Let's start using C++17

.github/workflows/build_cpp.yml

@@ -105,7 +105,7 @@ jobs:
         sudo apt install gcc-11 g++-11
         sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-11 110 --slave /usr/bin/g++ g++ /usr/bin/g++-11 --slave /usr/bin/gcov gcov /usr/bin/gcov-11
     - name: Configure
-      run: cmake ${{ github.workspace }}/dlib/test -B build
+      run: cmake ${{ github.workspace }}/dlib/test -B build -DCMAKE_CXX_STANDARD=17 -DCMAKE_CXX_EXTENSIONS=OFF
     - name: Build just tests
       run: cmake --build build --config Release --target dtest --parallel 4
     - name: Test

README.md

@@ -6,7 +6,7 @@ Dlib is a modern C++ toolkit containing machine learning algorithms and tools fo
 
 ## Compiling dlib C++ example programs
 
-Go into the examples folder and type:
+You need a C++17 compiler and CMake. Go into the examples folder and type:
 
 ```bash
 mkdir build; cd build; cmake .. ; cmake --build .
@@ -21,9 +21,9 @@ mkdir build; cd build; cmake .. -DUSE_AVX_INSTRUCTIONS=1; cmake --build .
 
 Doing so will make some things run faster.
 
-Finally, Visual Studio users should usually do everything in 64bit mode.  By default Visual Studio is 32bit, both in its outputs and its own execution, so you have to explicitly tell it to use 64bits.  Since it's not the 1990s anymore you probably want to use 64bits.  Do that with a cmake invocation like this:
+Visual Studio users can explicitly select a 64bit target and compiler with a CMake invocation like this:
 ```bash
-cmake .. -G "Visual Studio 14 2015 Win64" -T host=x64
+cmake .. -G "Visual Studio 17 2022" -A x64 -T host=x64
 ```
 
 ## Compiling your own C++ programs that use dlib

dlib/CMakeLists.txt

@@ -806,10 +806,13 @@ if (NOT TARGET dlib)
       add_library(dlib_all_source_cpp STATIC all/source.cpp) 
       target_link_libraries(dlib_all_source_cpp dlib)
       target_compile_options(dlib_all_source_cpp PUBLIC ${active_preprocessor_switches})
-      target_compile_features(dlib_all_source_cpp PUBLIC cxx_std_14)
+      target_compile_features(dlib_all_source_cpp PUBLIC cxx_std_17)
    endif()
 
-   target_compile_features(dlib PUBLIC cxx_std_14)
+   target_compile_features(dlib PUBLIC cxx_std_17)
+   if (DLIB_USE_CUDA)
+      target_compile_features(dlib PRIVATE cuda_std_17)
+   endif()
    # Build-tree clients share dlib's compiler, so forward the options detected
    # above to them. Do not put those options in the installed target, where
    # dlibConfig.cmake recreates them for the consuming compiler's frontend.

dlib/algs.h

@@ -10,12 +10,14 @@
 
 // this file contains miscellaneous stuff                      
 
-// Give people who forget the -std=c++14 option a reminder
-#if (defined(__GNUC__) && ((__GNUC__ >= 5 && __GNUC_MINOR__ >= 0) || (__GNUC__ > 5))) || \
-    (defined(__clang__) && ((__clang_major__ >= 3 && __clang_minor__ >= 4) || (__clang_major__ >= 3)))
-    #if __cplusplus < 201402L
-        #error "Dlib requires C++14 support.  Give your compiler the -std=c++14 option to enable it."
+// Give people who forget to enable C++17 a reminder.
+#if defined(_MSC_VER)
+    // MSVC only updates __cplusplus when /Zc:__cplusplus is enabled.
+    #if !defined(_MSVC_LANG) || _MSVC_LANG < 201703L
+        #error "Dlib requires C++17 support.  Give your compiler the /std:c++17 option to enable it."
     #endif
+#elif __cplusplus < 201703L
+    #error "Dlib requires C++17 support.  Give your compiler the -std=c++17 option to enable it."
 #endif
 
 #if defined __NVCC__
@@ -32,10 +34,6 @@
 
 #ifdef _MSC_VER
 
-#if  _MSC_VER < 1900
-#error "dlib versions newer than v19.1 use C++11 and therefore require Visual Studio 2015 or newer."
-#endif
-
 // Disable the following warnings for Visual Studio
 
 // this is to disable the "'this' : used in base member initializer list"

dlib/java/cmake_swig_jni

@@ -178,7 +178,7 @@ FILE(MAKE_DIRECTORY "${CMAKE_CURRENT_BINARY_DIR}/${binary_dir_name}")
 
 set_source_files_properties(${CMAKE_CURRENT_BINARY_DIR}/swig.i PROPERTIES CPLUSPLUS ON)
 swig_add_module(${output_library_name} java ${CMAKE_CURRENT_BINARY_DIR}/swig.i ${source_files})
-target_compile_features(${output_library_name} PUBLIC cxx_std_14)
+target_compile_features(${output_library_name} PUBLIC cxx_std_17)
 
 include_directories(${JNI_INCLUDE_DIRS})
 swig_link_libraries(${output_library_name}  ${additional_link_libraries})

dlib/serialize.h

@@ -2231,7 +2231,7 @@ namespace dlib
             //when deserializing unique_ptr, this is fresh state, so reset the pointers, even if item is non-empty
             bool is_non_empty;
             deserialize(is_non_empty, in);
-            item.reset(is_non_empty ? new T() : nullptr); //can't use make_unique since dlib does not use C++14 as a minimum requirement.
+            item.reset(is_non_empty ? new T() : nullptr);
 
             if (is_non_empty)
                 deserialize(*item, in);

dlib/test/WINDOWS_build_and_run_all_unit_tests.bat

@@ -9,18 +9,18 @@ echo testing python >> test_log.txt
 rmdir /S /Q build_python
 mkdir build_python
 cd build_python
-cmake -G "Visual Studio 14 2015 Win64" -T host=x64 ../../../tools/python  
+cmake -G "Visual Studio 17 2022" -A x64 -T host=x64 ../../../tools/python
 cmake --build . --config Release || exit /B
 ping 127.0.0.1 -n 5 -w 1000 > null
 cd ..
 
 
 
-echo testing vc2015 >> test_log.txt
-rmdir /S /Q build_vc2015_64
-mkdir build_vc2015_64
-cd build_vc2015_64
-cmake -G "Visual Studio 14 2015 Win64" -T host=x64 .. 
+echo testing vc2022 >> test_log.txt
+rmdir /S /Q build_vc2022_64
+mkdir build_vc2022_64
+cd build_vc2022_64
+cmake -G "Visual Studio 17 2022" -A x64 -T host=x64 ..
 cmake --build . --config Release || exit /B
 ping 127.0.0.1 -n 5 -w 1000 > null
 cmake --build . --config Debug || exit /B

dlib/test/constexpr_if.cpp

@@ -192,7 +192,7 @@ namespace
     public:
         constexpr_if_test (
         ) : tester ("test_constexpr_if",
-                    "Runs tests on the C++14 approximation of C++17 if constexpr() statements but better.")
+                    "Runs tests on dlib's compile time dispatch utilities.")
         {}
 
         void perform_test (

dlib/test/makefile

@@ -6,7 +6,7 @@ TARGET = dtest
 
 # these are the compile time flags passed to gcc
 CXXFLAGS ?= -ggdb -Wall
-CPPFLAGS ?= -std=c++14 -DDEBUG -DDLIB_NO_GUI_SUPPORT -I../..
+CPPFLAGS ?= -std=c++17 -DDEBUG -DDLIB_NO_GUI_SUPPORT -I../..
 
 # These are the link time flags passed to gcc
 LFLAGS = -lpthread   -lnsl 

docs/docs/compile.xml

@@ -13,7 +13,7 @@
                from github, go to the base folder of the dlib repository,
                and run <tt>python setup.py install</tt>.  Once either of these commands
                finishes running you are ready to use dlib from Python.  Note
-               that you need to have CMake and a working C++ compiler
+               that you need to have CMake and a working C++17 compiler
                installed for this to work.  
             </p>
 
@@ -53,7 +53,7 @@ cd build
 cmake ..
 cmake --build . --config Release
 </code_box>
-Note that you need to have a C++14 compiler installed on your system.  There are free C++14 compilers
+Note that you need to have a C++17 compiler installed on your system.  There are free C++17 compilers
 for most operating systems.  For example, Visual Studio is free on Windows and GCC is free and
 works well on Mac OS X and Linux systems.  If you have multiple compilers/IDEs installed then you can
 tell CMake which one you want it to use via the -G option.
@@ -65,10 +65,9 @@ tell CMake which one you want it to use via the -G option.
    makes it run much faster (also see <a href="faq.html#Whyisdlibslow">this FAQ</a>). 
 </p>
 <p>
-   Finally, note that when using Visual Studio, CMake will by default generate a 32bit executable.  
-   This means the programs you compile will only be able to use 2GB of RAM.  To avoid this, you need
-   to tell CMake to generate a 64bit executable.  You do this by using a command like 
-   <code_box>cmake -G "Visual Studio 14 2015 Win64" -T host=x64 ..</code_box> instead of <code_box>cmake ..</code_box>  
+   When using Visual Studio, select a 64bit target to avoid the memory limits of 32bit executables.
+   You can do this by using a command like
+   <code_box>cmake -G "Visual Studio 17 2022" -A x64 -T host=x64 ..</code_box> instead of <code_box>cmake ..</code_box>
    You can see the list of valid arguments to <tt>-G</tt> by running <tt>cmake</tt> with no options.   Note also the <tt>-T host=x64</tt>
    option, which tells Visual Studio to let the compiler use more than 2GB of RAM.  That is important if you don't want the compiler to
    crash from running out of RAM in some situations.
@@ -119,7 +118,7 @@ tell CMake which one you want it to use via the -G option.
         <h3>Compiling on Linux From Command Line</h3>
         From within the examples folder, you can compile nearly all of the examples with a single command like so:
 <code_box>
-g++ -std=c++14 -O3 -I.. ../dlib/all/source.cpp -lpthread -lX11 example_program_name.cpp
+g++ -std=c++17 -O3 -I.. ../dlib/all/source.cpp -lpthread -lX11 example_program_name.cpp
 </code_box>
 
 On non-Linux systems like Solaris, you might have to link to other libraries.  For example, I have seen systems
@@ -136,9 +135,10 @@ sudo apt-get install libx11-dev
             (via the -l option) to the following libraries: gdi32, comctl32, user32, winmm, ws2_32, or imm32.
         </p>
 
-        <h3>Compiling on Windows Using Visual Studio 2015 or Newer</h3>
+        <h3>Compiling on Windows Using Visual Studio</h3>
         <p>
-           All you need to do is create an empty console project. Then add dlib/all/source.cpp to it and add the
+           Create an empty console project and enable C++17 with <tt>/std:c++17</tt>.
+           Then add dlib/all/source.cpp to it and add the
            folder containing the dlib folder to the #include search path.  Then you can compile any example program 
            by adding it to your project.  
         </p>

docs/docs/containers.xml

@@ -739,7 +739,7 @@
          <spec_file>dlib/optional.h</spec_file>
          <description>
             This is a standard's compliant backport of <a href="https://en.cppreference.com/w/cpp/utility/optional">std::optional</a> 
-            that works with C++14.  It includes C++23 monadic interfaces
+            that works with C++17.  It includes C++23 monadic interfaces
          </description>
       </component>
 

docs/docs/faq.xml

@@ -128,8 +128,8 @@
 
       <question text="How can I use dlib in Visual Studio?">
          <p>
-            First, note that you need a version of Visual Studio with decent
-            C++11 support.  This means you need Visual Studio 2015 or newer.
+            First, note that you need a version of Visual Studio with C++17 support.
+            Enable it with <tt>/std:c++17</tt>, or use CMake to select it automatically.
          </p>
          There are instructions on the <a href="compile.html">How to Compile</a> page.  
          If you do not understand the instructions in the "<b>Compiling on Windows Using Visual Studio</b>" section
@@ -554,22 +554,8 @@ cross_validate_trainer_threaded(trainer,
 
    <questions group="Deep Learning">
       <question text="Why can't I use the DNN module with Visual Studio?">
-         You can, but you need to use Visual Studio 2015 Update 3 or newer since prior versions
-         had bad C++11 support.  To make this as confusing as possible,
-         Microsoft has released multiple different versions of "Visual Studio
-         2015 Update 3".  As of October 2016, the version available from the
-         Microsoft web page has good enough C++11 support to compile the DNN
-         tools in dlib.  So make sure you have a version no older than October
-         2016.
-         <p>
-            To make this even more complicated, Visual Studio 2017 had
-            regressions in its C++11 support.  So all versions of Visual Studio
-            2017 prior to December 2017 would just hang if you tried to compile
-            the DNN examples.  Happily, the newest versions of Visual Studio
-            2017 appear to have good C++11 support and will compile the DNN
-            codes without any issue.  So make sure your Visual Studio is
-            fully updated.
-         </p>
+         You can, but you need a version of Visual Studio with C++17 support.
+         Keep Visual Studio fully updated and use CMake to enable the required compiler options.
          <p>
             Finally, it should be noted that you should give the <tt>-T host=x64</tt>
             cmake option when generating a Visual Studio project.  If you don't

docs/docs/metaprogramming.xml

@@ -1002,7 +1002,7 @@ or you can omit the error message and call it like:
          <file>dlib/constexpr_if.h</file>
          <spec_file link="true">dlib/constexpr_if.h</spec_file>
          <description>        
-            This function allows you to do compile time dispatch similar to what you can do with constexpr if in C++17, but only requires a C++14 compiler.
+            This function allows you to do compile time dispatch similar to what you can do with if constexpr in C++17.
          </description>
 
       </component>

docs/docs/release_notes.xml

@@ -11,6 +11,17 @@
 <!-- **************************************************************************************  -->
 
 <current>
+New Features and Improvements:
+
+Non-Backwards Compatible Changes:
+   - Dlib now requires C++17.
+
+Bug fixes:
+</current>
+
+<!-- **************************************************************************************  -->
+
+<old name="20.0" date="May 27, 2025">
 New Features and Improvements:
    - Machine Learning
       - Added auto_train_multiclass_svm_linear_classifier(), a tool that trains a multiclass
@@ -69,7 +80,7 @@ Bug fixes:
    - Fix pixel saturation in interpolate_quadratic (PR #2806)
    - Fix convolution backward filter cache issue when using cudnn 7 and above (PR #2828)
    - Fix a bug in the gradient computation code for the layer_normalize layer.
-</current>
+</old>
 
 <!-- **************************************************************************************  -->
 

setup.py

@@ -18,7 +18,7 @@
     --no USE_AVX_INSTRUCTIONS: will set -DUSE_AVX_INSTRUCTIONS=no
 Additional options:
     --compiler-flags: pass flags onto the compiler, e.g. --compiler-flags "-Os -Wall" passes -Os -Wall onto GCC.
-    -G: Set the CMake generator.  E.g. -G "Visual Studio 14 2015"
+    -G: Set the CMake generator.  E.g. -G "Visual Studio 17 2022"
     --clean: delete any previous build folders and rebuild.  You should do this if you change any build options
              by setting --compiler-flags or --no since the last time you ran a build.  This will
              ensure the changes take effect.