[Python-modules-commits] r14065 - in packages/numpy/trunk/debian (3 files)
morph at users.alioth.debian.org
morph at users.alioth.debian.org
Tue Jul 27 15:39:02 UTC 2010
Date: Tuesday, July 27, 2010 @ 15:38:51
Author: morph
Revision: 14065
* debian/patches/changeset_r8526.diff
- cherry-picked from upstream SVN patches to define the quad prec little
endian double on Alpha, fixing the FTBFS on that arch; thanks to David
Cournapeau for the patch and huge support! Closes: #590481
Added:
packages/numpy/trunk/debian/patches/changeset_r8526.diff
Modified:
packages/numpy/trunk/debian/changelog
packages/numpy/trunk/debian/patches/series
Modified: packages/numpy/trunk/debian/changelog
===================================================================
--- packages/numpy/trunk/debian/changelog 2010-07-27 02:47:54 UTC (rev 14064)
+++ packages/numpy/trunk/debian/changelog 2010-07-27 15:38:51 UTC (rev 14065)
@@ -7,8 +7,12 @@
- added debhelper script dh_numpy that using .../dist/python-numpy adds to
python:Depends the correct versioned depends on python-numpy; thansk to
Piotr Ożarowski for providing the script
+ * debian/patches/changeset_r8526.diff
+ - cherry-picked from upstream SVN patches to define the quad prec little
+ endian double on Alpha, fixing the FTBFS on that arch; thanks to David
+ Cournapeau for the patch and huge support! Closes: #590481
- -- Sandro Tosi <morph at debian.org> Tue, 27 Jul 2010 00:42:07 +0200
+ -- Sandro Tosi <morph at debian.org> Tue, 27 Jul 2010 17:28:22 +0200
python-numpy (1:1.4.1-3) unstable; urgency=low
Added: packages/numpy/trunk/debian/patches/changeset_r8526.diff
===================================================================
--- packages/numpy/trunk/debian/patches/changeset_r8526.diff (rev 0)
+++ packages/numpy/trunk/debian/patches/changeset_r8526.diff 2010-07-27 15:38:51 UTC (rev 14065)
@@ -0,0 +1,44 @@
+Index: /trunk/numpy/core/src/npymath/npy_math_private.h
+===================================================================
+--- python-numpy-1.4.1.orig/numpy/core/src/npymath/npy_math_private.h (revision 8510)
++++ python-numpy-1.4.1/numpy/core/src/npymath/npy_math_private.h (revision 8526)
+@@ -362,4 +362,39 @@
+ typedef npy_uint64 ldouble_exp_t;
+ typedef npy_uint32 ldouble_sign_t;
++#elif defined(HAVE_LDOUBLE_IEEE_QUAD_LE)
++ /*
++ * IEEE quad precision, Little Endian. Bit representation is
++ * | s |eeeeeeeeeee|mmmmmmmm................mmmmmmm|
++ * |1 bit| 15 bits | 112 bits |
++ * | a[1] | a[0] |
++ */
++ typedef npy_uint64 IEEEl2bitsrep_part;
++
++ union IEEEl2bitsrep {
++ npy_longdouble e;
++ IEEEl2bitsrep_part a[2];
++ };
++
++ #define LDBL_MANL_INDEX 0
++ #define LDBL_MANL_MASK 0xFFFFFFFFFFFFFFFF
++ #define LDBL_MANL_SHIFT 0
++
++ #define LDBL_MANH_INDEX 1
++ #define LDBL_MANH_MASK 0x0000FFFFFFFFFFFF
++ #define LDBL_MANH_SHIFT 0
++
++ #define LDBL_EXP_INDEX 1
++ #define LDBL_EXP_MASK 0x7FFF000000000000
++ #define LDBL_EXP_SHIFT 48
++
++ #define LDBL_SIGN_INDEX 1
++ #define LDBL_SIGN_MASK 0x8000000000000000
++ #define LDBL_SIGN_SHIFT 63
++
++ #define LDBL_NBIT 0
++
++ typedef npy_uint64 ldouble_man_t;
++ typedef npy_uint64 ldouble_exp_t;
++ typedef npy_uint32 ldouble_sign_t;
+ #endif
+
Modified: packages/numpy/trunk/debian/patches/series
===================================================================
--- packages/numpy/trunk/debian/patches/series 2010-07-27 02:47:54 UTC (rev 14064)
+++ packages/numpy/trunk/debian/patches/series 2010-07-27 15:38:51 UTC (rev 14065)
@@ -6,3 +6,4 @@
10_use_local_python.org_object.inv_sphinx.diff
changeset_r8510.diff
changeset_r8511.diff
+changeset_r8526.diff
More information about the Python-modules-commits
mailing list