At the end of the last article (In September 2017 actually), I had a fully automated regression testing system for IPXWrapper. In September of 2019, after two years of not touching the system and doing a little work on IPXWrapper itself, I felt it was time to install Windows updates in the VM images... and that's where everything went wrong....

The GNU Make shell function executes a shell command, expanding to whatever it wrote to standard output. If the command fails, thats just fiiine - however much it did or didn't output is what you get.

command_output := $(shell command)

I use the shell function for (among other things) getting compiler flags necessary for using libraries (e.g. from pkg-config), and Make's behaviour of ignoring these errors is rather annoying, since it can mask the actual error with pages of output before getting to a build command that actually fails. Sometimes it can even mess up the compiler flags to the point you (or at least I) spend ages hunting down a phantom problem that doesn't exist.

Weirdly, I couldn't find any good solution on the web, with a bit of experimentation I wound up developing the following macro:

# Wrapper around the $(shell) function that aborts the build if the command
# exits with a nonzero status.
shell-or-die = $\
	$(eval sod_out := $(shell $(1); echo $$?))$\
	$(if $(filter 0,$(lastword $(sod_out))),$\
		$(wordlist 1, $(shell echo $$(($(words $(sod_out)) - 1))), $(sod_out)),$\
		$(error $(1) exited with status $(lastword $(sod_out))))

WX_CXXFLAGS := $(call shell-or-die,wx-config --cxxflags base core aui propgrid adv)
WX_LIBS     := $(call shell-or-die,wx-config --libs     base core aui propgrid adv)

If you don't mind depending on GNU Make 4.2 (a bit too new for my tastes), if can be simplified to:

shell-or-die = $\
	$(eval sod_out := $(shell $(1)))$\
	$(if $(filter 0,$(.SHELLSTATUS)),$\
		$(sod_out),$\
		$(error $(1) exited with status $(.SHELLSTATUS)))

In 2014 I wrote a fairly comprehensive test suite for IPXWrapper, which tests it end-to-end, from the APIs through to the network traffic they generate and process. It depends on a meticulously configured set of Windows and Linux machines, which I had duplicated using several different versions of Windows.

Eventually bit-rot set in and some of the Windows VMs became unusable for quick testing; sat installing updates whenever I booted them, broke themselves in odd ways, etc. Also my workstation doesn't have enough RAM for Chrome and several Windows VMs at the same time. No machine does.

Today I was writing some C++ and wanted to add a private struct within a class for storing some data, but not just any struct - I wanted a private abstract base struct with a couple of implementations.

So I wrote something like this:

...

That didn't compile, GCC gave me the following error and Google wasn't terribly helpful when I searched for it:

/home/solemnwarning/test.cpp:8:3: error: ‘struct Thing::PrivateAbstract::A’ redeclared with different access