CodingStyle


In general, use the same coding style as the surrounding code.

However, do not make any unnecessary changes as that complicates
the VCS (git) history and makes it harder to merge patches. So
do not modify code just to make it conform to a coding style.

    Indentation

Turn on a "fill tabs with spaces" option in your editor.

Remove tabs and trailing spaces from any lines that are modified.

Note that some files are indented with 2 spaces (when they
have large indentation) while most are indented with 4 spaces.

    Language

TCC is mostly implemented in C90. Do not use any non-C90 features
that are not already in use.

Non-C90 features currently in use, as revealed by
./configure --extra-cflags="-std=c90 -Wpedantic":

- long long (including "LL" constants)
- inline
- very long string constants
- assignment between function pointer and 'void *'
- "//" comments
- empty macro arguments (DEF_ASMTEST in i386-tok.h)
- unnamed struct and union fields (in struct Sym), a C11 feature

    Testing

A simple "make test" is sufficient for some simple changes. However,
before committing a change consider performing some of the following
additional tests:

- Build and run "make test" on several architectures.

- Build with ./configure --enable-cross.

- If the generation of relocations has been changed, try compiling
  with TCC and linking with GCC/Clang. If the linker has been
  modified, try compiling with GCC/Clang and linking with TCC.

- Test with ASan/UBSan to detect memory corruption and undefined behaviour:

make clean
./configure
make
make test
cp libtcc.a libtcc.a.hide

make clean
./configure --extra-cflags="-fsanitize=address,undefined -g"
make
cp libtcc.a.hide libtcc.a
make test

- Test with Valgrind to detect some uses of uninitialised values:

make clean
./configure
make
# On Intel, because Valgrind does floating-point arithmetic differently:
( cd tests && gcc -I.. tcctest.c && valgrind -q ./a.out > test.ref )
make test TCC="valgrind -q --leak-check=full `pwd`/tcc -B`pwd` -I`pwd`"

  (Because of how VLAs are implemented, invalid reads are expected
  with 79_vla_continue.)