Fix compilation with Emscripten

.github/setenv.sh

@@ -298,6 +298,10 @@ function c4_cfg_test()
                   -DCMAKE_C_FLAGS="-std=c99 -m$bits" -DCMAKE_CXX_FLAGS="-m$bits"
             cmake --build $build_dir --target help | sed 1d | sort
             ;;
+        em++)
+            emcmake cmake -S $PROJ_DIR -B $build_dir -DCMAKE_INSTALL_PREFIX="$install_dir" \
+                  -DCMAKE_BUILD_TYPE=$BT $CMFLAGS -DCMAKE_CXX_FLAGS="-s DISABLE_EXCEPTION_CATCHING=0"
+            ;;
         *)
             echo "unknown compiler"
             exit 1
@@ -367,7 +371,7 @@ function _c4_parallel_build_flags()
                 echo "-IDEBuildOperationMaxNumberOfConcurrentCompileTasks=$NUM_JOBS_BUILD"
             fi
             ;;
-        *g++*|*gcc*|*clang*)
+        *g++*|*gcc*|*clang*|em++)
             if [ -z "$NUM_JOBS_BUILD" ] ; then
                 echo "-j $(nproc)"
             else
@@ -394,7 +398,7 @@ function _c4_generator_build_flags()
             # https://stackoverflow.com/questions/51153525/xcode-10-unable-to-attach-db-error
             echo "-UseModernBuildSystem=NO"
             ;;
-        *g++*|*gcc*|*clang*)
+        *g++*|*gcc*|*clang*|em++)
             ;;
         "") # allow empty compiler
             ;;

.github/workflows/emscripten.yml

@@ -0,0 +1,95 @@
+name: emscripten
+
+defaults:
+  #if: "!contains(github.event.head_commit.message, 'skip ci')"  # SKIP
+  run:
+    # Use a bash shell so we can use the same syntax for environment variable
+    # access regardless of the host operating system
+    shell: bash -e -x {0}
+
+on:
+  # https://git.luolix.topmunity/t/how-to-trigger-an-action-on-push-or-pull-request-but-not-both/16662
+  workflow_dispatch:
+  push:
+    branches:
+    - master
+  pull_request:
+    branches:
+    - master
+
+env:
+  PROJ_PFX_TARGET: c4core-
+  PROJ_PFX_CMAKE: C4CORE_
+  CMAKE_FLAGS:
+  NUM_JOBS_BUILD: # 4
+  EMSCRIPTEN_CACHE_FOLDER: 'emsdk-cache'
+
+
+# ubuntu-20.04:
+#   # https://github.com/actions/virtual-environments/blob/main/images/linux/Ubuntu2004-README.md
+#   gcc: 7.5.0, 8.4.0, 9.3.0, 10.2.0
+#   clang: 8.0.1, 9.0.1, 10.0.0
+# ubuntu-18.04:
+#   # https://github.com/actions/virtual-environments/blob/main/images/linux/Ubuntu1804-README.md
+#   gcc: 7.5.0, 8.4.0, 9.3.0, 10.1.0
+#   clang: 6.0.0, 8.0.0, 9.0.0
+# macos-11.0: macOS Big Sur 11.0
+#   # https://github.com/actions/virtual-environments/blob/main/images/macos/macos-11.0-Readme.md
+#   Xcode 12.1 11.7
+#   clang/LLVM 10.0.1
+#   gcc-8 gcc-9
+# macos-10.15: macOS Catalina 10.15
+#   # https://github.com/actions/virtual-environments/blob/main/images/macos/macos-10.15-Readme.md
+#   Xcode 12.1 11.7
+#   clang/LLVM 11.0.0
+#   gcc-8 gcc-9
+# windows-2019:
+#   # https://github.com/actions/virtual-environments/blob/main/images/win/Windows2019-Readme.md
+#   vs2019
+# windows-2016:
+#   # https://github.com/actions/virtual-environments/blob/main/images/win/Windows2016-Readme.md
+#   vs2017
+
+jobs:
+
+  #----------------------------------------------------------------------------
+  emscripten:
+    name: emscripten/${{matrix.emver}}/c++${{matrix.std}}/${{matrix.bt}}
+    continue-on-error: true
+    if: always()  # https://stackoverflow.com/questions/62045967/github-actions-is-there-a-way-to-continue-on-error-while-still-getting-correct
+    runs-on: ${{matrix.os}}
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - {std: 11, cxx: em++, emver: 2.0.34, bt: Debug  , os: ubuntu-latest, bitlinks: static32}
+          - {std: 11, cxx: em++, emver: 2.0.34, bt: Release, os: ubuntu-latest, bitlinks: static32}
+          - {std: 20, cxx: em++, emver: 2.0.34, bt: Debug  , os: ubuntu-latest, bitlinks: static32}
+          - {std: 20, cxx: em++, emver: 2.0.34, bt: Release, os: ubuntu-latest, bitlinks: static32}
+          - {std: 11, cxx: em++, emver: 3.0.0 , bt: Debug  , os: ubuntu-latest, bitlinks: static32}
+          - {std: 11, cxx: em++, emver: 3.0.0 , bt: Release, os: ubuntu-latest, bitlinks: static32}
+          - {std: 20, cxx: em++, emver: 3.0.0 , bt: Debug  , os: ubuntu-latest, bitlinks: static32}
+          - {std: 20, cxx: em++, emver: 3.0.0 , bt: Release, os: ubuntu-latest, bitlinks: static32}
+    env:
+      STD: "${{matrix.std}}"
+      CXX_: "${{matrix.cxx}}"
+      BT: "${{matrix.bt}}"
+      BITLINKS: "${{matrix.bitlinks}}"
+      VG: "${{matrix.vg}}"
+      SAN: "${{matrix.san}}"
+      LINT: "${{matrix.lint}}"
+      OS: "${{matrix.os}}"
+    steps:
+      - {name: checkout, uses: actions/checkout@v2, with: {submodules: recursive}}
+      - name: setup emscripten cache
+        id: cache-system-libraries
+        uses: actions/cache@v2
+        with: {path: "${{env.EMSCRIPTEN_CACHE_FOLDER}}", key: "${{matrix.emver}}-${{runner.os}}"}
+      - name: setup emscripten
+        uses: mymindstorm/setup-emsdk@v11
+        with: {version: "${{matrix.emver}}", actions-cache-folder: "${{env.EMSCRIPTEN_CACHE_FOLDER}}"}
+      - {name: show info, run: source .github/setenv.sh && c4_show_info}
+      - name: static32-configure---------------------------------------------------
+        run: source .github/setenv.sh && c4_cfg_test static32
+      - {name: static32-build, run: source .github/setenv.sh && c4_build_test static32}
+      - {name: static32-run, run: source .github/setenv.sh && c4_run_test static32}

README.md

@@ -23,6 +23,19 @@ I've been using these utilities in this or similar forms for some years now,
 and I've found them incredibly useful in my projects. I'm packing these as a
 separate library, as all of my projects use it.
 
+c4core is [extensively unit-tested in Linux, Windows and
+MacOS](https://github.com/biojppm/c4core/actions). The tests cover
+x64, x86, arm, wasm (emscripten), aarch64, ppc64le and s390x
+architectures, and include analysing c4core with:
+  * valgrind
+  * clang-tidy
+  * clang sanitizers:
+    * memory
+    * address
+    * undefined behavior
+    * thread
+  * [LGTM.com](https://lgtm.com/projects/g/biojppm/c4core)
+
 
 ## Obtaining c4core
 

changelog/current.md

@@ -22,6 +22,7 @@
       ```
 - Add `csubstr::is_unsigned_integer()` and `csubstr::is_real()` ([PR #49](https://github.com/biojppm/c4core/pull/49)).
 - CMake: add alias target c4core::c4core, guaranteeing that the same code can be used with `add_subdirectory()` and `find_package()`. (see [rapidyaml #173](https://github.com/biojppm/rapidyaml/issues/173))
+- Add support for compilation with emscripten (WebAssembly+javascript) ` ([PR #52](https://github.com/biojppm/c4core/pull/52)).
 
 
 ### Fixes
@@ -37,3 +38,6 @@
 - In `c4/charconv.hpp`: do not use C4_ASSERT in `to_c_fmt()`, which is `constexpr`.
 
 
+### Thanks
+
+- @cschreib

src/c4/bitmask.hpp

@@ -88,10 +88,11 @@ bm2stream(Stream &s, Enum value, EnumOffsetType offst=EOFFS_PFX)
 
 // some utility macros, undefed below
 
+/// @cond dev
 
-/** Execute @p code if the @p num of characters is available in the str
+/* Execute `code` if the `num` of characters is available in the str
  * buffer. This macro simplifies the code for bm2str().
- * @todo improve this by writing from the end and moving only once. */
+ * @todo improve performance by writing from the end and moving only once. */
 #define _c4prependchars(code, num)                                      \
     if(str && (pos + num <= sz))                                        \
     {                                                                   \
@@ -100,19 +101,29 @@ bm2stream(Stream &s, Enum value, EnumOffsetType offst=EOFFS_PFX)
         /* now write in the beginning of the string */                  \
         code;                                                           \
     }                                                                   \
-    else if(str && sz) { C4_ERROR("cannot write to string pos=%d num=%d sz=%d", (int)pos, (int)num, (int)sz); } \
+    else if(str && sz)                                                  \
+    {                                                                   \
+        C4_ERROR("cannot write to string pos=%d num=%d sz=%d",          \
+                 (int)pos, (int)num, (int)sz);                          \
+    }                                                                   \
     pos += num
 
-/** Execute @p code if the @p num of characters is available in the str
+/* Execute `code` if the `num` of characters is available in the str
  * buffer. This macro simplifies the code for bm2str(). */
 #define _c4appendchars(code, num)                                       \
     if(str && (pos + num <= sz))                                        \
     {                                                                   \
         code;                                                           \
     }                                                                   \
-    else if(str && sz) { C4_ERROR("cannot write to string pos=%d num=%d sz=%d", (int)pos, (int)num, (int)sz); } \
+    else if(str && sz)                                                  \
+    {                                                                   \
+        C4_ERROR("cannot write to string pos=%d num=%d sz=%d",          \
+                 (int)pos, (int)num, (int)sz);                          \
+    }                                                                   \
     pos += num
 
+/// @endcond
+
 
 /** convert a bitmask to string.
  * return the number of characters written. To find the needed size,
@@ -131,13 +142,13 @@ size_t bm2str
 
     auto syms = esyms<Enum>();
     size_t pos = 0;
-    typename EnumSymbols<Enum>::Sym const* zero = nullptr;
+    typename EnumSymbols<Enum>::Sym const* C4_RESTRICT zero = nullptr;
 
     // do reverse iteration to give preference to composite enum symbols,
     // which are likely to appear later in the enum sequence
     for(size_t i = syms.size()-1; i != size_t(-1); --i)
     {
-        auto const p = syms[i];
+        auto const &C4_RESTRICT p = syms[i]; // do not copy, we are assigning to `zero`
         I b = static_cast<I>(p.value);
         if(b == 0)
         {

src/c4/charconv.hpp

@@ -1209,8 +1209,10 @@ inline bool atod(csubstr str, double * C4_RESTRICT v)
 inline size_t atof_first(csubstr str, float * C4_RESTRICT v)
 {
     csubstr trimmed = str.first_real_span();
-    if(trimmed.len == 0) return csubstr::npos;
-    if(atof(trimmed, v)) return static_cast<size_t>(trimmed.end() - str.begin());
+    if(trimmed.len == 0)
+        return csubstr::npos;
+    if(atof(trimmed, v))
+        return static_cast<size_t>(trimmed.end() - str.begin());
     return csubstr::npos;
 }
 
@@ -1222,8 +1224,10 @@ inline size_t atof_first(csubstr str, float * C4_RESTRICT v)
 inline size_t atod_first(csubstr str, double * C4_RESTRICT v)
 {
     csubstr trimmed = str.first_real_span();
-    if(trimmed.len == 0) return csubstr::npos;
-    if(atod(trimmed, v)) return static_cast<size_t>(trimmed.end() - str.begin());
+    if(trimmed.len == 0)
+        return csubstr::npos;
+    if(atod(trimmed, v))
+        return static_cast<size_t>(trimmed.end() - str.begin());
     return csubstr::npos;
 }
 

src/c4/cpu.hpp

@@ -103,6 +103,10 @@
 #   define C4_WORDSIZE 8
 #   define C4_BYTE_ORDER _C4EB
 
+#elif defined(__EMSCRIPTEN__)
+#   define C4_BYTE_ORDER _C4EL
+#   define C4_WORDSIZE 4
+
 #elif defined(SWIG)
    #error "please define CPU architecture macros when compiling with swig"
 

src/c4/std/vector_fwd.hpp

@@ -14,7 +14,11 @@ template<typename T, typename Alloc> class vector;
 #elif defined(_LIBCPP_VERSION)
 namespace std {
 template<typename> class allocator;
+#if defined(__EMSCRIPTEN__)
+inline namespace __2 {
+#else
 inline namespace __1 {
+#endif
 template<typename T, typename Alloc> class vector;
 } // namespace __1
 } // namespace std

test/test_charconv.cpp

@@ -1162,7 +1162,7 @@ void test_rtoa(substr buf, Real f, rtoa_fn_t<Real> rtoa_fn, int precision, const
     size_t ret;
 
     INFO("num=" << f << " precision=" << precision
-         << "'  hexa='" << hexa << "'  hexa_alternative='" << hexa_alternative << "'");
+         << "  hexa='" << hexa << "'  hexa_alternative='" << hexa_alternative << "'");
 
     memset(buf.str, 0, buf.len);
     ret = rtoa_fn(buf, f, precision, FTOA_SCIENT);
@@ -1186,7 +1186,7 @@ void test_rtoa(substr buf, Real f, rtoa_fn_t<Real> rtoa_fn, int precision, const
     std::string report;
     from_chars(buf.left_of(ret), &report);
     bool ok = buf.left_of(ret) == to_csubstr(hexa) || buf.left_of(ret) == to_csubstr(hexa_alternative);
-    CHECK_MESSAGE(ok, "ret='" << report);
+    CHECK_MESSAGE(ok, "ret='" << report << "'");
 }
 
 
@@ -1196,6 +1196,18 @@ TEST_CASE("ftoa.basic")
     substr buf(bufc);
     C4_ASSERT(buf.len == sizeof(bufc)-1);
 
+    // earlier versions of emscripten's sprintf() do not respect some
+    // precision values when printing in hexadecimal format.
+    //
+    // @see https://github.com/biojppm/c4core/pull/52
+    #if defined(__EMSCRIPTEN__) && __EMSCRIPTEN_major__ < 3
+    #define _c4emscripten_alt(alt) , alt
+    #define _c4emscripten_alt2(alt1, alt2) , alt2
+    #else
+    #define _c4emscripten_alt(alt)
+    #define _c4emscripten_alt2(alt1, alt2) , alt1
+    #endif
+
     float f = 1.1234123f;
     double d = 1.1234123;
 
@@ -1205,11 +1217,11 @@ TEST_CASE("ftoa.basic")
     test_rtoa(buf, f, &ftoa, 1, /*scient*/"1.1e+00", /*flt*/"1.1", /*flex*/"1.1", /*hexa*/"0x1.2p+0");
     test_rtoa(buf, d, &dtoa, 1, /*scient*/"1.1e+00", /*flt*/"1.1", /*flex*/"1.1", /*hexa*/"0x1.2p+0");
 
-    test_rtoa(buf, f, &ftoa, 2, /*scient*/"1.12e+00", /*flt*/"1.12", /*flex*/"1.12", /*hexa*/"0x1.20p+0");
-    test_rtoa(buf, d, &dtoa, 2, /*scient*/"1.12e+00", /*flt*/"1.12", /*flex*/"1.12", /*hexa*/"0x1.20p+0");
+    test_rtoa(buf, f, &ftoa, 2, /*scient*/"1.12e+00", /*flt*/"1.12", /*flex*/"1.12", /*hexa*/"0x1.20p+0" _c4emscripten_alt("0x1.1f8p+0"));
+    test_rtoa(buf, d, &dtoa, 2, /*scient*/"1.12e+00", /*flt*/"1.12", /*flex*/"1.12", /*hexa*/"0x1.20p+0" _c4emscripten_alt("0x1.1f8p+0"));
 
-    test_rtoa(buf, f, &ftoa, 3, /*scient*/"1.123e+00", /*flt*/"1.123", /*flex*/"1.123", /*hexa*/"0x1.1f9p+0");
-    test_rtoa(buf, d, &dtoa, 3, /*scient*/"1.123e+00", /*flt*/"1.123", /*flex*/"1.123", /*hexa*/"0x1.1f9p+0");
+    test_rtoa(buf, f, &ftoa, 3, /*scient*/"1.123e+00", /*flt*/"1.123", /*flex*/"1.123", /*hexa*/"0x1.1f9p+0" _c4emscripten_alt("0x1.1f98p+0"));
+    test_rtoa(buf, d, &dtoa, 3, /*scient*/"1.123e+00", /*flt*/"1.123", /*flex*/"1.123", /*hexa*/"0x1.1f9p+0" _c4emscripten_alt("0x1.1f98p+0"));
 
     test_rtoa(buf, f, &ftoa, 4, /*scient*/"1.1234e+00", /*flt*/"1.1234", /*flex*/"1.1234", /*hexa*/"0x1.1f98p+0");
     test_rtoa(buf, d, &dtoa, 4, /*scient*/"1.1234e+00", /*flt*/"1.1234", /*flex*/"1.1234", /*hexa*/"0x1.1f98p+0");
@@ -1226,8 +1238,8 @@ TEST_CASE("ftoa.basic")
     test_rtoa(buf, f, &ftoa, 2, /*scient*/"1.01e+00", /*flt*/"1.01", /*flex*/"1.01", /*hexa*/"0x1.03p+0");
     test_rtoa(buf, d, &dtoa, 2, /*scient*/"1.01e+00", /*flt*/"1.01", /*flex*/"1.01", /*hexa*/"0x1.03p+0");
 
-    test_rtoa(buf, f, &ftoa, 3, /*scient*/"1.012e+00", /*flt*/"1.012", /*flex*/"1.012", /*hexa*/"0x1.033p+0", /*hexa*/"0x1.032p+0");
-    test_rtoa(buf, d, &dtoa, 3, /*scient*/"1.012e+00", /*flt*/"1.012", /*flex*/"1.012", /*hexa*/"0x1.033p+0", /*hexa*/"0x1.032p+0");
+    test_rtoa(buf, f, &ftoa, 3, /*scient*/"1.012e+00", /*flt*/"1.012", /*flex*/"1.012", /*hexa*/"0x1.033p+0" _c4emscripten_alt2("0x1.032p+0", "0x1.0328p+0"));
+    test_rtoa(buf, d, &dtoa, 3, /*scient*/"1.012e+00", /*flt*/"1.012", /*flex*/"1.012", /*hexa*/"0x1.033p+0" _c4emscripten_alt2("0x1.032p+0", "0x1.0328p+0"));
 
     test_rtoa(buf, f, &ftoa, 4, /*scient*/"1.0123e+00", /*flt*/"1.0123", /*flex*/"1.0123", /*hexa*/"0x1.0329p+0");
     test_rtoa(buf, d, &dtoa, 4, /*scient*/"1.0123e+00", /*flt*/"1.0123", /*flex*/"1.0123", /*hexa*/"0x1.0329p+0");