CMake C++ Custom Library on Windows “Undefined Reference” – No Error on Linux

Here is the short version with a quick setup of the situation and the fix. After that, I’ll elaborate a bit.



I have a custom C++ library and a separate project for tests (all based on Qt 6). The test project requires the library for execution.

Here is a short excerpt of the CMake scripts, first the library, then the tests.

project(wt2-shared VERSION 2.0.0 DESCRIPTION "WorkTracker2 Shared 

# To export symbols.

# Snip header + source definitions

add_library(${PROJECT_NAME} ${SOURCES} ${HEADERS})

target_include_directories(${PROJECT_NAME} PUBLIC include/)
target_link_libraries(wt2-shared Qt6::Core Qt6::Sql)
project(wt2-shared-test VERSION 2.0.0 DESCRIPTION "WorkTracker2 
Shared Library Tests")

# Snip header + source definitions

add_executable(${PROJECT_NAME} ${SOURCES} ${HEADERS})

target_include_directories(${PROJECT_NAME} PRIVATE ${INCLUDES})
target_link_libraries(${PROJECT_NAME} Qt6::Core Qt6::Test wt2-shared)


This error only occurred on Windows, and it does not matter which toolchain I used, be it MinGW or MSVC. The result was always the same.

The following shows the MinGW error.

[100%] Linking CXX executable wt2-shared-test.exe    
    undefined reference to `__imp__ZN4Data3Sql13SqlDataSourceC1E7QString'
    undefined reference to `__imp__ZN4Data3Sql13SqlDataSource4loadEv'
    undefined reference to `__imp__ZN4Data3Sql13SqlDataSourceC1E7QString'	
    undefined reference to `__imp__ZN4Data3Sql13SqlDataSource4loadEv'
collect2.exe: error: ld returned 1 exit status
mingw32-make[2]: *** 
wt2-shared-test/wt2-shared-test.exe] Error 1


The add_library definition in the CMakeLists.txt was incomplete.
To make it work, I added SHARED because I want a shared library.


Continue reading, though, to get the full picture. There is more to it than just making the library a shared one.

Read More »

CMake CLI Parameter “generator-name” Usage

This topic shouldn’t even require a blog post, but I find the CMake CLI usage rather odd when it comes to specifying a generator. Here’s a shortened "-h" output.

> cmake -h

cmake [options] <path-to-source>
cmake [options] <path-to-existing-build>
cmake [options] -S <path-to-source> -B <path-to-build>

Specify a source directory to (re-)generate a build system for it in the
current working directory.  Specify an existing build directory to
re-generate its build system.

-S <path-to-source>          = Explicitly specify a source directory.
-B <path-to-build>           = Explicitly specify a build directory.
-C <initial-cache>           = Pre-load a script to populate the cache.
-D <var>[:<type>]=<value>    = Create or update a cmake cache entry.
-U <globbing_expr>           = Remove matching entries from CMake cache.
-G <generator-name>          = Specify a build system generator.
-T <toolset-name>            = Specify toolset name if supported by
-A <platform-name>           = Specify platform name if supported by



The following generators are available on this platform (* marks default):
Visual Studio 16 2019        = Generates Visual Studio 2019 project files.
                                Use -A option to specify architecture.
Visual Studio 15 2017 [arch] = Generates Visual Studio 2017 project files.
                                Optional [arch] can be "Win64" or "ARM".


Borland Makefiles            = Generates Borland makefiles.
* NMake Makefiles              = Generates NMake makefiles.
NMake Makefiles JOM          = Generates JOM makefiles.
MSYS Makefiles               = Generates MSYS makefiles.
MinGW Makefiles              = Generates a make file for use with
Unix Makefiles               = Generates standard UNIX makefiles.

Kate - Ninja                 = Generates Kate project files.
Kate - Unix Makefiles        = Generates Kate project files.
Eclipse CDT4 - NMake Makefiles
                            = Generates Eclipse CDT 4.0 project files.
Eclipse CDT4 - MinGW Makefiles
                            = Generates Eclipse CDT 4.0 project files.
Eclipse CDT4 - Ninja         = Generates Eclipse CDT 4.0 project files.
Eclipse CDT4 - Unix Makefiles= Generates Eclipse CDT 4.0 project files.

These generator options do not look like valid parameter values to the "-G" option. But they are. So, if you want to compile on Windows using MinGW, you have to use this.

> cmake -S ../src -B ./ -G "MinGW Makefiles"

Or, if you prefer Visual Studio project files:

> cmake -S ../ -B ./ -G "Visual Studio 15 2017 Win64"

This "syntax" looks weird, and it tripped me for about 10 minutes until I found a sample and understood how it works.

Terraform Azure Error: parsing json result from the Azure CLI: Error waiting for the Azure CLI: exit status 1; Failed to load token files

There are some instances where I have managed to screw up my Azure CLI configuration file with Terraform. It must have something to do with parallel usage of Terraform or Terraform simultaneously with the az tool. Either way, I ran into the following error.

$ terraform refresh
Acquiring state lock. This may take a few moments...

Error: Error building account: Error getting authenticated object ID: Error parsing json result from the Azure CLI: Error aiting for the Azure CLI: exit status 1

  on line 16, in provider "azurerm":
  16: provider "azurerm" {

I wondered: "What might block the Azure access? Am I maybe not logged in?" So, I went ahead and tried to log in.

$ az login
Failed to load token files. If you have a repro, please log an issue
at At the same time, you 
can clean up by running 'az account clear' and then 'az login'. 

(Inner Error: Failed to parse /home/rlo/.azure/accessTokens.json with exception: Extra data: line 1 column 18614 (char 18613))

The error probably comes from parallel access to my Azure CLI configuration file. When I opened the /home/rlo/.azure/accessTokens.json, I found some dangling garbage at the end of it that broke the JSON format.

Here’s a snippet of the last few lines.

        "refreshToken": "0.A...",
        "oid": "<oid>",
        "userId": "<userId>",
        "isMRRT": true,
        "_clientId": "<clientId>",
        "_authority": "<uid>"

I took out the trash bc1"}], saved the file, and it worked again. Many access to resources. Such joy 😉

Qt5 QtCreator Error on Linux: stddef.h: No such file or directory – Code model could not parse an included file

The following is an error that has shown itself every time I have installed the Qt5 framework and the QtCreator development environment on a Linux based machine. It never mattered which flavor of Linux; QtCreator always showed this error.

Warning: The code model could not parse an included file, which might lead to incorrect code completion and highlighting, for example. 

fatal error: 'stddef.h' file not found 
note: in file included from /home/rlo/Code/C++/WorkTracker2/WorkTracker2Shared/src/data/taskrepository.h:1: 
note: in file included from /home/rlo/Code/C++/WorkTracker2/WorkTracker2Shared/src/data/taskrepository.h:3: 
note: in file included from /usr/include/c++/9/optional:38: 
note: in file included from /usr/include/c++/9/stdexcept:38: 
note: in file included from /usr/include/c++/9/exception:143: 
note: in file included from /usr/include/c++/9/bits/exception_ptr.h:38: 

Although that message never caused any issues compiling the code, I found it rather annoying, and at some point, annoying enough to search for a solution.

As it turns out, this message appears when you have Clang libraries installed. QtCreator detects that and automatically uses Clang to parse the source code and provide inline error messages and code completion.

You can get rid of this error when you explicitly add the STL header files’ include-path to your project. In my case, I have added the following to my *.pro file.

unix {
    INCLUDEPATH += /usr/lib/gcc/x86_64-linux-gnu/9/include

Azure PostgreSQL Error: PSQLException The connection attempt failed

A few days ago at work, I was investigating a strange issue where one of our services could not connect to the Azure Managed PostgreSQL Database from the Kubernetes cluster. Oddly enough, other services of that cluster did not exhibit this behavior.

org.postgresql.util.PSQLException: The connection attempt failed.
        at org.postgresql.core.v3.ConnectionFactoryImpl.openConnectionImpl( ~[postgresql-42.2.16.jar!/:42.2.16]
        at org.postgresql.core.ConnectionFactory.openConnection( ~[postgresql-42.2.16.jar!/:42.2.16]
        at org.postgresql.jdbc.PgConnection.<init>( ~[postgresql-42.2.16.jar!/:42.2.16]
        at org.postgresql.Driver.makeConnection( ~[postgresql-42.2.16.jar!/:42.2.16]
        at org.postgresql.Driver.connect( ~[postgresql-42.2.16.jar!/:42.2.16]
        at ~[ehg-hermes.jar:0.13.0-SNAPSHOT]
        at org.springframework.boot.loader.Launcher.launch( ~[ehg-hermes.jar:0.13.0-SNAPSHOT]
        at org.springframework.boot.loader.Launcher.launch( ~[ehg-hermes.jar:0.13.0-SNAPSHOT]
        at org.springframework.boot.loader.JarLauncher.main( ~[ehg-hermes.jar:0.13.0-SNAPSHOT]
Caused by: null
        at org.postgresql.core.PGStream.receiveChar( ~[postgresql-42.2.16.jar!/:42.2.16]
        at org.postgresql.core.v3.ConnectionFactoryImpl.enableGSSEncrypted( ~[postgresql-42.2.16.jar!/:42.2.16]
        at org.postgresql.core.v3.ConnectionFactoryImpl.tryConnect( ~[postgresql-42.2.16.jar!/:42.2.16]
        at org.postgresql.core.v3.ConnectionFactoryImpl.openConnectionImpl( ~[postgresql-42.2.16.jar!/:42.2.16]
        ... 46 common frames omitted

As it turns out, it was an issue with the PSQL JDBC driver version that comes bundled with Spring Boot version 2.3.4-RELEASE. All the other services were still built with a slightly older release and therefore used an older PSQL JDBC driver.

The key indicator of what is going on is this method call.


A bit of research led me to a question on StackOverflow that pointed me in the right direction, and ultimately I ended up on Microsoft’s Azure documentation. If you scroll down, you will find a section named "GSS error".

The solution to this problem is simple. If you do not want or cannot change the Spring Boot or PSQL JDBC driver version, e.g., because of automated CVE scans that break builds (the reason we upgraded this one service), then you can solve it with a configuration change. Append gssEncMode=disable to the JDBC connection string.

Example: jdbc:postgresql://

CMake on Windows 10 with Qt5 Error: CMAKE_PREFIX_PATH

Quick one: Set the proper CMAKE_PREFIX_PATH value for Qt5 development on Windows 10 with MinGW and CMake.

Here is the lovely error you get from CMake.

CMake Error at CMakeLists.txt:13 (find_package):
 By not providing "FindQt5.cmake" in CMAKE_MODULE_PATH this project 
 asked CMake to find a package configuration file provided by "Qt5", 
 CMake did not find one.

 Could not find a package configuration file provided by "Qt5" with 
any of
 the following names:


 Add the installation prefix of "Qt5" to CMAKE_PREFIX_PATH or set 
 to a directory containing one of the above files. If "Qt5" provides a
 separate development package or SDK, be sure it has been installed.

CMake asks you to tell it where to find the Qt5 *.cmake configuration files. One option is to set the CMAKE_PREFIX_PATH variable. The Qt5 documentation has the following to say about that:

The easiest way to use CMake is to set the CMAKE_PREFIX_PATH environment variable to the install prefix of Qt 5.

Now, what the hell is the "install prefix"? It turns out it is the path where you can find the compiler-specific Qt binaries, include files and the other stuff. In my case, using MinGW, it is located at "C:\Apps\Qt\5.14.2\mingw73_64". If you are using the MSVC binaries, select the corresponding directory, probably something like "C:\Apps\Qt\5.14.2\msvc_2017".

Maven Failsafe Plugin environmentVariables Not Working (org.postgresql.Driver claims to not accept jdbcUrl)

A few months ago I had written about how one can setup a PostgreSQL database in a docker container during the Maven testing phase (part 1 and part 2). Today, I wanted to iterate on this topic using Testcontainers. Unfortunately, before I could get to that I ran into issues with the original project. For some reason I was now getting the following error:

Caused by: java.lang.RuntimeException: Driver org.postgresql.Driver claims to not accept jdbcUrl, jdbc:postgresql://localhost:${it-database.port}/docker_db_testing_tests

Maven did not replace the variable it-database in the integration test file. The question is: "why now"? It has worked before. Now, one thing that I changed is that this time I was using Linux instead of Windows. Either way, the fix was simple, although not obvious. It seems to be an issue with the Maven Failsafe plugin.

Not working:




Here is the link to the commit.

Spring Multipart File – Can I Read InputStream Multiple Times?

The short answer is Yes.

Here is the long version and why I even asked myself this question.

If you are familiar with ServletRequest then you probably know that calling its getInputStream method only works once. If you need to read the body data multiple times then it is up to you to cache it in a buffer or employ workarounds such as a "caching servlet request". Unfortunately, this fact is not stated in the Javadoc of ServletRequest#getInputStream so it is no wonder this question gets asked.

Spring’s MultipartFile is a bit different here. It, too, has a getInputStream method, but this one can be called multiple times. Again, it is not obvious from the documentation which is why I am making this mental note for myself and others who are researching this question because they know about the behavior of ServletRequest and assume – as I did – it is the same for MultipartFile. Fortunately, it is not.

In my case I needed to compute a hash of an uploaded file and then move the file to Azure’s Blob Storage. The Azure API used an InputStream and I assumed, once I had consumed the multipart InputStream that I could not use that API anymore.

As a side note: Using DigestInputStream it is possible to do this in one go, move the data to storage and while doing that compute the hash. In my case, I needed the hash first to compare it with a value that was provided on upload. Only when they match can the data be transferred to storage.

Automate Game Screenshot Capture: Windows API SendInput Function with C++

I am trying to write video game reviews after I have finished a game and I like to add some impressions in the form of screenshots to the reviews. There is one problem though: sometimes it is impossible to press the keyboard shortcut to capture a screenshot because the game requires my full attention – and all my fingers. Therefore, I miss out on a lot of action sequences. What does a programmer do in such a situation? Write a tool that scratches the itch.

Read More »

Mockito “when” vs. “verify”

Recently, during a code review at work there was a discussion about whether Mockito’s verify is necessary when when (or given, which is a synonym) is parameterized. The quick answer is "no".

Imagine this contrived example. I have two classes, Calculator and ComplexOperationService, where the latter requires the former. The goal is to write a unit test for ComplexOperationService and mock all usages of Calculator.

Here are the two classes.

public class Calculator {

    public int sum(final int a, final int b) {
        return a + b;

public class ComplexOperationService {
    private final Calculator calculator;

    public void doComplexOperation(final int a, final int b) {
        System.out.println(calculator.sum(a, b));
Read More »

Integration Testing with Docker Maven Plugin, PostgreSQL, Flyway (Update)

Recently I have written about how one can create a PostgreSQL Docker image with the Docker Maven Plugin to run integration tests that require a database. While this worked all nice and well during development, the concept has one flaw: the plugin will push the database Docker image to a Docker registry during the deploy phase. I do not want this temporary image to end up there. This is the behavior of the Docker Maven Plugin and I have not found a way to work around this. By “work around this” I mean somehow configure the plugin to ignore this custom PSQL image during the “deploy” phase. Unfortunately, there is only a global <filter> that applies to all the phases of the plugin.

There is a proper solution however, at least for what I was using the database.

Read More »

Windows Subsystem for Linux (WSL) Docker: error storing credentials – err: exit status 1, out: Cannot autolaunch D-Bus without X11 $DISPLAY

Recently at work, when copying an application from our internal Docker Registry to Azure, I ran into the following error in my WSL Ubuntu installation.

Login at
Error saving credentials: error storing credentials - err: exit status 1, out: `Cannot autolaunch D-Bus without X11 $DISPLAY`
ERROR: source registry login failed

The easiest fix I found was to install the gnupg2 and pass packages.

sudo apt install gnupg2 pass 

One important thing to note regarding security: the output mentioned storing the credentials in plain text as a result somewhere in the WSL user’s /home directory. If you are very conscious about where passwords are stored, do not use this solution or remove the password file afterwards. That’s good enough for me at the moment, I just needed to get this to work somehow.

Integration Testing With Docker Maven Plugin, PostgreSQL, Flyway

Some things in software development require more than mocks and unit testing. If your application uses a database it makes sense to also hit that database in automated testing to ensure custom SQL queries work correctly, Hibernate relations are set up properly and also that database migrations are successful.

This blog post was written with a focus on the latter. I will be using Spring Boot talking to a PostgreSQL database. The database structure is managed via Flyway and, basically customary for Java applications, Maven serves as the build and dependency management tool. Docker will also play a role because we’ll be creating and running a PostgreSQL docker image for testing. From Maven. Every time the test is executed. And to spice things up, we’ll also create a custom database and user in that dockerized PSQL image.

I have created a working sample on Github and you can follow every single step by taking a look at the commit history. There you can see individual changes, starting from an empty Spring Boot application with no database to the final solution with Spring Data JPA and Flyway.

In the following sections and snippets, I will highlight the important parts of each step.

Read More »

Comparing Java Optional vs C++ STL optional

Optionals in Java have been around for some time now, basically since the release of version 8. My other language of choice, C++, has received this feature in version C++17. Since I am currently in the process of writing some C++ code, I was curious how they were implemented there. Optionals are trying to solve a problem that is likely to plague any language. What shall a method or function return if there is no value? Or shall it not return anything but instead start crying like a petulant child and throw an exception?

As an introduction, let me dive a little bit into why we need optionals (or do we?) and compare two different implementations of this concept, one being java.util.Optional and the other C++ std::optional. I chose to compare these two language for several reasons:

  1. I work with Java in my day job, so I have a good idea of how it works there.
  2. As mentioned, C++ is one of the languages I know quite well too.
  3. The main reason: both optional implementations are add-on classes rather than language features. More on that later.
Read More »