I am trying to compile an all static build of source-customized ffmpeg libraries with the additional codecs from VideoLAN, such as their x264 codec. I am ultimately building for use in Windows with Visual Studio 2015 C++, so I am comparing a few compiling guides for my instructions.
In the guide linked below (I located via Google searches) shows how to compile x264 with the VS2015 toolchain in MSYS2 without any special steps: https://gist.github.com/RangelReale/3e6392289d8ba1a52b6e70cdd7e10282
While in this other guide, offered by the compiling guide pages from ffmpeg's web site, it includes steps for "fixing the build scripts" for x264: https://www.roxlu.com/2019/062/compiling-ffmpeg-with-x264-on-windows-10-using-msvc
Problem being, when I "fix the build scripts" the configure script no longer finds a C compiler. Building with the first guide seems to work, but it was also written nearly 2 years ago, so perhaps the ffmpeg guide is doing some important fix? I don't know.
I am:
and at this point the two compiling guides vary, the first guide just cd's into x264 and runs configure while the other guide does some curl download of a config.guess file and then some sed manipulation of ./configure. Considering these instructions are "endorsed" by the ffmpeg site, I am left wondering what that is doing and how critical is it that these "special" steps cause my x264 configure to fail.
All this manipulations with "config.guess" and "configure" are only needed because later on they run simple "msys2_shell.cmd" i.e. in MSYS-mode. Correct/supported way in x264 is to run shell in MINGW32 or MINGW64 mode with "-mingw32" and "-mingw64" options like you do (or run mingw32.exe / mingw64.exe instead of msys2_shell.cmd). This change result in different output of "uname" command which is important for x264's configure.