Issues with Linux port

 

Linux port

 The company who employs me has decided to port parts of our application to Linux. At a first shot we will use WSL and Visual Studio but issues are not over:

• CMake is the lingua franca of generating cross platform build scripts. CMake is a beast of itself however and not sure why this got so popular.
• WSL keeps sometimes its old configuration and source files. It seems that when the source files are read only they are read only on the target WSL system as well. If you edit a file afterwards the update of this file will fail so one won't see their changes. Either make the source files not read only beforehand or clean all files and directories on the WSL host and then start fresh again. Alternatively make the source files writable on Linux through chmod. 
• MSVC uses __declspec(dllexport) to export functions from DLL's; GCC doesn't have that.
• MSVC is pushing security enhanced versions of the CRT through its code analyzer. According to cppreference these functions are standardized albeit as extension (i.e. Annex K of C11). Unfortunately glibc has not implemented them; partially because some dubious reasoning. The API isn't perfect and there are pre-exisiting bounds checking crt but still it's standardized and some people (we) use them. So one ends up writing Windows and Linux specific code even in a layer which supposed to be platform independent.
• Many of the MSVC C API is Windows specific (e.g. _splitpath; _makepath; _tchdir). Using C alone on Windows platform may therefore still not platform independent.
• Warning suppression's in pre-compiled header in GCC are ignored in code. This is quite unhandy; especially since some suppression's one want to apply globally to all sources and are therefore are primary candidate to put in pre-compiled header. I have filed a bug 123287 report and it's stated that this has been solved for GCC 15.x.
• On Windows wchar_t are 2 bytes and represent UCS-2 or UTF-16. On Linux wchar_t it is 4 bytes and probably represents UTF-32. To be compatible with existing persistence storage we had to use char16_t on certain places in the code. Unfortunately it seems that some character code conversion facilities are deprecated so this solution will not hold out for long.
• std::basic_ifstream and std::basic_ofstream don't accept std::wstring as function name on Linux. This seems to be a MSVC extension so change code to use std::filesystem::path which is a conformance improvement.
• __FUNCTION__ is an extension which both MSVC and GCC understand. On GCC it is not a macro so prepending it with 'L' to get the wide character variant does not work. It also only gives the function name without class in case member function which makes in unattractive. So specific MSVC and GCC code is needed. There is a standard: __func__. However again it only gives the function name. source_location is another alternative but this gives too much information for the function name. 

GCC might still contain some basic bugs. The warning about #pragma once in main file (when building a pre-compiled header) is only solved in version 14.

Using clang-cl in Visual Studio

clang-cl

 clang-cl is the command line tool in Visual Studio capable of invoking the clang compiler with the arguments of msvc. In Visual Studio projects one can just flip the toolset and the clang compiler will be chosen. clang has the following positive aspects:

• better C++ conformance. Examples are that msvc is leniant towards missing 'typename' for dependent types and 'template' for nesting templates; clang picks them up. There are other issues.
• offers some code improvements like correct member order in constructors and virtuals which override base class
• detects some performance improvements like advising to use shared_ptr by reference in loops
• more precise compilation warnings and errors

It has also some drawbacks: 

• some noisy warnings 
• does not understand all msvc code. For example the msvc's #import extension is not understood

 Despite using msvc's code analysis the clang compiler was still able to pick up other issues. Some clang warnings are far fetched and one can choose to disable them. This is especially needed for external libraries which one cannot easily patch. To disable warnings one can use the following:

#ifdef __clang__
#pragma clang diagnostic ignored "-Wimplicit-exception-spec-mismatch"
#pragma clang diagnostic ignored "-Wmissing-field-initializers"
#pragma clang diagnostic ignored "-W#pragma-messages"
#pragma clang diagnostic ignored "-Wunused-but-set-variable"
#pragma clang diagnostic ignored "-Wunused-local-typedef"
#endif

 The first one for example is necessary to suppress warnings in MFC. 'delete' should be specified with 'noexcept' but the MFC delete lacks this.

Watch out for hypes in ICT

Hypes

 ICT has a rich history of hypes where people thought that this would be a panacea for all problems. These hypes lasted for some time like paradigms in Thomas Kuhn's theory about evolution of science. From the top of my head we had the following hypes in the past:

• relational / SQL databases (70's)
• structural design
• object oriented design (80's)
• component based development (90's)
• design patterns (1995)
• scrum / agile (2001)
• AI (2022)

 Many of these hypes were initially promising but not to the extend of solving all problems. They are now part of the current solution domain. We also know now that there are still problems to tackle. 

 Let's see what AI will bring us in the future. For now it's on the level of coding assist but not on the level of designing whole systems. In that part it still cannot replace programmers. There are already studies mitigating the effect of using AI.  It also still makes mistakes. From personal experience it can introduce errors in a code base if you let it run without crosschecking.

 Scrum has brought nothing to ICT except misery. The company I work for took a major loss after embracing it. 

Careful with AI tooling

AI tooling

 Since some period I started working with AI tooling. Mostly I use Gemini and Copilot inside Visual Studio. The experience is a bit of mixed feelings about this. Gemini had some good suggestions but failed also many times. Copilot has good code completion suggestions but misses the mark also. Copilot's function name suggestion are very welcome.

 On the other hand AI tooling is still full of mistakes. Some examples:

• I asked Gemini for camera sharpness algorithm. It came up with a good algorithm but the actual OpenCV function calls and parameters were incorrect.
• I asked Gemini to get the real sample time from an 'IMediaSample'. It suggest to use the non existing 'GetSampleTime'. There is b.t.w. a 'GetMediaTime' function but this returns the stream time; i.e. the time since the graph was running and not the time from the start of the video. 
• I asked Gemini lately of conversion from UCS-2 to UTF-16 and it wrongly suggested to use wstring_convert. However wstring_covert is hardbound to std::string as byte_string

 Even worse that sometimes AI tooling can suggest plain bugs. I was implementing a swap of width and height and Copilot's code complete came up with the following code snippet:

// NOTE: incorrect 

Size sz = ...;
if (sz.GetWidth() < sz.GetHeight())
{
   sz.SetWidth(sz.GetHeight());
   sz.SetHeight(sz.GetWidth());
}

This doesn't swap but sets the width and height on the old height value.

 AI tooling can be helpful but are still not on the level to be trusted blindly. They also now help with limited scope; e.g. code blocks; algorithms and functions. I am not aware if they can help in refactoring and extending architecture spanning solutions.

 

Debugging GDI drawing

GDI debugging

 The other day I had to debug a hard to track drawing bug. The application is built with the MFC framework so it still uses GDI on places to draw custom controls.

 The incorrect drawing artifact was displayed after an invocation of 'DrawText' with the flag 'DT_CALCRECT'. This was unexpected since with the flag the function doesn't draw and only measures the size. Eventually I realized that GDI batches invocations so perhaps the buggy overdrawing had already taken place before. What was needed to prove this hypothesis:

•  suppress GDI's caching mechanism through 'GdiSetBatchLimit'.
•  use direct drawing; so no memory device context

 With this in place indeed it could be seen that the mistake happened earlier in the code and that the 'DrawText' invocation was merely a flush of the GDI batch.

 Be aware that suppressing  GDI's batch might not always work. When the window where the drawing took place was on the primary monitor the batch mode could be turned off but on the second monitor it still cached its calls.

Watch out for an old VC++ runtime

 VC_redist.x64.exe

 For C/ C++ applications the VC++ runtime needs to be installed on the computer. The other day we experienced crashes when a component developed with a late version of VS2022 was crashing on a fresh installation of Windows 11. It turned out that this Windows 11 still uses an old version of the VC++ runtime which could crash the application (most notably in grabbing a std::mutex lock). After updating the PC with a recent version of 'VC_redist.x64.exe' the problem was solved.

 

ark.intel.com

 

ark.intel.com

 Intel had a wonderful website where one could easily lookup the processor and see what capabilities (e.g. SSE 4.2; AVX; AVX2) it had. In a recent visit it was completely overhauled and they have removed (or hidden) the easy possibility to lookup your processor with one click. Thanks Intel for modernizing their website and destroying a valuable functionality.