Which makes libc++ since version 16 not use transitive includes for
backwards compatibility, both for faster build times and to detect if
some code is missing a transitive include (which could blow up for the
user even w/o this macro defined, as compiling with -std=c++23 also
removes quite a lot such includes).
This affects Android, macOS/iOS and Emscripten. On the CI right now all
VMs use older libc++, which means it does nothing, but it's done there
anyway to future-proof. Locally on Arch it affects the Android and
Emscripten builds already, as well as an explicit clang + libc++ build.
FUCKING stupid defaults. CMake's default no less stupid. I'm MAD, why
would anybody even want to have the build run sequentially on todays
machines?! If your build is crap and can't run in parallel, FIX YOUR
CODE, but don't make the other 99% users suffer!!
I kinda get the point, to avoid rogue third parties tainting my
ŕepositories with nasty backdoors, yes, but unless *everything* on the
way to the server including DNS queries is encrypted, then it's still
just a half of the solution and any silly corporate firewall can still
prevent me from cloning stuff from github dot com.
Took me a while (several years?) to figure out a way to benchmark this
without basically duplicating the testing effort and without new
variants being too hard to add.
This makes it possible to:
- finally use Magnum as a CMake subproject on Windows and have your
executables not fail to run with a "DLL missing" error (and the
setting is put to cache so superprojects just implicitly make use of
that)
- run tests on Windows without having to install first
- use dynamic plugins from a CMake subproject on any platform without
having to install first or load them by filename --- and the plugin
directory is now easily discovered as relative to
libraryLocation() of the library implementing given plugin interface
Currently a separate job with everything that depends on GL disabled.
And against a minimal Vulkan driver that has just four symbols, because
I can't run the stuff there anyway.
What's done:
* The build matrix now has four targets -- Linux desktop, OSX desktop,
iOS GLES2 and iOS GLES3.
* Putting the commands into separate bash scripts, which are called
based on what's the actual build matrix item.
* Displaying *all* test output (colored) for visual verification.