Make Sage build with clang (3.7+) and make it the default on OS X

[ohanar]

Apple stopped supporting GCC in favour of clang. This is to track the status of building Sage with clang with the aim of making it the default compiler used on OS X and other OS where it is the default compiler.

---

Packages failing their tests

• eclib, reported upstream at https://github.com/JohnCremona/eclib/issues/19
• giac - patch to pass tests with gcc/pari-2.8+ causes failures with clang fixed by #24174
• ccache #22836

---

Supporting clang not only on OSX:

• porting to clang/FreeBSD -- #22679

Depends on #24579

CC: @nexttime @jpflori @dimpase @mkoeppe @isuruf

Component: build

Keywords: clang porting

Author: François Bissey

Branch: ac7816d

Reviewer: Jeroen Demeyer, John Palmieri, Dima Pasechnik

Issue created by migration from https://trac.sagemath.org/ticket/12426

[ohanar]

Description changed:

--- 
+++ 
@@ -1 +1,27 @@
 It is not unlikely that apple will eventually stop supporting gcc in favour of clang. This is a metaticket to track the status of building sage with clang.
+
+Spkgs needing fixing:
+* atlas
+* eclib
+* ecm
+* f2c -- #12427
+* flint -- #12429
+* flintqs -- #12428
+* gap
+* gfan -- #12440
+* givaro
+* lcal -- #12435, #12436, #12437
+* libfplll
+* libm4riec
+* linbox
+* matplotlib -- #12434
+* palp -- #12431
+* polybori
+* pynac
+* python -- #12430
+* ratpoints
+* sage
+* singular
+* symmetrica -- #12439
+* zlib -- #12432
+* zn_poly -- #12433

[ohanar]

Description changed:

--- 
+++ 
@@ -10,9 +10,9 @@
 * gap
 * gfan -- #12440
 * givaro
-* lcal -- #12435, #12436, #12437
+* lcalc -- #12435, #12436, #12437
 * libfplll
-* libm4riec
+* libm4rie
 * linbox
 * matplotlib -- #12434
 * palp -- #12431

[ohanar]

Description changed:

--- 
+++ 
@@ -21,7 +21,7 @@
 * python -- #12430
 * ratpoints
 * sage
-* singular
+* singular -- #12441, #12442
 * symmetrica -- #12439
 * zlib -- #12432
 * zn_poly -- #12433

[ohanar]

Description changed:

--- 
+++ 
@@ -20,7 +20,7 @@
 * pynac
 * python -- #12430
 * ratpoints
-* sage
+* sage -- #12443
 * singular -- #12441, #12442
 * symmetrica -- #12439
 * zlib -- #12432

[ohanar]

Description changed:

--- 
+++ 
@@ -4,22 +4,24 @@
 * atlas
 * eclib
 * ecm
+* gap
+* libfplll
+* libm4rie
+* linbox
+* polybori
+* pynac
+* ratpoints
+
+Needing review:
 * f2c -- #12427
 * flint -- #12429
 * flintqs -- #12428
-* gap
 * gfan -- #12440
-* givaro
+* givaro -- #12444
 * lcalc -- #12435, #12436, #12437
-* libfplll
-* libm4rie
-* linbox
 * matplotlib -- #12434
 * palp -- #12431
-* polybori
-* pynac
 * python -- #12430
-* ratpoints
 * sage -- #12443
 * singular -- #12441, #12442
 * symmetrica -- #12439

[ohanar]

Description changed:

--- 
+++ 
@@ -6,6 +6,7 @@
 * ecm
 * gap
 * libfplll
+* libgcrypt
 * libm4rie
 * linbox
 * polybori

[jdemeyer]

comment:7

I guess many of these tickets are useful for porting in general.

[ohanar]

Description changed:

--- 
+++ 
@@ -1,19 +1,18 @@
 It is not unlikely that apple will eventually stop supporting gcc in favour of clang. This is a metaticket to track the status of building sage with clang.

 Spkgs needing fixing:
-* atlas
-* eclib
-* ecm
-* gap
+* atlas -- probably impossible without Clint's help
+* eclib -- builds, but completely broken, no clue why
+* gap -- #7041
 * libfplll
-* libgcrypt
-* libm4rie
+* libgcrypt -- easy fix, but probably gone soon
 * linbox
 * polybori
 * pynac
-* ratpoints
+* ratpoints -- #12473

 Needing review:
+* ecm -- #12472
 * f2c -- #12427
 * flint -- #12429
 * flintqs -- #12428

[ohanar]

Description changed:

--- 
+++ 
@@ -9,12 +9,12 @@
 * linbox
 * polybori
 * pynac
-* ratpoints -- #12473
+* ratpoints -- #7038, #12473

 Needing review:
 * ecm -- #12472
-* f2c -- #12427
-* flint -- #12429
+* f2c -- #7027
+* flint -- #7024
 * flintqs -- #12428
 * gfan -- #12440
 * givaro -- #12444

[ohanar]

Description changed:

--- 
+++ 
@@ -20,7 +20,7 @@
 * givaro -- #12444
 * lcalc -- #12435, #12436, #12437
 * matplotlib -- #12434
-* palp -- #12431
+* palp -- #7071
 * python -- #12430
 * sage -- #12443
 * singular -- #12441, #12442

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

Description changed:

--- 
+++ 
@@ -1,4 +1,4 @@
-It is not unlikely that apple will eventually stop supporting gcc in favour of clang. This is a metaticket to track the status of building sage with clang.
+It is not unlikely that Apple will eventually stop supporting GCC in favour of clang. This is a metaticket to track the status of building Sage with clang.

 Spkgs needing fixing:
 * atlas -- probably impossible without Clint's help
@@ -9,21 +9,21 @@
 * linbox
 * polybori
 * pynac
-* ratpoints -- #7038, #12473
+* ratpoints -- #12682, #12473

 Needing review:
 * ecm -- #12472
-* f2c -- #7027
-* flint -- #7024
-* flintqs -- #12428
+* f2c -- #7027; package will be removed soon, see #12714
+* flint -- #7024; the issue is fixed in FLINT >= 1.5.2
+* flintqs -- #12428; package should be removed anyway, since FLINT includes a more recent version
 * gfan -- #12440
 * givaro -- #12444
-* lcalc -- #12435, #12436, #12437
+* lcalc -- #12681, #12436, #12437
 * matplotlib -- #12434
 * palp -- #7071
 * python -- #12430
 * sage -- #12443
-* singular -- #12441, #12442
+* singular -- #12680, #12442
 * symmetrica -- #12439
 * zlib -- #12432
 * zn_poly -- #12433

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

Description changed:

--- 
+++ 
@@ -1,29 +1,30 @@
 It is not unlikely that Apple will eventually stop supporting GCC in favour of clang. This is a metaticket to track the status of building Sage with clang.

 Spkgs needing fixing:
-* atlas -- probably impossible without Clint's help
+* ATLAS -- probably impossible without Clint's help
 * eclib -- builds, but completely broken, no clue why
-* gap -- #7041
+* GAP -- #7041
 * libfplll
 * libgcrypt -- easy fix, but probably gone soon
-* linbox
-* polybori
+* LinBox; related: #12762 (doesn't build with GCC 4.7.0 because code doesn't conform to C++11)
+* PolyBoRi; probably fixed by #12750 / #12754 / #12655
 * pynac
 * ratpoints -- #12682, #12473

 Needing review:
-* ecm -- #12472
+* GMP-ECM -- #12472
 * f2c -- #7027; package will be removed soon, see #12714
-* flint -- #7024; the issue is fixed in FLINT >= 1.5.2
-* flintqs -- #12428; package should be removed anyway, since FLINT includes a more recent version
-* gfan -- #12440
-* givaro -- #12444
-* lcalc -- #12681, #12436, #12437
+* FLINT -- #7024; the issue is fixed in FLINT >= 1.5.2
+* FLINTQS -- #12428; package should be removed anyway, since FLINT includes a more recent version
+* Gfan -- #12440; see also #12760 (doesn't build with GCC 4.7.0 / conform to C++11)
+* Givaro -- #12444; see also #12761 (doesn't build with GCC 4.7.0 / conform to C++11)
+* Lcalc -- #12681, #12436, #12437
 * matplotlib -- #12434
 * palp -- #7071
-* python -- #12430
-* sage -- #12443
-* singular -- #12680, #12442
-* symmetrica -- #12439
+* Python -- #12430
+* Sage -- #12443
+* Singular -- #12680, #12442
+* Symmetrica -- #12439
 * zlib -- #12432
 * zn_poly -- #12433
+

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

Changed keywords from clang to clang C++11 porting

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

comment:13

Added further ticket references.

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

Description changed:

--- 
+++ 
@@ -28,3 +28,7 @@
 * zlib -- #12432
 * zn_poly -- #12433

+---
+
+Probably related:  #14676
+

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

comment:15

Description presumably needs some updating...

[jdemeyer]

Description changed:

--- 
+++ 
@@ -7,6 +7,7 @@
 * libfplll
 * libgcrypt -- easy fix, but probably gone soon
 * LinBox; related: #12762 (doesn't build with GCC 4.7.0 because code doesn't conform to C++11)
+* MPIR: #13948
 * PolyBoRi; probably fixed by #12750 / #12754 / #12655
 * pynac
 * ratpoints -- #12682, #12473

[jdemeyer]

Description changed:

--- 
+++ 
@@ -1,33 +1,34 @@
 It is not unlikely that Apple will eventually stop supporting GCC in favour of clang. This is a metaticket to track the status of building Sage with clang.

-Spkgs needing fixing:
+---
+
+Packages failing to build with Clang:
+
+* matplotlib -- #12434
+* MPIR: #13948
+* ratpoints -- #12682, #12473
+
 * ATLAS -- probably impossible without Clint's help
 * eclib -- builds, but completely broken, no clue why
 * GAP -- #7041
 * libfplll
-* libgcrypt -- easy fix, but probably gone soon
 * LinBox; related: #12762 (doesn't build with GCC 4.7.0 because code doesn't conform to C++11)
-* MPIR: #13948
 * PolyBoRi; probably fixed by #12750 / #12754 / #12655
 * pynac
-* ratpoints -- #12682, #12473
-
-Needing review:
 * GMP-ECM -- #12472
-* f2c -- #7027; package will be removed soon, see #12714
-* FLINT -- #7024; the issue is fixed in FLINT >= 1.5.2
-* FLINTQS -- #12428; package should be removed anyway, since FLINT includes a more recent version
 * Gfan -- #12440; see also #12760 (doesn't build with GCC 4.7.0 / conform to C++11)
 * Givaro -- #12444; see also #12761 (doesn't build with GCC 4.7.0 / conform to C++11)
 * Lcalc -- #12681, #12436, #12437
-* matplotlib -- #12434
-* palp -- #7071
-* Python -- #12430
-* Sage -- #12443
 * Singular -- #12680, #12442
 * Symmetrica -- #12439
-* zlib -- #12432
-* zn_poly -- #12433
+
+---
+
+Spkgs ignoring CC/CXX environment variables:
+
+* FLINT -- #7024; the issue is fixed in FLINT >= 1.5.2
+* FLINTQS -- #12428; package should be removed anyway, since FLINT includes a more recent version
+* Sage -- #12443

 ---

[jdemeyer]

Description changed:

--- 
+++ 
@@ -4,22 +4,19 @@

 Packages failing to build with Clang:

+* ATLAS -- probably impossible without Clint's help
+* eclib -- builds, but completely broken, no clue why
+* libfplll
+* LinBox
+* pynac
+* GMP-ECM -- #12472
+* Gfan -- #12440
+* Givaro -- #12444
+* Lcalc -- #12436, #12437
 * matplotlib -- #12434
 * MPIR: #13948
 * ratpoints -- #12682, #12473
-
-* ATLAS -- probably impossible without Clint's help
-* eclib -- builds, but completely broken, no clue why
-* GAP -- #7041
-* libfplll
-* LinBox; related: #12762 (doesn't build with GCC 4.7.0 because code doesn't conform to C++11)
-* PolyBoRi; probably fixed by #12750 / #12754 / #12655
-* pynac
-* GMP-ECM -- #12472
-* Gfan -- #12440; see also #12760 (doesn't build with GCC 4.7.0 / conform to C++11)
-* Givaro -- #12444; see also #12761 (doesn't build with GCC 4.7.0 / conform to C++11)
-* Lcalc -- #12681, #12436, #12437
-* Singular -- #12680, #12442
+* Singular -- #12442
 * Symmetrica -- #12439

 ---

[jdemeyer]

Description changed:

--- 
+++ 
@@ -15,7 +15,7 @@
 * Lcalc -- #12436, #12437
 * matplotlib -- #12434
 * MPIR: #13948
-* ratpoints -- #12682, #12473
+* ratpoints -- #12473
 * Singular -- #12442
 * Symmetrica -- #12439

[jdemeyer]

Description changed:

--- 
+++ 
@@ -4,7 +4,7 @@

 Packages failing to build with Clang:

-* ATLAS -- probably impossible without Clint's help
+* ATLAS -- probably impossible without Clint's help, but might be a configure-only problem
 * eclib -- builds, but completely broken, no clue why
 * libfplll
 * LinBox

[jdemeyer]

Description changed:

--- 
+++ 
@@ -11,7 +11,6 @@
 * pynac
 * GMP-ECM -- #12472
 * Gfan -- #12440
-* Givaro -- #12444
 * Lcalc -- #12436, #12437
 * matplotlib -- #12434
 * MPIR: #13948

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

comment:22

eclib has been fixed upstream last year (IIRC) as well; currently re-checking whether our current version still builds...

[jdemeyer]

Description changed:

--- 
+++ 
@@ -6,7 +6,7 @@

 * ATLAS -- probably impossible without Clint's help, but might be a configure-only problem
 * eclib -- builds, but completely broken, no clue why
-* libfplll
+* libfplll -- fixed by #12835
 * LinBox
 * pynac
 * GMP-ECM -- #12472

[jdemeyer]

Description changed:

--- 
+++ 
@@ -7,7 +7,7 @@
 * ATLAS -- probably impossible without Clint's help, but might be a configure-only problem
 * eclib -- builds, but completely broken, no clue why
 * libfplll -- fixed by #12835
-* LinBox
+* LinBox -- assignment to `const` variable
 * pynac
 * GMP-ECM -- #12472
 * Gfan -- #12440

[jdemeyer]

Description changed:

--- 
+++ 
@@ -13,7 +13,8 @@
 * Gfan -- #12440
 * Lcalc -- #12436, #12437
 * matplotlib -- #12434
-* MPIR: #13948
+* MPIR -- #13948
+* R -- problematic due to mixture of C and Fortran code
 * ratpoints -- #12473
 * Singular -- #12442
 * Symmetrica -- #12439

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

comment:26

So on Darwin at least, we'd still have to build gfortran... (i.e., GCC with only the Fortran frontend and libgfortran.)

Or resort to f2c again... :-)

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

comment:27

Replying to @nexttime:

eclib has been fixed upstream last year (IIRC) as well; currently re-checking whether our current version still builds...

(Still) works for me, clang 3.2; make check (and some doctests) passed.

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

Description changed:

--- 
+++ 
@@ -5,11 +5,10 @@
 Packages failing to build with Clang:

 * ATLAS -- probably impossible without Clint's help, but might be a configure-only problem
-* eclib -- builds, but completely broken, no clue why
 * libfplll -- fixed by #12835
 * LinBox -- assignment to `const` variable
 * pynac
-* GMP-ECM -- #12472
+* GMP-ECM -- #12472 (seems to meanwhile work -- GMP-ECM 6.4.4, clang 3.2, **with** asm redc)
 * Gfan -- #12440
 * Lcalc -- #12436, #12437
 * matplotlib -- #12434

[jdemeyer]

Description changed:

--- 
+++ 
@@ -7,11 +7,9 @@
 * ATLAS -- probably impossible without Clint's help, but might be a configure-only problem
 * libfplll -- fixed by #12835
 * LinBox -- assignment to `const` variable
-* pynac
-* GMP-ECM -- #12472 (seems to meanwhile work -- GMP-ECM 6.4.4, clang 3.2, **with** asm redc)
+* pynac -- C++ template problem
 * Gfan -- #12440
 * Lcalc -- #12436, #12437
-* matplotlib -- #12434
 * MPIR -- #13948
 * R -- problematic due to mixture of C and Fortran code
 * ratpoints -- #12473

[jdemeyer]

Description changed:

--- 
+++ 
@@ -6,7 +6,7 @@

 * ATLAS -- probably impossible without Clint's help, but might be a configure-only problem
 * libfplll -- fixed by #12835
-* LinBox -- assignment to `const` variable
+* Givaro -- causes build failure for LinBox: assignment to `const` variable
 * pynac -- C++ template problem
 * Gfan -- #12440
 * Lcalc -- #12436, #12437

[jdemeyer]

Description changed:

--- 
+++ 
@@ -10,6 +10,7 @@
 * pynac -- C++ template problem
 * Gfan -- #12440
 * Lcalc -- #12436, #12437
+* LinBox -- C++ problem
 * MPIR -- #13948
 * R -- problematic due to mixture of C and Fortran code
 * ratpoints -- #12473

[jdemeyer]

Description changed:

--- 
+++ 
@@ -11,7 +11,7 @@
 * Gfan -- #12440
 * Lcalc -- #12436, #12437
 * LinBox -- C++ problem
-* MPIR -- #13948
+* MPIR -- #13948 (**of particular importance since it's needed to build GCC**)
 * R -- problematic due to mixture of C and Fortran code
 * ratpoints -- #12473
 * Singular -- #12442

[jdemeyer]

Description changed:

--- 
+++ 
@@ -4,7 +4,7 @@

 Packages failing to build with Clang:

-* ATLAS -- probably impossible without Clint's help, but might be a configure-only problem
+* ATLAS -- probably impossible without Clint's help. Also mixes Fortran and C code.
 * libfplll -- fixed by #12835
 * Givaro -- causes build failure for LinBox: assignment to `const` variable
 * pynac -- C++ template problem
@@ -22,10 +22,9 @@
 Spkgs ignoring CC/CXX environment variables:

 * FLINT -- #7024; the issue is fixed in FLINT >= 1.5.2
-* FLINTQS -- #12428; package should be removed anyway, since FLINT includes a more recent version
+* FLINTQS -- #12428
 * Sage -- #12443

 ---

-Probably related:  #14676
-
+Distantly related: #14676

[jdemeyer]

Description changed:

--- 
+++ 
@@ -5,7 +5,6 @@
 Packages failing to build with Clang:

 * ATLAS -- probably impossible without Clint's help. Also mixes Fortran and C code.
-* libfplll -- fixed by #12835
 * Givaro -- causes build failure for LinBox: assignment to `const` variable
 * pynac -- C++ template problem
 * Gfan -- #12440

[jdemeyer]

Description changed:

--- 
+++ 
@@ -10,7 +10,6 @@
 * Gfan -- #12440
 * Lcalc -- #12436, #12437
 * LinBox -- C++ problem
-* MPIR -- #13948 (**of particular importance since it's needed to build GCC**)
 * R -- problematic due to mixture of C and Fortran code
 * ratpoints -- #12473
 * Singular -- #12442

[jdemeyer]

Description changed:

--- 
+++ 
@@ -1,4 +1,4 @@
-It is not unlikely that Apple will eventually stop supporting GCC in favour of clang. This is a metaticket to track the status of building Sage with clang.
+Apple stopped supporting GCC in favour of clang. This is a metaticket to track the status of building Sage with clang.

 ---

[jdemeyer]

Description changed:

--- 
+++ 
@@ -19,7 +19,6 @@

 Spkgs ignoring CC/CXX environment variables:

-* FLINT -- #7024; the issue is fixed in FLINT >= 1.5.2
 * FLINTQS -- #12428
 * Sage -- #12443

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

comment:39

I think the upcoming GCC 4.9 will help us with Clang as well... ;-)

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

comment:41

Checking current status with Sage 6.2 and clang 3.4.1 ...

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

comment:42

Replying to @nexttime:

Checking current status with Sage 6.2 and clang 3.4.1 ...

The following package(s) may have failed to build:

package: atlas-3.10.1.20140210

package: boehm_gc-7.2d.p0

package: elliptic_curves-0.7

package: gf2x-1.1.p0

package: gfan-0.5.p0

package: lcalc-1.23.p12

package: libgap-4.7.4

package: polybori-0.8.3

package: pynac-0.3.2

package: ratpoints-2.1.3.p3

package: symmetrica-2.0.p9

[More to come.]

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

comment:43

Goes on with

package: eclib-20140128

package: linbox-1.3.2.p0

package: numpy-1.7.0

package: singular-3.1.6.p1

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

comment:44

Replying to @nexttime:

package: elliptic_curves-0.7

No idea what's going on there (ImportError: No module named _sqlite3); just recompiling sqlite with GCC doesn't help.

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

comment:45

Many of the hundreds of C++ errors are of the form

error: friend declaration specifying a default argument must be the only declaration
error: friend declaration specifying a default argument must be a definition

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

comment:46

Replying to @nexttime:

Replying to @nexttime:

package: elliptic_curves-0.7

No idea what's going on there (ImportError: No module named _sqlite3); just recompiling sqlite with GCC doesn't help.

Ah, that's a nice one: Python fails to build the _sqlite3 extension because of

clang: error: unknown argument: '-R/path/to/sage-6.2-clang-3.4.1/local/lib'

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

Description changed:

--- 
+++ 
@@ -2,14 +2,18 @@

 ---

-Packages failing to build with Clang:
+Packages failing to build (or causing other packages to fail to build) with Clang:

 * ATLAS -- probably impossible without Clint's help. Also mixes Fortran and C code.
+* eclib -- default parameters in `friend` declarations (?)
 * Givaro -- causes build failure for LinBox: assignment to `const` variable
-* pynac -- C++ template problem
+* gf2x -- uses GCC builtins clang doesn't support (apparently)
 * Gfan -- #12440
 * Lcalc -- #12436, #12437
+* libGAP -- some preprocessor / macro weirdness?
 * LinBox -- C++ problem
+* pynac -- C++ template problem
+* Python -- fails to build `_sqlite3` (uses `-R<path>` instead of `-Wl,-rpath,<path>`)
 * R -- problematic due to mixture of C and Fortran code
 * ratpoints -- #12473
 * Singular -- #12442

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

comment:48

Linbox (without the errors caused by Givaro headers):

../../linbox/solutions/solve.h:314:9: error: no matching conversion for functional-style cast from 'const char [19]' to 'LinBox::LinBoxFailure'

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

comment:49

NumPy fails to build due to (at least) various problems with xmmintrin.h.

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

Description changed:

--- 
+++ 
@@ -12,6 +12,7 @@
 * Lcalc -- #12436, #12437
 * libGAP -- some preprocessor / macro weirdness?
 * LinBox -- C++ problem
+* NumPy -- various errors with (GCC's) `xmmintrin.h` (including unsupported builtins)
 * pynac -- C++ template problem
 * Python -- fails to build `_sqlite3` (uses `-R<path>` instead of `-Wl,-rpath,<path>`)
 * R -- problematic due to mixture of C and Fortran code

[83660e46-0051-498b-a8c1-f7a7bd232b5a]

Description changed:

--- 
+++ 
@@ -8,11 +8,11 @@
 * eclib -- default parameters in `friend` declarations (?)
 * Givaro -- causes build failure for LinBox: assignment to `const` variable
 * gf2x -- uses GCC builtins clang doesn't support (apparently)
-* Gfan -- #12440
+* Gfan (0.5) -- #12440
 * Lcalc -- #12436, #12437
 * libGAP -- some preprocessor / macro weirdness?
 * LinBox -- C++ problem
-* NumPy -- various errors with (GCC's) `xmmintrin.h` (including unsupported builtins)
+* NumPy (1.7.0) -- various errors with (GCC's) `xmmintrin.h` (including unsupported builtins)
 * pynac -- C++ template problem
 * Python -- fails to build `_sqlite3` (uses `-R<path>` instead of `-Wl,-rpath,<path>`)
 * R -- problematic due to mixture of C and Fortran code