Does PostSharp add code during precompilation of assembly? - postsharp

When assembly using PostSharp [Log] aspect was decompiled, the decompiled code appeared without new code added to it. Doesn't PostSharp add code corresponding to aspect during precompilation?

PostSharp post-processes the compiler output by reading and disassembling the intermediate assembly, execute the required transformations and validations, and rewriting the final assembly to disk.
PostSharp integrates itself in the build process via PostSharp.targets. NuGet adds the PostSharp.targets import to your project when you install PostSharp nuget package.
Let's consider this method:
[Log]
public void Method() { }
If everything works correctly then the decompiled C# code of this method starts like this (PostSharp 5.0):
public void Method()
{
bool flag = <>z__Program.DefaultCategory.IsEnabled(LogLevel.Debug);
LogMemberInfo logMemberInfo;
If you don't see this code, it may indicate that PostSharp is not correctly installed to your project or it is not correctly configured.
Adding Detailed Logging to your Solution article explains how to correctly configure and add PostSharp to your project.

Related

Getting error log while previewing report from Jasper [duplicate]

I am getting a NoClassDefFoundError when I run my Java application. What is typically the cause of this?
While it's possible that this is due to a classpath mismatch between compile-time and run-time, it's not necessarily true.
It is important to keep two or three different exceptions straight in our head in this case:
java.lang.ClassNotFoundException This exception indicates that the class was not found on the classpath. This indicates that we were trying to load the class definition, and the class did not exist on the classpath.
java.lang.NoClassDefFoundError This exception indicates that the JVM looked in its internal class definition data structure for the definition of a class and did not find it. This is different than saying that it could not be loaded from the classpath. Usually this indicates that we previously attempted to load a class from the classpath, but it failed for some reason - now we're trying to use the class again (and thus need to load it, since it failed last time), but we're not even going to try to load it, because we failed loading it earlier (and reasonably suspect that we would fail again). The earlier failure could be a ClassNotFoundException or an ExceptionInInitializerError (indicating a failure in the static initialization block) or any number of other problems. The point is, a NoClassDefFoundError is not necessarily a classpath problem.
This is caused when there is a class file that your code depends on and it is present at compile time but not found at runtime. Look for differences in your build time and runtime classpaths.
Here is the code to illustrate java.lang.NoClassDefFoundError. Please see Jared's answer for detailed explanation.
NoClassDefFoundErrorDemo.java
public class NoClassDefFoundErrorDemo {
public static void main(String[] args) {
try {
// The following line would throw ExceptionInInitializerError
SimpleCalculator calculator1 = new SimpleCalculator();
} catch (Throwable t) {
System.out.println(t);
}
// The following line would cause NoClassDefFoundError
SimpleCalculator calculator2 = new SimpleCalculator();
}
}
SimpleCalculator.java
public class SimpleCalculator {
static int undefined = 1 / 0;
}
NoClassDefFoundError In Java
Definition:
Java Virtual Machine is not able to find a particular class at runtime which was available at compile time.
If a class was present during compile time but not available in java classpath during runtime.
Examples:
The class is not in Classpath, there is no sure shot way of knowing it but many times you can just have a look to print System.getproperty("java.classpath") and it will print the classpath from there you can at least get an idea of your actual runtime classpath.
A simple example of NoClassDefFoundError is class belongs to a missing JAR file or JAR was not added into classpath or sometimes jar's name has been changed by someone like in my case one of my colleagues has changed tibco.jar into tibco_v3.jar and the program is failing with java.lang.NoClassDefFoundError and I were wondering what's wrong.
Just try to run with explicitly -classpath option with the classpath you think will work and if it's working then it's a sure short sign that someone is overriding java classpath.
Permission issue on JAR file can also cause NoClassDefFoundError in Java.
Typo on XML Configuration can also cause NoClassDefFoundError in Java.
when your compiled class which is defined in a package, doesn’t present in the same package while loading like in the case of JApplet it will throw NoClassDefFoundError in Java.
Possible Solutions:
The class is not available in Java Classpath.
If you are working in J2EE environment than the visibility of Class among multiple Classloader can also cause java.lang.NoClassDefFoundError, see examples and scenario section for detailed discussion.
Check for java.lang.ExceptionInInitializerError in your log file. NoClassDefFoundError due to the failure of static initialization is quite common.
Because NoClassDefFoundError is a subclass of java.lang.LinkageError it can also come if one of it dependency like native library may not available.
Any start-up script is overriding Classpath environment variable.
You might be running your program using jar command and class was not defined in manifest file's ClassPath attribute.
Resources:
3 ways to solve NoClassDefFoundError
java.lang.NoClassDefFoundError Problem patterns
I have found that sometimes I get a NoClassDefFound error when code is compiled with an incompatible version of the class found at runtime. The specific instance I recall is with the apache axis library. There were actually 2 versions on my runtime classpath and it was picking up the out of date and incompatible version and not the correct one, causing a NoClassDefFound error. This was in a command line app where I was using a command similar to this.
set classpath=%classpath%;axis.jar
I was able to get it to pick up the proper version by using:
set classpath=axis.jar;%classpath%;
One interesting case in which you might see a lot of NoClassDefFoundErrors is when you:
throw a RuntimeException in the static block of your class Example
Intercept it (or if it just doesn't matter like it is thrown in a test case)
Try to create an instance of this class Example
static class Example {
static {
thisThrowsRuntimeException();
}
}
static class OuterClazz {
OuterClazz() {
try {
new Example();
} catch (Throwable ignored) { //simulating catching RuntimeException from static block
// DO NOT DO THIS IN PRODUCTION CODE, THIS IS JUST AN EXAMPLE in StackOverflow
}
new Example(); //this throws NoClassDefFoundError
}
}
NoClassDefError will be thrown accompanied with ExceptionInInitializerError from the static block RuntimeException.
This is especially important case when you see NoClassDefFoundErrors in your UNIT TESTS.
In a way you're "sharing" the static block execution between tests, but the initial ExceptionInInitializerError will be just in one test case. The first one that uses the problematic Example class. Other test cases that use the Example class will just throw NoClassDefFoundErrors.
This is the best solution I found so far.
Suppose we have a package called org.mypackage containing the classes:
HelloWorld (main class)
SupportClass
UtilClass
and the files defining this package are stored physically under the directory D:\myprogram (on Windows) or /home/user/myprogram (on Linux).
The file structure will look like this:
When we invoke Java, we specify the name of the application to run: org.mypackage.HelloWorld. However we must also tell Java where to look for the files and directories defining our package. So to launch the program, we have to use the following command:
I was using Spring Framework with Maven and solved this error in my project.
There was a runtime error in the class. I was reading a property as integer, but when it read the value from the property file, its value was double.
Spring did not give me a full stack trace of on which line the runtime failed.
It simply said NoClassDefFoundError. But when I executed it as a native Java application (taking it out of MVC), it gave ExceptionInInitializerError which was the true cause and which is how I traced the error.
#xli's answer gave me insight into what may be wrong in my code.
I get NoClassFoundError when classes loaded by the runtime class loader cannot access classes already loaded by the java rootloader. Because the different class loaders are in different security domains (according to java) the jvm won't allow classes already loaded by the rootloader to be resolved in the runtime loader address space.
Run your program with 'java -javaagent:tracer.jar [YOUR java ARGS]'
It produces output showing the loaded class, and the loader env that loaded the class. It's very helpful tracing why a class cannot be resolved.
// ClassLoaderTracer.java
// From: https://blogs.oracle.com/sundararajan/entry/tracing_class_loading_1_5
import java.lang.instrument.*;
import java.security.*;
// manifest.mf
// Premain-Class: ClassLoadTracer
// jar -cvfm tracer.jar manifest.mf ClassLoaderTracer.class
// java -javaagent:tracer.jar [...]
public class ClassLoadTracer
{
public static void premain(String agentArgs, Instrumentation inst)
{
final java.io.PrintStream out = System.out;
inst.addTransformer(new ClassFileTransformer() {
public byte[] transform(ClassLoader loader, String className, Class classBeingRedefined, ProtectionDomain protectionDomain, byte[] classfileBuffer) throws IllegalClassFormatException {
String pd = (null == protectionDomain) ? "null" : protectionDomain.getCodeSource().toString();
out.println(className + " loaded by " + loader + " at " + new java.util.Date() + " in " + pd);
// dump stack trace of the thread loading class
Thread.dumpStack();
// we just want the original .class bytes to be loaded!
// we are not instrumenting it...
return null;
}
});
}
}
The technique below helped me many times:
System.out.println(TheNoDefFoundClass.class.getProtectionDomain().getCodeSource().getLocation());
where the TheNoDefFoundClass is the class that might be "lost" due to a preference for an older version of the same library used by your program. This most frequently happens with the cases, when the client software is being deployed into a dominant container, armed with its own classloaders and tons of ancient versions of most popular libs.
Java ClassNotFoundException vs NoClassDefFoundError
[ClassLoader]
Static vs Dynamic class loading
Static(Implicit) class loading - result of reference, instantiation, or inheritance.
MyClass myClass = new MyClass();
Dynamic(Explicit) class loading is result of Class.forName(), loadClass(), findSystemClass()
MyClass myClass = (MyClass) Class.forName("MyClass").newInstance();
Every class has a ClassLoader which uses loadClass(String name); that is why
explicit class loader uses implicit class loader
NoClassDefFoundError is a part of explicit class loader. It is Error to guarantee that during compilation this class was presented but now (in run time) it is absent.
ClassNotFoundException is a part of implicit class loader. It is Exception to be elastic with scenarios where additionally it can be used - for example reflection.
In case you have generated-code (EMF, etc.) there can be too many static initialisers which consume all stack space.
See Stack Overflow question How to increase the Java stack size?.
Two different checkout copies of the same project
In my case, the problem was Eclipse's inability to differentiate between two different copies of the same project. I have one locked on trunk (SVN version control) and the other one working in one branch at a time. I tried out one change in the working copy as a JUnit test case, which included extracting a private inner class to be a public class on its own and while it was working, I open the other copy of the project to look around at some other part of the code that needed changes. At some point, the NoClassDefFoundError popped up complaining that the private inner class was not there; double-clicking in the stack trace brought me to the source file in the wrong project copy.
Closing the trunk copy of the project and running the test case again got rid of the problem.
I fixed my problem by disabling the preDexLibraries for all modules:
dexOptions {
preDexLibraries false
...
I got this error when I add Maven dependency of another module to my project, the issue was finally solved by add -Xss2m to my program's JVM option(It's one megabyte by default since JDK5.0). It's believed the program does not have enough stack to load class.
In my case I was getting this error due to a mismatch in the JDK versions. When I tried to run the application from Intelij it wasn't working but then running it from the command line worked. This is because Intelij was attempting to run it with the Java 11 JDK that was setup but on the command line it was running with the Java 8 JDK. After switching that setting under File > Project Structure > Project Settings > Project SDK, it worked for me.
Update [https://www.infoq.com/articles/single-file-execution-java11/]:
In Java SE 11, you get the option to launch a single source code file
directly, without intermediate compilation. Just for your convenience,
so that newbies like you don't have to run javac + java (of course,
leaving them confused why that is).
NoClassDefFoundError can also occur when a static initializer tries to load a resource bundle that is not available in runtime, for example a properties file that the affected class tries to load from the META-INF directory, but isn’t there. If you don’t catch NoClassDefFoundError, sometimes you won’t be able to see the full stack trace; to overcome this you can temporarily use a catch clause for Throwable:
try {
// Statement(s) that cause(s) the affected class to be loaded
} catch (Throwable t) {
Logger.getLogger("<logger-name>").info("Loading my class went wrong", t);
}
I was getting NoClassDefFoundError while trying to deploy application on Tomcat/JBOSS servers. I played with different dependencies to resolve the issue, but kept getting the same error. Marked all javax.* dependencies as provided in pom.xml, And war literally had no Dependency in it. Still the issue kept popping up.
Finally realized that src/main/webapps/WEB-INF/classes had classes folder which was getting copied into my war, so instead of compiled classes, this classes were getting copied, hence no dependency change was resolving the issue.
Hence be careful if any previously compiled data is getting copied, After deleting classes folder and fresh compilation, It worked!..
If someone comes here because of java.lang.NoClassDefFoundError: org/apache/log4j/Logger error, in my case it was produced because I used log4j 2 (but I didn't add all the files that come with it), and some dependency library used log4j 1. The solution was to add the Log4j 1.x bridge: the jar log4j-1.2-api-<version>.jar which comes with log4j 2. More info in the log4j 2 migration.
This error can be caused by unchecked Java version requirements.
In my case I was able to resolve this error, while building a high-profile open-source project, by switching from Java 9 to Java 8 using SDKMAN!.
sdk list java
sdk install java 8u152-zulu
sdk use java 8u152-zulu
Then doing a clean install as described below.
When using Maven as your build tool, it is sometimes helpful -- and usually gratifying, to do a clean 'install' build with testing disabled.
mvn clean install -DskipTests
Now that everything has been built and installed, you can go ahead and run the tests.
mvn test
I got NoClassDefFound errors when I didn't export a class on the "Order and Export" tab in the Java Build Path of my project. Make sure to put a checkmark in the "Order and Export" tab of any dependencies you add to the project's build path. See Eclipse warning: XXXXXXXXXXX.jar will not be exported or published. Runtime ClassNotFoundExceptions may result.
It could also be because you copy the code file from an IDE with a certain package name and you want to try to run it using terminal. You will have to remove the package name from the code first.
This happens to me.
Everyone talks here about some Java configuration stuff, JVM problems etc., in my case the error was not related to these topics at all and had a very trivial and easy to solve reason: I had a wrong annotation at my endpoint in my Controller (Spring Boot application).
I have had an interesting issue wiht NoClassDefFoundError in JavaEE working with Liberty server. I was using IMS resource adapters and my server.xml had already resource adapter for imsudbJXA.rar.
When I added new adapter for imsudbXA.rar, I would start getting this error for instance objects for DLIException, IMSConnectionSpec or SQLInteractionSpec.
I could not figure why but I resolved it by creating new server.xml for my work using only imsudbXA.rar. I am sure using multiple resource adapters in server.xml is fine, I just had no time to look into that.
I had this error but could not figure out the solution based on this thread but solved it myself.
For my problem I was compiling this code:
package valentines;
import java.math.BigInteger;
import java.util.ArrayList;
public class StudentSolver {
public static ArrayList<Boolean> solve(ArrayList<ArrayList<BigInteger>> problems) {
//DOING WORK HERE
}
public static void main(String[] args){
//TESTING SOLVE FUNCTION
}
}
I was then compiling this code in a folder structure that was like /ProjectName/valentines
Compiling it worked fine but trying to execute: java StudentSolver
I was getting the NoClassDefError.
To fix this I simply removed: package valentines;
I'm not very well versed in java packages and such but this how I fixed my error so sorry if this was already answered by someone else but I couldn't interpret it to my problem.
My solution to this was to "avail" the classpath contents for the specific classes that were missing. In my case, I had 2 dependencies, and though I was able to compile successfully using javac ..., I was not able to run the resulting class file using java ..., because a Dynamic class in the BouncyCastle jar could not be loaded at runtime.
javac --classpath "ext/commons-io-2.11.0;ext/bc-fips-1.0.2.3" hello.java
So at compile time and by runtime, the JVM is aware of where to fetch Apache Commons and BouncyCastle dependencies, however, when running this, I got
Error: Unable to initialize main class hello
Caused by: java.lang.NoClassDefFoundError:
org/bouncycastle/jcajce/provider/BouncyCastleFipsProvider
And I therefore manually created a new folder named ext at the same location, as per the classpath, where I then placed the BouncyCastle jar to ensure it would be found at runtime. You can place the jar relative to the class file or the jar file as long as the resulting manifest has the location of the jar specified. Note I only need to avail the one jar containing the missing class file.
Java was unable to find the class A in runtime.
Class A was in maven project ArtClient from a different workspace.
So I imported ArtClient to my Eclipse project.
Two of my projects was using ArtClient as dependency.
I changed library reference to project reference for these ones (Build Path -> Configure Build Path).
And the problem gone away.
I had the same problem, and I was stock for many hours.
I found the solution. In my case, there was the static method defined due to that. The JVM can not create the another object of that class.
For example,
private static HttpHost proxy = new HttpHost(proxyHost, Integer.valueOf(proxyPort), "http");
I got this message after removing two files from the SRC library, and when I brought them back I kept seeing this error message.
My solution was: Restart Eclipse. Since then I haven't seen this message again :-)

dagger android support to androidx.fragment

How to inject a fragment from the package androidx.fragment.app.Fragment ?
I'm using the dagger-android framework to inject my dependencies in my code.
As the documentation says I do this to inject my fragment
#Override
public void onAttach(Activity activity) {
AndroidInjection.inject(this);
super.onAttach(activity);
// ...
}
the problem is that AndroidSupportInjection class accept only fragments of the package android.support.v4.app.Fragment or if I use AndroidInjection class only accept fragments of the package android.app.Fragment and I want to use fragments of the androidx.fragment.app.Fragment package.
Also DaggerFrament extend from android.support.v4.app.Fragment and want to use a fragment from androidx
And If I try to implement HasSupportFragmentInjector also this interface use a fragment from android.support
add the below code to your gradle.properties
android.useAndroidX=true
android.enableJetifier=true
And if you are trying to inject into a Fragment you have to replace AndroidInjection.inject(this) with AndroidSupportInjection.inject(this)
I had the same problem in case of HasFragmentInjector. You need to use HasSupportFragmentInjector for fragment injection. This is because, HasFragmentInjector uses android.app.Fragment which is not effected by jetifier. You need to add android-dagger-support library, jetifier converts all the support packages to androidx in Studio 3.3 (if jetifier is enabled).
If jetifier does not change support packages to androidx packages. You can download jetifier tool from here and convert the android-dagger-support.aar file manually by using the following command.
./jetifier-standalone -i dagger-android-support-<version>.aar -o <output-name>
Then add the library to your project. This is the HasSupportFragment class after conversion
import androidx.fragment.app.Fragment;
import dagger.android.AndroidInjector;
public interface HasSupportFragmentInjector {
AndroidInjector<Fragment> supportFragmentInjector();
}
Somehow, jetifier tool was not converting libraries in AndroidStudio.
I had to do it manually.
I had a similar error and it was due to the Dagger version. On version 2.17 there is an strange issue, but if you roll back to version 2.16 it compiles perfectly (apart from the flags on gradle.properties that Paul posted).
From there using the tutorials you won't have trouble. Forgot to mention that on my project I had the non-androidX version of everything, then I ran the androidX migration that android studio offers, and after that I had to switch the Dagger version, but I suppose that if you do it from the start it's the same.
Hope this helps, if you switch and it doesn't work, post a little bit of your dagger implementation and plugins versions and I will try to help more!
Add the following to your gradle.properties file
android.useAndroidX = true
android.enableJetifier = true
Just for reference. i had the same problem. It was Jetifier issue. please upgrade your gradle build tools plugin to 3.3.0
classpath 'com.android.tools.build:gradle:3.3.0'
Sample code: https://github.com/jega-ms/android-dagger2-mvp-rx
This is what I did to work with androidx namespace for Dagger 2.21
Downloaded the jetifier tool from here: https://developer.android.com/studio/command-line/jetifier
Unzip into a folder and open a terminal pointing into the extracted bin folder
From Android Studio, open a class like DaggerFragment to check the path where the file is stored, for me (in MacOS) is something like this:
From terminal execute this command (replacing with the correct variables and path)
./jetifier-standalone -i /Users/{YOUR_USER}/.gradle/caches/{PATH_TO_DAGGER_ANDROID_SUPPORT_FOLDER}/dagger-android-support-2.21.aar -o dagger-android-support-2.21.aar
The converted dagger-android-support-2.21.aar will appear in your bin folder
Now open your app build.gradle file and change this line
implementation fileTree(dir: 'libs', include: ['*.jar', '*.aar'])
adding the , '*.aar' part in the include array
Move the generated dagger-android-support-2.21.aar from bin to libs folder inside your project.
Remove (or comment) this line from the dependencies
implementation "com.google.dagger:dagger-android-support:2.21
Now you can proceed invalidating the cache and rebuild the project and now DaggerFragment will point to your converted version which uses androidx.fragment.app.Fragment
NOTE: Obviously this is a temporary workaround and you should move to the official version as soon this is fixed in Dagger
The solution to my particular problem was to use android dagger classes as interfaces instead of extend of them:
class MyFragment() : HasSupportFragmentInjector {
#Inject
lateinit var childFragmentInjector: DispatchingAndroidInjector<Fragment>
override fun onAttach(context: Context?) {
AndroidSupportInjection.inject(this)
super.onAttach(context)
}
override fun supportFragmentInjector(): AndroidInjector<Fragment> {
return childFragmentInjector
}
........
}
To my Activities
class MyActivity : HasSupportFragmentInjector {
#Inject
internal lateinit var fragmentInjector: DispatchingAndroidInjector<Fragment>
override fun onCreate(savedInstanceState: Bundle?) {
AndroidInjection.inject(this)
super.onCreate(savedInstanceState)
}
override fun supportFragmentInjector(): AndroidInjector<Fragment> = fragmentInjector
......
}
and also I have this in my gradle.properties file:
android.useAndroidX = true
android.enableJetifier = true

Nunitlite-runner.exe - Dll file has no TestFixtures

I'm trying run tests from nunitlite-runner.exe but i got Error message
1) Invalid : /myfile.dll
Has no TestFixtures
I have in source code
[TestFixture]
public class MyClassName
{
[Order(1)]
[Test]
public void Case1()
{
//
}
[Order(2)]
[Test]
public void Case2()
{
//
}
}
How can i fix code or run my tests?
I got this problem with nunit3-console.exe. I had, in my enthusiasm to upgrade to the latest and greatest, updated my NUnit project references to the latest version (3.9 at this writing), along with my console exe (3.7). But it seems that the console exe is a build or two behind the core libraries, so you can't reference a later version of NUnit than the version you have of the console exe.
I downgraded my NUnit project reference to 3.7, and that fixed it for me.

How do I load Nuget packages that a custom DLL depends on?

I'm developing a Function App via the Portal (not local development). I have a custom DLL that depends on 1 nuget package: Entity Framework 6.1.3
I have uploaded my DLL to "../bin" and my code compiles successfully when I reference my DbContext object. So far, so good.
I also have a Project.json file and I see it acquiring the nuget packages when I save. So far, so good.
{
"frameworks": {
"net46":{
"dependencies": {
"EntityFramework": "6.1.3"
}
}
}
}
My Run.csx code compiles successfully and looks like this:
#r "../bin/Library.dll"
using System;
public static void Run(TimerInfo myTimer, TraceWriter log)
{
log.Info("My code: Started");
log.Info(typeof(Library.MyContext).ToString());
}
However, the code doesn't actually run (I don't even see the "My code: Started" log item). The error I receive is:
2017-02-27T06:37:28.731 Exception while executing function:
Functions.TimerTriggerCSharp1. mscorlib: Exception has been thrown by
the target of an invocation. f-TimerTriggerCSharp1__-205940111: Could
not load file or assembly 'EntityFramework, Version=6.0.0.0,
Culture=neutral, PublicKeyToken=b77a5c561934e089' or one of its
dependencies. The system cannot find the file specified.
My custom DLL is the simplest possible EF-referencing DLL I can possibly make. All you need to recreate it is this:
Custom DLL packages.config:
<?xml version="1.0" encoding="utf-8"?>
<packages>
<package id="EntityFramework" version="6.1.3" targetFramework="net46" />
</packages>
Custom DLL Class1.cs
using System.Data.Entity;
namespace Library
{
public class MyContext : DbContext
{
}
}
What am I doing wrong that is keeping my custom DLL from being able to utilize the downloaded EntityFramework nuget package?
Just to verify that my nuget references are actually working, if I comment out most of my Run.csx code and replace it with this line, all executes correctly and logs what you would expect:
log.Info($"My code: {typeof(System.Data.Entity.DbContext).ToString()}");
As somebody suggested, I have tried changing my Run.csx references to look like this and it doesn't change the runtime error I get (it does compile so the path is correct):
#r "../../../data/Functions/packages/nuget/entityframework/6.1.3/lib/net45/EntityFramework.dll"
#r "../bin/My.dll"
I can also change my Run.csx file to contain this code and it does successfully execute:
using System;
public static void Run(TimerInfo myTimer, TraceWriter log)
{
log.Info(typeof(MyContext).ToString());
}
public class MyContext : System.Data.Entity.DbContext
{
}
In order to consume the assemblies coming from your referenced packages, you can deploy your custom dependency as a private assembly:
Deploy the assembly into a bin folder, in your function folder (e.g. wwwroot\myfunction\bin)
Reference the assembly without the relative path, by file name only (e.g. Library.dll)
If you wish to use shared assemblies, deployed to a common location and referenced as you have above, you'd need to deploy the assembly with its dependencies (essentially the output from your project build).
Another option that I would recommend if you want to take advantage of the shared model is to deploy your dependency as a NuGet package (which you can deploy to either a custom source hosted somewhere or as a file), that package would then specify its package dependencies and they would all be resolved automatically.
Looks like configuration problem, you need to define the reference in the project file to be copied to output.
In the Solution Explorer, expand the project, References, right click the EntityFramework, Properties, Set Copy Local = true.
Sometimes the default value is false, so then it will not in the output folder of the project.
I hope this helps.

How do I reference a UWP+NET46 portable library from a .NET 4.6 console application?

I have a portable class library project that targets .NET 4.6 and Universal Windows Platform. This class library contains just one class with the following line of code in its constructor:
Directory.CreateDirectory(Path.Combine(Path.GetTempPath(), Guid.NewGuid().ToString()));
Now I create a new .NET 4.6 console application project in the same solution and add a project reference to the portable class library. Calling the method that houses the above line of code results in the following exception at runtime:
Could not load file or assembly 'System.IO.FileSystem, Version=4.0.0.0, Culture=neutral, PublicKeyToken=b03f5f7f11d50a3a' or one of its dependencies. The system cannot find the file specified.
What am I doing wrong here? There are no compile-time errors or warnings.
Things I have tried: add missing(?) NuGet package manually
It seems that System.IO.FileSystem is a library delivered via NuGet, as part of the Microsoft.NETCore mega-package. Okay, perhaps I need to explicitly add this package to any project that uses my portable class library. I attempt to do so.
Could not install package 'Microsoft.NETCore.Platforms 1.0.0'. You are trying to install this package into a project that targets '.NETFramework,Version=v4.6', but the package does not contain any assembly references or content files that are compatible with that framework. For more information, contact the package author.
No luck with this approach.
Things I have tried: create a project.json file
While there is no clear info on the web, I read a few tidbits about a new project.json based NuGet harness or build system. Just to experiment, I created the following project.json file in my console application project:
{
"dependencies": {
},
"frameworks": {
"net46": { }
},
"runtimes": {
"win-anycpu": { }
}
}
It works! The runtime error goes away! However, I soon found that this was either not the right solution or not a complete solution. I started writing some code to read configuration section values, which involved making use of the IConfigurationSectionHandler interface, and got the following compile-time error:
error CS0246: The type or namespace name 'IConfigurationSectionHandler' could not be found (are you missing a using directive or an assembly reference?)
This interface is part of the System assembly. I see a reference to this assembly, but it has a yellow exclamation mark icon, and a warning appears in the warnings window:
The referenced component 'System' could not be found.
This is where I ran out of ideas. Am I missing something totally obvious?
I have found the solution. My initial attempt was to install the Microsoft.NETCore package into the console application, resulting in the error shown in my original post.
However, if I install only the narrowly-scoped packages, e.g. System.IO.FileSystem, then I achieve success and the application works correctly. Apparently there is something special about the Microsoft.NETCore "master package" that prevents it from correctly installing into dependent projects.