM to exe without registering it?
I suppose that properly used manifest files may help, but I didn't find a way to successfully use it ...

Getting Registration-Free COM to work can be tricky, but works when configured properly. The key issue is creating manifests, which document all required dependencies. In your case, you'll need two manifests:
Client manifest for your application
Server manifest for the LogParser library. This part requires a tool for analyzing type libraries, such as the OLE/COM Object Viewer (oleview.exe). It allows looking into the embedded type library inside LogParser.dll.
Let's take the (slightly modified) C# example, which is documented in the LogParser help file. The client is named "logqryclient.exe" in this case, and the Runtime Callable Wrapper has been created via the type library importer (tlbimp).
using System;
using Interop.MSUtil;
namespace logqryclient
{
class Program
{
static void Main(string[] args)
{
try
{
// Instantiate the LogQuery object
ILogQuery oLogQuery = new LogQueryClassClass();
// Create the query
string query = #"SELECT TOP 50 SourceName, EventID, Message FROM System";
// Execute the query
ILogRecordset oRecordSet = oLogQuery.Execute(query, null);
// Browse the recordset
for (; !oRecordSet.atEnd(); oRecordSet.moveNext())
{
ILogRecord rec = oRecordSet.getRecord();
Console.WriteLine(rec.toNativeString(","));
}
// Close the recordset
oRecordSet.close();
}
catch (System.Runtime.InteropServices.COMException exc)
{
Console.WriteLine("Unexpected error: " + exc.Message);
}
}
}
}
To use this code without registering the COM classes, you'll first need to place the LogParser.dll into the same directory as the client executable.
Next, you'll need to create an accompanying server manifest (named "LogParser.manifest" here). This documents all necessary classes and marshalling information for the interfaces (required for thread switching). As mentioned earlier, you'll need a type library analyzer to gain access to the class and interface identifiers.
In the above case, you'll need identifiers for:
ILogQuery interface & LogQueryClass class
ILogRecordset interface
ILogRecord interface
Hence, the server manifest could look as follows:
<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<assembly xmlns="urn:schemas-microsoft-com:asm.v1" manifestVersion="1.0">
<assemblyIdentity type="win32" name="LogParser" version="1.0.0.0" />
<file name = "LogParser.dll">
<!-- LogQueryClass -->
<comClass
clsid="{8CFEBA94-3FC2-45CA-B9A5-9EDACF704F66}"
threadingModel = "Apartment" />
<!-- Embedded type library -->
<typelib
tlbid="{A7E75D86-41CD-4B6E-B4BD-CC2ED34B3FB0}"
version="1.0"
helpdir=""/>
</file>
<!-- Marshalling information for interfaces -->
<comInterfaceExternalProxyStub
name="ILogQuery"
iid="{3BDE06BC-89E4-42FD-BE64-832A5F33D7D3}"
proxyStubClsid32="{00020424-0000-0000-C000-000000000046}"
baseInterface="{00000000-0000-0000-C000-000000000046}"
tlbid = "{A7E75D86-41CD-4B6E-B4BD-CC2ED34B3FB0}" />
<comInterfaceExternalProxyStub
name="ILogRecordset"
iid="{C9452B1B-093C-4842-ABD1-F81410926874}"
proxyStubClsid32="{00020424-0000-0000-C000-000000000046}"
baseInterface="{00000000-0000-0000-C000-000000000046}"
tlbid = "{A7E75D86-41CD-4B6E-B4BD-CC2ED34B3FB0}" />
<comInterfaceExternalProxyStub
name="ILogRecord"
iid="{185FFF88-E24A-4984-9621-AA41BEAE8513}"
proxyStubClsid32="{00020424-0000-0000-c000-000000000046}"
baseInterface="{00000000-0000-0000-c000-000000000046}"
tlbid = "{A7E75D86-41CD-4B6E-B4BD-CC2ED34B3FB0}" />
</assembly>
To allow the client to find the server manifest and ultimately the LogParser library, embed the following client manifest into the "logqryclient.exe" client:
<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<assembly xmlns="urn:schemas-microsoft-com:asm.v1" manifestVersion="1.0">
<assemblyIdentity
type = "win32"
name = "logqryclient"
version = "1.0.0.0" />
<dependency>
<dependentAssembly>
<assemblyIdentity
type="win32"
name="LogParser"
version="1.0.0.0" />
</dependentAssembly>
</dependency>
</assembly>
Now, all required information is located in the manifests, so that you can run the code in registration-free configuration.

Related

RTI DDS creating own data types

I am working on a .Net example where I define my own data type using RTI Connext DDS.
Instead of creating the application from the beginning, I got help from the source code of the hello_world_xml_dynamic example in rti_workspace directory. I have made several changes to the USER_QOS_PROFILES.xml file to create my own data type and changes its name to MY_PROFILES.xml
But when I compile the application and run it from the command line, I get the following error:
DDS_DomainParticipantFactory_create_participant_from_config_w_paramsI:ERROR: Profile library 'MyParticipantLibrary::PublicationParticipant' not found
! Unable to create DDS domain participant
The line of code that catching the error:
if (this.participant == null)
{
this.participant = DDS.DomainParticipantFactory.get_instance().
create_participant_from_config(
"MyParticipantLibrary::PublicationParticipant");
if (this.participant == null)
{
Console.Error.WriteLine("! Unable to create DDS domain participant");
return;
}
}
this is the configuration file MY_PROFILES.xml :
<!--
RTI Data Distribution Service Deployment
-->
<dds xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:noNamespaceSchemaLocation="http://community.rti.com/schema/6.0.1/rti_dds_profiles.xsd">
<!-- Qos Library -->
<qos_library name="qosLibrary">
<qos_profile name="DefaultProfile">
</qos_profile>
</qos_library>
<!-- types -->
<types>
<struct name="FlightData">
<member name="Latitude" type="double"/>
<member name="Longitude" type="double"/>
<member name="Altitude" type="double"/>
</struct>
</types>
<!-- Domain Library -->
<domain_library name="MyDomainLibrary" >
<domain name="FlightDataDomain" domain_id="0">
<register_type name="FlightDataType"
type_ref="FlightData" />
<topic name="FlightDataTopic"
register_type_ref="FlightDataType">
<topic_qos name="FlightData_qos"
base_name="qosLibrary::DefaultProfile"/>
</topic>
</domain>
</domain_library>
<!-- Participant library -->
<domain_participant_library name="MyParticipantLibrary">
<domain_participant name="PublicationParticipant"
domain_ref="MyDomainLibrary::FlightDataDomain">
<publisher name="MyPublisher">
<data_writer name="FlightDataWriter"
topic_ref="FlightDataTopic"/>
</publisher>
</domain_participant>
<domain_participant name="SubscriptionParticipant"
domain_ref="MyDomainLibrary::FlightDataDomain">
<subscriber name="MySubscriber">
<data_reader name="FlightDataReader"
topic_ref="FlightDataTopic">
<datareader_qos name="FlightData_reader_qos"
base_name="qosLibrary::DefaultProfile"/>
</data_reader>
</subscriber>
</domain_participant>
</domain_participant_library>
</dds>
where am i making a mistake?
Your XML file looks correct. From the 'not found' error message, it seems that you may not have taken the right steps to instruct your application to load that profiles-file MY_PROFILES.xml to actually learn about your desired Participant. You can easily verify that this is the case by introducing an error in your XML file (for example by incorrectly renaming one tag) and rerun your application. If it does not complain about the syntax or schema of the XML, then your file did not get loaded and this hypothesis is correct.
If that turns out to be your problem indeed, then you have several options to fix that. They are listed in the User's Manual section 18.5 How to Load XML-Specified QoS Settings.

Add Friend Classes to Exclusion List of Obfuscation

I'm using Dotfuscator CE with Visual Studio 2015 Update 3 to obfuscate my .Net assemblies.
We know that Public types and members are not be obfuscated by default.
I'm curious to know how can we add Friend Classes in Exclusion list so that those should not be obfuscated?
Here is the config file file I'm using to obfuscate my DLL.
<?xml version="1.0" encoding="utf-8" standalone="no"?>
<!DOCTYPE dotfuscator SYSTEM "http://www.preemptive.com/dotfuscator/dtd/dotfuscator_v2.3.dtd">
<dotfuscator version="2.3">
<propertylist>
<property name="SourceDirectory" value="This Path Will Be Replaced By Visual Studio" />
<property name="SourceFile" value="This Filename Will Be Replaced By Visual Studio" />
</propertylist>
<global>
<option>quiet</option>
</global>
<input>
<asmlist>
<inputassembly refid="e4ca1ab5-26cb-4ab7-9621-87063f75a38f">
<option>honoroas</option>
<option>stripoa</option>
<option>library</option>
<option>transformxaml</option>
<file dir="${SourceDirectory}" name="${SourceFile}" />
</inputassembly>
</asmlist>
</input>
<output>
<file dir="${SourceDirectory}" />
</output>
<renaming>
<option>xmlserialization</option>
<mapping>
<mapoutput overwrite="true">
<file dir="${SourceDirectory}\Dotfuscated" name="Map.xml" />
</mapoutput>
</mapping>
<referencerulelist>
<referencerule rulekey="{6655B10A-FD58-462d-8D4F-5B1316DFF0FF}" />
<referencerule rulekey="{7D9C8B02-2383-420f-8740-A9760394C2C1}" />
<referencerule rulekey="{229FD6F8-5BCC-427b-8F72-A7A413ECDF1A}" />
<referencerule rulekey="{2B7E7C8C-A39A-4db8-9DFC-6AFD38509061}" />
<referencerule rulekey="{494EA3BA-B947-44B5-BEE8-A11CC85AAF9B}" />
<referencerule rulekey="{89769974-93E9-4e71-8D92-BE70E855ACFC}" />
<referencerule rulekey="{4D81E604-A545-4631-8B6D-C3735F793F80}" />
</referencerulelist>
</renaming>
<sos mergeruntime="true">
<option>version:v4</option>
<option>sendanalytics</option>
<option>dontsendtamper</option>
</sos>
<smartobfuscation>
<smartobfuscationreport verbosity="all" overwrite="false" />
</smartobfuscation>
</dotfuscator>
Actually I've a Model class with Friend access specifier. I post its object via PostAsJsonAsync method e.g.
Dim result As HttpResponseMessage = client.PostAsJsonAsync(APIEndPoints.LOGIN, _LoginModel).Result
Here is the Friend Class:
Friend Class LoginModel
Public AccessKey As String
Public Password As String
End Class
API method that receives the request and model:
[HttpPost]
[Route("authenticate")]
public async Task<JsonResult> Authenticate([FromBody] LoginViewModel lvm)
// Here lvm.Accesskey is null
When API receives the request and LoginModel too, its fields are null. If I make my LoginModel public then it works.
Note: this only happens when I obfuscate my DLL, otherwise the implementation works with Friend class too.
Also note: Friend classes are common in VB.Net. They works like public classes when accessed within an assembly but they are private outside the assembly.
Based on your clarification, it sounds like you want to exclude not only the names of Friend types, but also the names of Public fields within those types. I had interpreted your original question as wanting to exclude anything marked Friend, no matter the context.
An important point here is that, in terms of Dotfuscator's rules, excluding a type does not automatically exclude its members.
Here's an exclusion rule set that excludes top-level Friend types and Public and Friend fields of those types:
<excludelist>
<type name=".*" regex="true" speclist="+notpublic">
<comment>Exclude top-level types that are only accessible to the assembly ("Friend" in VB, "internal" in C#, or "private" in IL).</comment>
<field name=".*" speclist="+public" regex="true">
<comment>Exclude public fields of types the parent rule matches</comment>
</field>
</type>
</excludelist>
You can also just exclude types and members you know will cause trouble when renamed, rather than excluding a large number of names using rules based on accessibility. Here's an example, assuming LoginModel is defined in assembly YourAssembly and namespace YourNamespace.Here:
<excludelist>
<type name="YourAssembly.YourNamespace.Here.LoginModel">
<field name="AccessKey" signature="string" />
<field name="Password" signature="string" />
</type>
</excludelist>
(I noticed you're using this same configuration for multiple input assemblies, but this rule is still safe because if the input assembly doesn't contain the specified type, then the rule will be ignored.)
For reference, the Professional Edition documentation on Exclusion Rules (and sub-topics of that page) might be useful - Community Edition and Professional Edition share the same configuration file format, for features that are supported by both editions.
Disclosure: I work on the Dotfuscator team for PreEmptive Solutions.
If you are trying to exclude your input assembly's Friend types and members because your assembly has a Friend Assembly, then be aware that Dotfuscator will automatically exclude such code elements from renaming (the only kind of obfuscation provided by Dotfuscator CE) and will issue the following warning:
WARNING: NameOfYourInputAsssembly has non-input Friend Assemblies and is in Library Mode; internal members will not be renamed or pruned. Consider adding Friend Assemblies as input for increased obfuscation.
(The term "internal" here is the C# equivalent of VB's "Friend" keyword).
As the warning suggests, you can instead include the Friend Assembly as another Input to Dotfuscator.
If you do so, Dotfuscator can then rename the Friend types and members, and update the Friend Assembly to refer to those types and members by the new names.
If you still would like to exclude Friend types and members, you can do so with the following renaming exclusion rule set, added as a child of the <renaming> tag in the config file:
<excludelist>
<type name=".*" regex="true" speclist="+notpublic">
<comment>Exclude types that are only accessible to the assembly ("Friend" in VB, "internal" in C#, or "private" in IL).</comment>
</type>
<type name=".*" regex="true" speclist="+nestedassembly">
<comment>Exclude nested types that are only accessible to the assembly ("Friend" in VB, "internal" in C#, or "private" in IL).</comment>
</type>
<type name=".*" regex="true" excludetype="false">
<comment>Select, but do not exclude, all types.</comment>
<method name=".*" speclist="+assembly" regex="true">
<comment>Exclude methods that are only accessible to the assembly ("Friend" in VB, "internal" in C#, or "assembly" in IL).</comment>
</method>
<field name=".*" speclist="+assembly" regex="true">
<comment>Exclude fields that are only accessible to the assembly ("Friend" in VB, "internal" in C#, or "assembly" in IL).</comment>
</field>
<propertymember name=".*" speclist="+assembly" regex="true">
<comment>Exclude properties that are only accessible to the assembly ("Friend" in VB, "internal" in C#, or "assembly" in IL).</comment>
</propertymember>
<eventmember name=".*" speclist="+assembly" regex="true">
<comment>Exclude events that are only accessible to the assembly ("Friend" in VB, "internal" in C#, or "assembly" in IL).</comment>
</eventmember>
</type>
</excludelist>
Edit: I had missed nested types in the previous revision of this answer.
Disclosure: I work on the Dotfuscator team for PreEmptive Solutions.

How to check if a MSBuild-Task fails if using ContinueOnError=true

I am running the MSBuild task with ContinueOnError=true:
<MSBuild Projects="#(ComponentToDeploy)"
Targets="$(DeploymentTargets)"
Properties="$(CommonProperties);%(AdditionalProperties)"
ContinueOnError="true"
Condition="%(Condition)"/>
So my build always succeeds.
Is there a way to find out if any error occurs?
I could not find any Output of the MSBuild task containing this information.
The only way I know is to parse the log file for errors but it looks like a workaround for me.
(I am using MSBuild 4.0)
This is an answer to the last feedback of #Ilya.
I'm using feedback/answer because of the length and formatting restrictions of the comments.
Log is scoped to individual targets or to be more specific tasks...
This was indeed the first question arose when I was reading your comment with the suggestion to use Log.HasLoggedErrors: "Was is the scope of the Log?".
Unfortunately I was not be able to finde a proper documentation. MSND does not help much...
Why did you know it is scoped to the task?
I'm not in doubt about your statement at all! I'm just wondering if there is a proper documentation somewhere..
(I haven't been using MSBuild for years ;-)
In any case, what are you building as project?
My test projects are very simple.
MyTest.project
<?xml version="1.0" encoding="utf-8" ?>
<Project xmlns="http://schemas.microsoft.com/developer/msbuild/2003" DefaultTargets="ElenasTarget" ToolsVersion="4.0">
<UsingTask AssemblyFile="$(MSBuildProjectDirectory)\MyCompany.Tools.MSBuild.Tasks.dll" TaskName="MSBuildWithHasLoggedErrors" />
<ItemGroup>
<MyProjects Include="CopyNotExistingFile.proj" />
</ItemGroup>
<Target Name="ElenasTarget">
<MSBuildWithHasLoggedErrors Projects="#(MyProjects)" ContinueOnError="true" >
<Output TaskParameter="HasLoggedErrors" PropertyName="BuildFailed" />
</MSBuildWithHasLoggedErrors>
<Message Text="BuildFailed=$(BuildFailed)" />
</Target>
</Project>
The CopyNotExistingFile.proj just tries to copy a file that does not exist:
<?xml version="1.0" encoding="utf-8" ?>
<Project xmlns="http://schemas.microsoft.com/developer/msbuild/2003" DefaultTargets="Target1" ToolsVersion="4.0">
<Target Name="Target1">
<Copy SourceFiles="C:\lalala.bum" DestinationFiles="C:\tralala.bam" />
</Target>
</Project>
And this is my custom task MSBuildWithHasLoggedErrors
namespace MyCompany.Tools.MSBuild.Tasks
{
public class MSBuildWithHasLoggedErrors : Microsoft.Build.Tasks.MSBuild
{
[Output]
public bool HasLoggedErrors { get; private set; }
public override bool Execute()
{
try
{
base.Execute();
HasLoggedErrors = Log.HasLoggedErrors;
}
catch (Exception e)
{
Log.LogErrorFromException(e, true);
return false;
}
return true;
}
}
}
If I build my MyTest.proj the HasLoggedErrorswill be set to false although an error (MSB3021) was logged(?) to the console logger:
Project "C:\Users\elena\mytest.proj" on node 1 (default targets).
Project "C:\Users\elena\mytest.proj" (1) is building "C:\Users\elena\CopyNotExistingFile.proj" (2) on node 1 (default targets).
Target1:
Copying file from "C:\lalala.bum" to "C:\tralala.bam".
C:\Users\elena\CopyNotExistingFile.proj(5,4): error MSB3021: Unable to copy file "C:\lalala.bum" to "C:\tralala.bam". Could not find file 'C:\lalala.bum'.
Done Building Project "C:\Users\elena\CopyNotExistingFile.proj" (default targets) -- FAILED.
ElenasTarget:
BuildFailed=False
Done Building Project "C:\Users\elena\mytest.proj" (default targets).
Build succeeded.
My expectation was HasLoggedErrors would be set to true.
one way is to build self but with different target, for example your DefaultTargets one launches your custom MSBuildWrapper task pointing to itself (ie $(MSBuildProjectFile)) but with a different target that does other builds, copies
I've already tried it (that were my investigations I meant in my post). Unfortunately it doesn't work either :-(
(I am aware you said in theory).
My new single project looks like this:
<?xml version="1.0" encoding="utf-8" ?>
<Project xmlns="http://schemas.microsoft.com/developer/msbuild/2003" DefaultTargets="ElenasTarget" ToolsVersion="4.0">
<UsingTask AssemblyFile="$(MSBuildProjectDirectory)\MyCompany.Tools.MSBuild.Tasks.dll" TaskName="MSBuildWithHasLoggedErrors" />
<Target Name="ElenasTarget">
<MSBuildWithHasLoggedErrors Projects="$(MSBuildProjectFile)" Targets="CopyNotExistingFile" ContinueOnError="true" >
<Output TaskParameter="HasLoggedErrors" PropertyName="BuildFailed" />
</MSBuildWithHasLoggedErrors>
<Message Text="BuildFailed=$(BuildFailed)" />
</Target>
<Target Name="CopyNotExistingFile" >
<Copy SourceFiles="C:\lalala.bum" DestinationFiles="C:\tralala.bam" />
</Target>
</Project>
If I build this project HasLoggedErrors will still be set to false.
(Furthermore, my "real" build I'm currently maintaining is much complexer containing several project files with targets... so I can't pack them all in a single project file ).
or writing custom logger and passing it through command line
That was my last hope!
My "real" build has a custom logger passed through the command line (I didn't use it for my test project for the sake of simplicity). That is actually producing the log (a XML file) I'm going to parse to find out if any errors have been logged.
BTW, I thought the console logger is a kind of "global" logger. Am I wrong?
Anyway, the custom logger does not help neither, the Log.HasLoggedErrors is still set to false.
Is there some way I am not aware of to reference a particular logger (e.g. my custom logger) to ask if it has logged any errors?
It really looks like Log is scoped to individual targets.
Hmm... if the reflection on the buildengine instance is the last resort I would still prefer parsing the log.
(Don't blame me! :-) )
My decision
After some investigations I've decided to stick with my initial solution: parse the log to find out if the build failed.
Check my comments to see why I prefer that to the suggestions have been provided so far.
If someone has some other ideas do not hesitate to share :-)
(Otherwise this question can be closed, I suppose...)
The MSBuildLastTaskResult reserved property will be set to True if the last task succeeded and False if the last task failed:
<MSBuild Projects="#(ComponentToDeploy)"
Targets="$(DeploymentTargets)"
Properties="$(CommonProperties);%(AdditionalProperties)"
ContinueOnError="true"
Condition="%(Condition)" />
<Message Text="MSBuild failed!" Condition="'$(MSBuildLastTaskResult)' == 'False'" />
I believe this was introduced with MSBuild v4.0.
I know this thread is a bit old, but another possible solution, as I presume you needed to know that build failed in order to execute some "final task", is to use:
<OnError ExecuteTargets="FinalReportTarget;CleanupTarget" />
That would fail the build in case of error, but execute the "FinalReportTarget" and "CleanupTarget".
ContinueOnError="true" is not needed in this case.
You could capture TargetOutputs and check them for error conditions afterwards, but that's still quite hackish.
If you only want to check if MSBuild task failed, use Exec task. Set IgnoreExitCode to true and check ExitCode output value. If not zero, something is wrong.
If you need the list of build errors, use /fileloggerparameters command line switch to log errors only to some specific file:
/flp1:logfile=errors.txt;errorsonly
But if another task inside some target (e.g. Copytask) raised an error the Log.HasLoggedErrors returns false.
Didn't know comments have length limits...
Log is scoped to individual targets or to be more specific tasks, and (as far as I'm aware) there is no way to get a "global" one, may be through reflection on the buildengine instance, or writing custom logger and passing it through command line. In any case, what are you building as project? HasLoggedErrors works as expected (and has been working unchanged for years), it shows if project being built logged any errors. It doesn't, and shouldn't, have any control over logging of other tasks (that might use other types of loggers). If you want a global one, one way is to build self but with different target, for example your DefaultTargets one launches your custom MSBuildWrapper task pointing to itself (ie $(MSBuildProjectFile)) but with a different target that does other builds, copies, etc, in theory it should simulate a global HasLoggedErrors...

Problem with struts 2 and json plugin

I'm using the json plugin that comes with struts 2 (json-lib-2.1.jar) and trying to follow the website to set it up.
Here's my struts.xml
<struts>
<package name="example" extends="json-default">
<action name="AjaxRetrieveUser" class="actions.view.RetrieveUser">
<result type="json"/>
</action>
</package>
</struts>
but I get this warning:
SEVERE: Unable to find parent packages json-default
Is there something else I'm supposed to do?
Edit:
I added this method to my RetrieveUser:
public Map<String,Object> getJsonModel()
{
return jsonModel;
}
And my struts.xml looks like this:
<struts>
<package name="example" extends="json-default">
<action name="AjaxRetrieveUser" class="actions.view.RetrieveUser">
<result type="json"/>
<param name="root">jsonModel</param>
</action>
</package>
</struts>
However, I don't think the response is going from the RetrieveUser class to the javascript. I'm using firebug and no request gets sent.
I believe that net.sf.json-lib is just a toolset you can use in your Java to build up JSON-ready objects, suitable to be returned by actions such as you describe.
Probably, you need to include struts-json-plugin - make sure its version matches your struts version.
I notice also that as written, your action will attempt to return RetrieveUser, serialized. Most implementations I've done/seen specify the root object to be returned, by adding
<param name="root">jsonUser</param>
Under the tag, and define this method in RetrieveUser
public Map<String, Object> getJsonUser()
[This is mentioned in the Sruts2 doc]. Hope that helps.
[edit] I use Map - you could also use the object structures provided by json-lib instead.
Re: Your edit. Probably need to see your calling javascript. And probably I will suggest that you make sure you have both a success and an error handler. Can you debug/log to show that the method is being called in java ? Do your logs show anything ? This is usually some sort of error....

Simulating the Maven2 filter mechanism using Ant

I have a properties file, let say my-file.properties.
In addition to that, I have several configuration files for my application where some information must be filled regarding the content of my-file.properties file.
my-file.properties:
application.version=1.0
application.build=42
user.name=foo
user.password=bar
Thus, in my configuration files, I will find some ${application.version}, ${user.name} that will be replaced by their value taken in the properties file...
When I build my application using Maven2, I only need to specify the properties file and say that my resources files are filtered (as in this answer to another problem). However, I need to achieve the same thing by using only Ant.
I've seen that Ant offers a filter task. However, it forces me to use the pattern #property.key# (i.e. #user.name# instead of #{user.name}) in my configuration files, which is not acceptable in my case.
How can I solve my problem?
I think expandproperties is what you are looking for. This acts just like Maven2's resource filters.
INPUT
For instance, if you have src directory (one of many files):
<link href="${css.files.remote}/css1.css"/>
src/test.txt
PROCESS
And in my ANT build file we have this:
<project default="default">
<!-- The remote location of any CSS files -->
<property name="css.files.remote" value="/css/theCSSFiles" />
...
<target name="ExpandPropertiesTest">
<mkdir dir="./filtered"/>
<copy todir="./filtered">
<filterchain>
<expandproperties/>
</filterchain>
<fileset dir="./src" />
</copy>
</target>
</project>
build.xml
OUTPUT
*When you run the ExpandPropertiesTest target you will have the following in your filtered directory: *
<link href="/css/theCSSFiles/css1.css"/>
filtered/test.txt
You can define a custom FilterReader. So you have a couple of choices:
Extend/copy the org.apache.tools.ant.filters.ReplaceTokens class and define a Map property that references another properties file containing all the replacements. This is still a bit of a chore as you have to define all the replacements.
Extend/copy the org.apache.tools.ant.filters.ReplaceTokens class with additional processing that just substitutes the matched token with a version with the correct garnish. Of course you'd have to be really careful where you use this type as it will match anything with the begin and end token.
So in the read() method of ReplaceTokens, replace:
final String replaceWith = (String) hash.get(key.toString());
with a call to a getReplacement() method:
...
final String replaceWith = getReplacement(key.toString);
...
private String getReplacement(String key) {
//first check if we have a replacement defined
if(has.containsKey(key)) {
return (String)hash.get(key);
}
//now use our built in rule, use a StringBuilder if you want to be tidy
return "$" + key + "}";
}
To use this, you'd ensure your class is packaged and on Ant's path and modify your filter:
<filterreader classname="my.custom.filters.ReplaceTokens">
<!-- Define the begin and end tokens -->
<param type="tokenchar" name="begintoken" value="$"/>
<param type="tokenchar" name="endtoken" value="}"/>
<!--Can still define explicit tokens, any not
defined explicitly will be replaced by the generic rule -->
</filterreader>
One hooooooorible way to make this work, inspired by the solution of Mnementh, is with the following code:
<!-- Read the property file -->
<property file="my-file.properties"/>
<copy todir="${dist-files}" overwrite="true">
<fileset dir="${src-files}">
<include name="*.properties"/>
</fileset>
<filterchain>
<filterreader classname="org.apache.tools.ant.filters.ReplaceTokens">
<!-- Define the begin and end tokens -->
<param type="tokenchar" name="begintoken" value="$"/>
<param type="tokenchar" name="endtoken" value="}"/>
<!-- Define one token per entry in the my-file.properties. Arggh -->
<param type="token" name="{application.version" value="${application.version}"/>
<param type="token" name="{user.name" value="${user.name}"/>
...
</filterreader>
</filterchain>
</copy>
Explanations:
I am using the ReplaceTokens reader to look for all $...} pattern. I cannot search for ${...} patterns, as the begintoken is a char and not a String. Then, I set the list of tokens starting with a { (i.e. I see {user.name instead of user.name). Hopefully, I have "only" about 20 lines in my-file.properties, so I need to define "only" 20 tokens in my Ant file...
Is there any simple and stupid solution to solve this simple and stupid problem??
Ant knows a concept named Filterchains, that is useful here. Use the ReplaceTokens-filter and specify the begintoken and endtoken as empty (normally that's '#'). That should do the trick.