simdutf 6.1.0

[BrewTestBot]

Created by brew bump

---

Created with brew bump-formula-pr.

• resource blocks have been checked for updates.

release notes

## What's Changed
* Optimize icelake latin1 to utf32 for small inputs using masked AVX512 instructions by @tantei3 in https://github.com/simdutf/simdutf/pull/645
* Add CMake toolchain files for various archs by @WojciechMula in https://github.com/simdutf/simdutf/pull/643
* add test for issue 633 by @pauldreik in https://github.com/simdutf/simdutf/pull/649
* Extract ASCII validation into a separate file by @WojciechMula in https://github.com/simdutf/simdutf/pull/646
* Sort out including scalar/* files by @WojciechMula in https://github.com/simdutf/simdutf/pull/650
* Simplify calculating length latin1 <=> UTF16/UTF32 by @WojciechMula in https://github.com/simdutf/simdutf/pull/652
* Refactor amalgamate.py by @WojciechMula in https://github.com/simdutf/simdutf/pull/653
* Fix fuzz test failure in lsx and lasx by @Xiao-Tao in https://github.com/simdutf/simdutf/pull/654
* Enable selective amalgamation by @WojciechMula in https://github.com/simdutf/simdutf/pull/637
* Optimized icelake AVX512 for utf32 to utf16 conversion by @tantei3 in https://github.com/simdutf/simdutf/pull/648
Full Changelog: simdutf/simdutf@v6.0.3...v6.1.0

[daeho-ro]

cd /tmp/simdutf-20250123-5191-lh9adf/simdutf-6.1.0/build/tests/reference && /opt/homebrew/Library/Homebrew/shims/mac/super/clang++  -I/tmp/simdutf-20250123-5191-lh9adf/simdutf-6.1.0/include -O3 -DNDEBUG -std=c++11 -arch arm64 -isysroot /Library/Developer/CommandLineTools/SDKs/MacOSX15.sdk -fPIC -MD -MT tests/reference/CMakeFiles/simdutf_tests_reference.dir/encode_latin1.cpp.o -MF CMakeFiles/simdutf_tests_reference.dir/encode_latin1.cpp.o.d -o CMakeFiles/simdutf_tests_reference.dir/encode_latin1.cpp.o -c /tmp/simdutf-20250123-5191-lh9adf/simdutf-6.1.0/tests/reference/encode_latin1.cpp
  fatal: not a git repository (or any of the parent directories): .git
  using env variable AMALGAMATE_SOURCE_PATH=/tmp/simdutf-20250123-5191-lh9adf/simdutf-6.1.0/src
  using env variable AMALGAMATE_OUTPUT_PATH=/tmp/simdutf-20250123-5191-lh9adf/simdutf-6.1.0/build/singleheader
  SCRIPTPATH=/tmp/simdutf-[202](https://github.com/Homebrew/homebrew-core/actions/runs/12939781735/job/36092978643?pr=205349#step:3:203)50123-5191-lh9adf/simdutf-6.1.0/singleheader PROJECTPATH=/tmp/simdutf-20250123-5191-lh9adf/simdutf-6.1.0
  We are about to amalgamate all simdutf files into one source file.
  See https://www.sqlite.org/amalgamation.html and https://en.wikipedia.org/wiki/Single_Compilation_Unit for rationale.
  Traceback (most recent call last):
    File "/tmp/simdutf-20250123-5191-lh9adf/simdutf-6.1.0/singleheader/amalgamate.py", line 478, in <module>
      main()
    File "/tmp/simdutf-20250123-5191-lh9adf/simdutf-6.1.0/singleheader/amalgamate.py", line 40, in main
      context.timestamp        = get_timestamp()
    File "/tmp/simdutf-20250123-5191-lh9adf/simdutf-6.1.0/singleheader/amalgamate.py", line 236, in get_timestamp
      raise ValueError(f"non-zero exit code {ret.returncode}")
  ValueError: non-zero exit code 128
  make[2]: *** [singleheader/simdutf.cpp] Error 1
  make[1]: *** [singleheader/CMakeFiles/simdutf-singleheader-files.dir/all] Error 2
  make[1]: *** Waiting for unfinished jobs....

[botantony]

This patch may fix the issue:

diff --git a/singleheader/amalgamate.py b/singleheader/amalgamate.py
index 5b18ea16..d600647e 100755
--- a/singleheader/amalgamate.py
+++ b/singleheader/amalgamate.py
@@ -7,6 +7,7 @@
 import os.path
 import subprocess
 import os
+import pathlib
 import re
 import shutil
 import datetime
@@ -229,15 +230,24 @@ def get_timestamp():
     # Forcing it to be UTC is difficult, because it needs to be portable
     # between gnu date and busybox date.
     try:
-        ret = subprocess.run(['git', 'show', '-s', '--format=%ci', 'HEAD'],
-                             stdout=subprocess.PIPE)
+        # avoid git going outside simdutf, which could happen when
+        # unpacking a release tarball inside a subdirectory of an unrelated
+        # git repository. that would lead to picking up the timestamp of the
+        # unrelated git repository.
+        parent = pathlib.Path(SCRIPTPATH).absolute().parent
+        GIT_CEILING_DIRECTORIES = str(parent)
+        ret = subprocess.run(['git', '-C', SCRIPTPATH, 'show', '-s', '--format=%ci', 'HEAD'],
+                             stdout=subprocess.PIPE,
+                             env=dict(os.environ, GIT_CEILING_DIRECTORIES=GIT_CEILING_DIRECTORIES))
 
         if ret.returncode != 0:
-            raise ValueError(f"non-zero exit code {ret.returncode}")
+            print(f"git called resulted in non-zero exit code {ret.returncode}")
+            print("timestamp based on current time")
+            return str(datetime.datetime.now())
 
         return ret.stdout.decode('utf-8').strip()
     except (UnicodeDecodeError, FileNotFoundError):
-        print("git not found, timestamp based on current time")
+        print("UnicodeDecodeError or FileNotFoundError, timestamp based on current time")
         return str(datetime.datetime.now())

[chenrui333]

@botantony yeah, that is what I was thinking yesterday :) can you filed an upstream PR on it? thanks! 🙏

[botantony]

Unfortunately I cannot do upstream PR due to lack of permissions (I am a contributor, not organization member).
Here's the final formula with applied patch:

class Simdutf < Formula
  desc "Unicode conversion routines, fast"
  homepage "https://github.com/simdutf/simdutf"
  url "https://github.com/simdutf/simdutf/archive/refs/tags/v6.1.0.tar.gz"
  sha256 "ef2903a7f085090c58f3acfa93a62733ae92a3f9b1d50800edec77a6816d7d67"
  license any_of: ["Apache-2.0", "MIT"]
  head "https://github.com/simdutf/simdutf.git", branch: "master"

  livecheck do
    url :stable
    regex(/^v?(\d+(?:\.\d+)+)$/i)
  end

  bottle do
    sha256 cellar: :any, arm64_sequoia: "e4e6f6c35a03b5298d8fc5e903f4f052f6a836d714182f8cdf8a6a249fd95a3c"
    sha256 cellar: :any, arm64_sonoma:  "78d81462239a71c50179a65e9ae63bda64ab170e1a0f54e97db4f765ab2e4265"
    sha256 cellar: :any, arm64_ventura: "0847766dd42149d771e60db322a329e54bb8b65c8feb1031b0a6cd35bca79b3a"
    sha256 cellar: :any, sonoma:        "1f6835d989644af3e6db2ac1c785b37c2e14bb8ad539d78e773baac200c8bc3f"
    sha256 cellar: :any, ventura:       "680f6afecf5b1b5e35330e5351e6eb34d87486e961f52f66b8fa5cf8cab9e056"
  end

  depends_on "cmake" => :build
  depends_on "icu4c@76"
  depends_on macos: :catalina

  uses_from_macos "python" => :build

  patch :DATA

  # VERSION=#{version} && curl -s https://raw.githubusercontent.com/simdutf/simdutf/v$VERSION/benchmarks/base64/CMakeLists.txt | grep -C 1 'VERSION'
  resource "base64" do
    url "https://github.com/aklomp/base64/archive/refs/tags/v0.5.2.tar.gz"
    sha256 "723a0f9f4cf44cf79e97bcc315ec8f85e52eb104c8882942c3f2fba95acc080d"
  end

  def install
    (buildpath/"base64").install resource("base64")

    args = %W[
      -DBUILD_SHARED_LIBS=ON
      -DCMAKE_INSTALL_RPATH=#{rpath}
      -DCMAKE_BUILD_WITH_INSTALL_RPATH=ON
      -DFETCHCONTENT_SOURCE_DIR_BASE64=#{buildpath}/base64
      -DPython3_EXECUTABLE=#{which("python3")}
      -DSIMDUTF_BENCHMARKS=ON
    ]
    system "cmake", "-S", ".", "-B", "build", *args, *std_cmake_args
    system "cmake", "--build", "build"
    system "cmake", "--install", "build"
    bin.install "build/benchmarks/benchmark" => "sutf-benchmark"
  end

  test do
    system bin/"sutf-benchmark", "--random-utf8", "1024", "-I", "20"
  end
end

__END__
diff --git a/singleheader/amalgamate.py b/singleheader/amalgamate.py
index 5b18ea16..d600647e 100755
--- a/singleheader/amalgamate.py
+++ b/singleheader/amalgamate.py
@@ -7,6 +7,7 @@
 import os.path
 import subprocess
 import os
+import pathlib
 import re
 import shutil
 import datetime
@@ -229,15 +230,24 @@ def get_timestamp():
     # Forcing it to be UTC is difficult, because it needs to be portable
     # between gnu date and busybox date.
     try:
-        ret = subprocess.run(['git', 'show', '-s', '--format=%ci', 'HEAD'],
-                             stdout=subprocess.PIPE)
+        # avoid git going outside simdutf, which could happen when
+        # unpacking a release tarball inside a subdirectory of an unrelated
+        # git repository. that would lead to picking up the timestamp of the
+        # unrelated git repository.
+        parent = pathlib.Path(SCRIPTPATH).absolute().parent
+        GIT_CEILING_DIRECTORIES = str(parent)
+        ret = subprocess.run(['git', '-C', SCRIPTPATH, 'show', '-s', '--format=%ci', 'HEAD'],
+                             stdout=subprocess.PIPE,
+                             env=dict(os.environ, GIT_CEILING_DIRECTORIES=GIT_CEILING_DIRECTORIES))

         if ret.returncode != 0:
-            raise ValueError(f"non-zero exit code {ret.returncode}")
+            print(f"git called resulted in non-zero exit code {ret.returncode}")
+            print("timestamp based on current time")
+            return str(datetime.datetime.now())

         return ret.stdout.decode('utf-8').strip()
     except (UnicodeDecodeError, FileNotFoundError):
-        print("git not found, timestamp based on current time")
+        print("UnicodeDecodeError or FileNotFoundError, timestamp based on current time")
         return str(datetime.datetime.now())

[botantony]

forgot to ping @chenrui333

[chenrui333]

oh, there is already a PR on this, simdutf/simdutf#657

[github-actions]

🤖 An automated task has requested bottles to be published to this PR.