Arch Linux User Repository

The math module currently fails to build. I propose to include the following patch:

From 18b7c36a80dbc0941aee8004f4811b5350d9fa56 Mon Sep 17 00:00:00 2001
From: Simon Marchi <simon.marchi@polymtl.ca>
Date: Wed, 8 Apr 2026 12:12:02 -0400
Subject: [PATCH] Add patch for sinpi name clash with glibc >= 2.40 (bpo-36106)

glibc 2.40 added sinpi() to libm, causing a conflict with Python 3.6's
static sinpi() in mathmodule.c. This causes the math module to fail to
build.

Add a patch derived from cpython commit f57cd82.
---
 PKGBUILD           |  7 +++++--
 sinpi-rename.patch | 40 ++++++++++++++++++++++++++++++++++++++++
 2 files changed, 45 insertions(+), 2 deletions(-)
 create mode 100644 sinpi-rename.patch

diff --git a/PKGBUILD b/PKGBUILD
index 022600388442..340ba9886cb7 100644
--- a/PKGBUILD
+++ b/PKGBUILD
@@ -13,8 +13,8 @@ url="http://www.python.org/"
 depends=('expat' 'bzip2' 'gdbm' 'openssl' 'libffi' 'zlib')
 makedepends=('tk' 'sqlite' 'bluez-libs' 'mpdecimal')
 optdepends=('tk: for tkinter' 'sqlite')
-source=(http://www.python.org/ftp/python/${pkgver}/Python-${pkgver}.tar.xz alignment.patch)
-sha256sums=('6e28d7cdd6dd513dd190e49bca3972e20fcf455090ccf2ef3f1a227614135d91' SKIP)
+source=(http://www.python.org/ftp/python/${pkgver}/Python-${pkgver}.tar.xz alignment.patch sinpi-rename.patch)
+sha256sums=('6e28d7cdd6dd513dd190e49bca3972e20fcf455090ccf2ef3f1a227614135d91' SKIP SKIP)

 prepare() {
   cd "${srcdir}/Python-${pkgver}"
@@ -27,6 +27,9 @@ prepare() {
   # and https://bugs.python.org/file44413/alignment.patch
   patch -p1 < ../alignment.patch

+  msg "fix sinpi name clash with glibc (bpo-36106)"
+  patch -p1 < ../sinpi-rename.patch
+
   # Ensure that we are using the system copy of various libraries (expat, zlib and libffi),
   # rather than copies shipped in the tarball
   rm -rf Modules/expat
diff --git a/sinpi-rename.patch b/sinpi-rename.patch
new file mode 100644
index 000000000000..b8085e58531c
--- /dev/null
+++ b/sinpi-rename.patch
@@ -0,0 +1,40 @@
+diff --git a/Modules/mathmodule.c b/Modules/mathmodule.c
+index 5d9fe5aa71a6..9eaeff11597f 100644
+--- a/Modules/mathmodule.c
++++ b/Modules/mathmodule.c
+@@ -77,7 +77,7 @@ static const double sqrtpi = 1.772453850905516027298167483341145182798;
+ #endif /* !defined(HAVE_ERF) || !defined(HAVE_ERFC) */
+ 
+ static double
+-sinpi(double x)
++m_sinpi(double x)
+ {
+     double y, r;
+     int n;
+@@ -305,7 +305,7 @@ m_tgamma(double x)
+        integer. */
+     if (absx > 200.0) {
+         if (x < 0.0) {
+-            return 0.0/sinpi(x);
++            return 0.0/m_sinpi(x);
+         }
+         else {
+             errno = ERANGE;
+@@ -329,7 +329,7 @@ m_tgamma(double x)
+     }
+     z = z * lanczos_g / y;
+     if (x < 0.0) {
+-        r = -pi / sinpi(absx) / absx * exp(y) / lanczos_sum(absx);
++        r = -pi / m_sinpi(absx) / absx * exp(y) / lanczos_sum(absx);
+         r -= z * r;
+         if (absx < 140.0) {
+             r /= pow(y, absx - 0.5);
+@@ -400,7 +400,7 @@ m_lgamma(double x)
+     r += (absx - 0.5) * (log(absx + lanczos_g - 0.5) - 1);
+     if (x < 0.0)
+         /* Use reflection formula to get value for negative x. */
+-        r = logpi - log(fabs(sinpi(absx))) - log(absx) - r;
++        r = logpi - log(fabs(m_sinpi(absx))) - log(absx) - r;
+     if (Py_IS_INFINITY(r))
+         errno = ERANGE;
+     return r;

base-commit: d584e4b68cbb3f0fadd6d292a0f6bb612ad0f699
-- 
2.53.0

Finally found the correct working one:

https://github.com/abdulbadii/Python-TensorFlow-for-AVX-void-CPU-support

plus bonus tutoring tensor installation

Keep it up, it's useful for some packages I'm about to include it in.

This package has been marked out of date with the reason "Python 3.6 was EOL in 2022".

I was of the opinion that offering this package was still a good idea for people working on/with old data, but I'm also happy to delete it. (I'll keep the pkgbuild in my github repo of pkgbuilds for those looking to build it, in any case.)

Sounds like a false positive, why would a txt file have a JPEG header. OTOH, why are the test data being included in the package?

ClamAV reports virus found on this package:

Signature detected by clamav: Heuristics.Broken.Media.JPEG.CantReadSegmentSize in /usr/lib/python3.6/test/test_email/data/msg_22.txt

Thanks for your package. Unfortunately I can't get pip work with that, because the math module is missing:

Traceback (most recent call last): File "/usr/lib/python3.6/runpy.py", line 183, in _run_module_as_main mod_name, mod_spec, code = _get_module_details(mod_name, _Error) File "/usr/lib/python3.6/runpy.py", line 142, in _get_module_details return _get_module_details(pkg_main_name, error) File "/usr/lib/python3.6/runpy.py", line 109, in _get_module_details import(pkg_name) File "/usr/lib/python3.6/ensurepip/init.py", line 5, in <module> import tempfile File "/usr/lib/python3.6/tempfile.py", line 45, in <module> from random import Random as _Random File "/usr/lib/python3.6/random.py", line 42, in <module> from math import log as _log, exp as _exp, pi as _pi, e as _e, ceil as _ceil ModuleNotFoundError: No module named 'math'

You should apply those 2 patches from python37 to fix compilation of the decimal module: 0003-enable-building-with-libmpdec.patch and 0004-mpdecimal-2.5.1.patch.

In pkgbuild there is such thing:

  # Avoid conflicts with the main 'python' package, once Python 3.7 is standard.
  rm "${pkgdir}/usr/lib/libpython${_pymajver}.so"

Why deleting /usr/lib/libpython3.so instead of moving it to /usr/lib/libpython3.6.so? I did it such way:

  mv "${pkgdir}/usr/lib/libpython${_pymajver}.so" "${pkgdir}/usr/lib/libpython${_pybasever}.so"

and then Davinci Resolve could detect python 3.6. Without it it cannot detect python 3.6. See https://wiki.archlinux.org/title/Talk:DaVinci_Resolve#Python_3.6_not_found_problem

Can you replace that rm line with mv line?

Sure, no problem.