Upgrade Singular

[jdemeyer]

Upgrade Singular to ​version 3-1-5 (released 12 July 2012).

Upstream bugs discovered:

• ​#437: negate() should be NTL::negate() (upstream needs testers)
• ​#438: Singular configured --without-debug doesn't build ("fixed" by removing the option)
• ​#439: factorize() returns wrong constant (fixed upstream, but not in a stable release)
• ​#440: Factoring over a finite field returns reducible factor (fixed upstream, but not in a stable release)
• ​#441: Factoring over a finite field sometimes returns wrong result (fixed upstream, but not in a stable release)
• ​#443: configure should check for nanosleep in -lrt (no feedback yet)

This new version fixes #13129, #12918, #12928, #12846, #12089, #13344.

Apply 13237_singular_3_1_5.patch​

New spkg: ​http://boxen.math.washington.edu/home/jdemeyer/spkg/singular-3-1-5.p0.spkg

singular-3-1-5.p0 (Jeroen Demeyer, Alexander Dreyer, 6 August 2012)

• Trac #13237: Upgrade to version 3-1-5.
• Removed patches which are now upstreamed:
  • patches/Singular.Makefile.in.shared.patch
  • patches/Singular.configure.patch
  • patches/factory.GNUmakefile.in.patch
  • patches/factory_configure
  • patches/make_parallel.patch
  • patches/os_x_ppc.patch
• In spkg-install, simplify apply_patches(), automatically apply all patches in patches/*.patch, move conditional patches to patches/conditional
• Rename some patches such that they all have the extension '.patch'
• Put the two patches for SAGE_DEBUG (Singular.Makefile.in.debug.patch and kernel.Makefile.in.debug.patch) into one file conditional/sage_debug.patch.
• Add several patches (see above):
  • NTL_negate.patch
  • singular_trac_439.patch
  • singular_trac_440.patch
  • singular_trac_441.patch
  • sage_trac_12089.patch
  • slibdir.patch
• Don't create the LIB->lib symlink, which is no longer needed.
• In spkg-install, remove distclean() step; merge clean_headers() and part of the old distclean() into remove_old_version().
• Only unset LD on Darwin.
• Remove the unsetting of TMPDIR (bug fixed upstream).
• Do not override user-set CFLAGS and CXXFLAGS.
• Echo all error messages to stderr instead of stdout.
• Fix various ./configure options and remove some unsupported options.
• Always configure --with-debug, as --without-debug doesn't work, see ​http://www.singular.uni-kl.de:8002/trac/ticket/438
• Don't create sage_singular symlink, which wasn't used anyway.
• Replace the $SAGE_LOCAL/bin/Singular script by a symlink.
• Remove workaround for GCC-4.0.x on Darwin (obsolete by the GCC spkg).
• Fix formatting of spkg-install (consistent indentation, no TABs).

[leif]

checking for FLINT >= 2.3... not found
configure: WARNING: Unable to find FLINT (which is strongly recommended) on your machine: please use --with-flint=PATH_TO_DIR_CONTAINING_LIB_AND_INCLUDE (see also ./configure --help if you do not understand what we are talking about)

:P

And lots of other warnings...

[leif]

Replying to leif:

And lots of other warnings...

These are the ones I get from configure and config.status:

configure: warning: Building of doc might fail. Need Texinfo
configure: WARNING: unrecognized options: --with-apint, --with-ntl, --without-MP, --without-lex, --without-bison, --without-Boost, --enable-Singular, --enable-factory, --enable-libfac, --enable-IntegerProgramming, --disable-doc, --with-malloc, --disable-debug, --enable-omalloc, --with-external-config_h, --with-track-custom, --enable-Plural, --with-factory, --with-libfac
configure: WARNING: Unable to find FLINT (which is strongly recommended) on your machine: please use --with-flint=PATH_TO_DIR_CONTAINING_LIB_AND_INCLUDE (see also ./configure --help if you do not understand what we are talking about)
config.status: WARNING:  'GNUmakefile.in' seems to ignore the --datarootdir setting
configure: WARNING: unrecognized options: --with-apint, --with-ntl, --without-MP, --without-lex, --without-bison, --without-Boost, --enable-Singular, --enable-factory, --enable-libfac, --enable-IntegerProgramming, --disable-doc, --with-malloc, --disable-debug, --enable-omalloc, --with-external-config_h, --with-track-custom, --enable-Plural, --with-factory, --with-libfac
configure: warning: factory.h not found! Install factory before building libfac!
configure: WARNING: unrecognized options: --with-gmp, --with-ntl, --without-bison, --enable-gmp, --enable-Singular, --enable-factory, --enable-libfac, --enable-IntegerProgramming, --disable-doc, --with-malloc, --disable-debug, --enable-omalloc, --with-external-config_h, --with-track-custom, --enable-Plural, --with-Singular
configure: WARNING: building without lex -- make might fail
configure: WARNING: building without python support
configure: WARNING: Unable to find FLINT (which is strongly recommended) on your machine: please use --with-flint=PATH_TO_DIR_CONTAINING_LIB_AND_INCLUDE (see also ./configure --help if you do not understand what we are talking about)
configure: WARNING: unrecognized options: --with-gmp, --with-ntl, --without-bison, --enable-gmp, --enable-Singular, --enable-factory, --enable-libfac, --enable-IntegerProgramming, --disable-doc, --with-malloc, --disable-debug, --enable-omalloc, --with-external-config_h, --with-track-custom, --enable-Plural, --with-Singular
configure: warning: Building of doc might fail. Need Texinfo
configure: WARNING: unrecognized options: --with-apint, --with-ntl, --without-MP, --without-lex, --without-bison, --without-Boost, --enable-Singular, --enable-factory, --enable-libfac, --enable-IntegerProgramming, --disable-doc, --with-malloc, --disable-debug, --enable-omalloc, --with-external-config_h, --with-track-custom, --enable-Plural, --with-factory, --with-libfac
configure: WARNING: Unable to find FLINT (which is strongly recommended) on your machine: please use --with-flint=PATH_TO_DIR_CONTAINING_LIB_AND_INCLUDE (see also ./configure --help if you do not understand what we are talking about)
config.status: WARNING:  'GNUmakefile.in' seems to ignore the --datarootdir setting
configure: WARNING: unrecognized options: --with-apint, --with-ntl, --without-MP, --without-lex, --without-bison, --without-Boost, --enable-Singular, --enable-factory, --enable-libfac, --enable-IntegerProgramming, --disable-doc, --with-malloc, --disable-debug, --enable-omalloc, --with-external-config_h, --with-track-custom, --enable-Plural, --with-factory, --with-libfac
configure: warning: factory.h not found! Install factory before building libfac!
configure: WARNING: unrecognized options: --with-gmp, --with-ntl, --without-bison, --enable-gmp, --enable-Singular, --enable-factory, --enable-libfac, --enable-IntegerProgramming, --disable-doc, --with-malloc, --disable-debug, --enable-omalloc, --with-external-config_h, --with-track-custom, --enable-Plural, --with-Singular
configure: WARNING: building without lex -- make might fail
configure: WARNING: building without python support
configure: WARNING: Unable to find FLINT (which is strongly recommended) on your machine: please use --with-flint=PATH_TO_DIR_CONTAINING_LIB_AND_INCLUDE (see also ./configure --help if you do not understand what we are talking about)
configure: WARNING: unrecognized options: --with-gmp, --with-ntl, --without-bison, --enable-gmp, --enable-Singular, --enable-factory, --enable-libfac, --enable-IntegerProgramming, --disable-doc, --with-malloc, --disable-debug, --enable-omalloc, --with-external-config_h, --with-track-custom, --enable-Plural, --with-Singular
configure: WARNING: unrecognized options: --with-apint, --with-ntl, --disable-debug
configure: WARNING: Unable to find FLINT (which is strongly recommended) on your machine: please use --with-flint=PATH_TO_DIR_CONTAINING_LIB_AND_INCLUDE (see also ./configure --help if you do not understand what we are talking about)
config.status: WARNING:  'GNUmakefile.in' seems to ignore the --datarootdir setting
configure: WARNING: unrecognized options: --with-apint, --with-ntl, --disable-debug

The unrecognized options should be addressed I think.

No idea how to interpret "factory.h not found! Install factory before building libfac!", as the build succeeded.

"Building without python support" can safely be ignored I guess, as we use our own Cython wrappers.

[jdemeyer]

Replying to leif:

The unrecognized options should be addressed I think.

I fixed the fixable warnings. Many of them are caused by recursive calling of configure, for example, the top-level configure calls omalloc/configure. Many options are supported by some (but not all) configure scripts. These will give warnings and I don't see how to avoid these.

[leif]

I assume that this is work in progress (i.e., you haven't tried to fix doctest errors yet).

Anyway, here's what I got on Ubuntu 10.04.4 LTS x86_64, with Sage 5.2.rc0 and GCC 4.7.0:

The following tests failed:

	sage -t  --long -force_lib devel/sage/doc/en/constructions/polynomials.rst # 1 doctests failed
	sage -t  --long -force_lib devel/sage/sage/rings/function_field/function_field.py # 1 doctests failed
	sage -t  --long -force_lib devel/sage/sage/rings/polynomial/multi_polynomial_libsingular.pyx # 3 doctests failed
	sage -t  --long -force_lib devel/sage/sage/rings/polynomial/multi_polynomial_ideal.py # 1 doctests failed
	sage -t  --long -force_lib devel/sage/sage/calculus/wester.py # 1 doctests failed
	sage -t  --long -force_lib devel/sage/sage/libs/singular/function.pyx # 2 doctests failed
	sage -t  --long -force_lib devel/sage/sage/combinat/words/words.py # 0 doctests failed

(The last failure was a segfault with a large traceback IIRC. Somewhat surprisingly successfully compiled with -O3, although many if not most compile commands override this by -O2; looks as if they use CPPFLAGS instead of CXXFLAGS in some parts.)

Let me know in case you're interested in further output; I guess you can reproduce most of the errors.

[jdemeyer]

Replying to leif:

I assume that this is work in progress (i.e., you haven't tried to fix doctest errors yet).

Exactly, I mainly worked on spkg-install.

[jdemeyer]

This is a new bug, the new answer is incorrect:

sage -t  --long -force_lib devel/sage/sage/rings/polynomial/multi_polynomial_libsingular.pyx
**********************************************************************
File "/release/merger/sage-5.2.rc1/devel/sage-main/sage/rings/polynomial/multi_polynomial_libsingular.pyx", line 3856:
    sage: p.factor()
Expected:
    (-2*v^2*u + 4*u^3 + v^2)^2
Got:
    (4) * (-2*v^2*u + 4*u^3 + v^2)^2
**********************************************************************

[leif]

Replying to jdemeyer:

This is a new bug, the new answer is incorrect:

Expected:
    (-2*v^2*u + 4*u^3 + v^2)^2
Got:
    (4) * (-2*v^2*u + 4*u^3 + v^2)^2

Yes, the most obvious one... :-)

Just noticed the failure in words.py apparently isn't reproducible. Traceback was

${SAGE_LOCAL}/lib/libcsage.so(print_backtrace+0x31)[0x2abd9e8f6664]
${SAGE_LOCAL}/lib/libcsage.so(sigdie+0x14)[0x2abd9e8f6696]
${SAGE_LOCAL}/lib/libcsage.so(sage_signal_handler+0x218)[0x2abd9e8f6250]
/lib/libpthread.so.0(+0xf8f0)[0x2abd9c9158f0]
${SAGE_LOCAL}/lib/libpython2.7.so.1.0(+0x135020)[0x2abd9c628020]
${SAGE_LOCAL}/lib/libpython2.7.so.1.0(+0x76aa7)[0x2abd9c569aa7]
${SAGE_LOCAL}/lib/libpython2.7.so.1.0(+0x1352b7)[0x2abd9c6282b7]
${SAGE_LOCAL}/lib/libpython2.7.so.1.0(_PyObject_GC_Malloc+0x115)[0x2abd9c6292b5]
${SAGE_LOCAL}/lib/libpython2.7.so.1.0(_PyObject_GC_New+0xd)[0x2abd9c6292ed]
${SAGE_LOCAL}/lib/libpython2.7.so.1.0(PyCFunction_NewEx+0xc5)[0x2abd9c584f05]
${SAGE_LOCAL}/lib/libpython2.7.so.1.0(Py_InitModule4_64+0x140)[0x2abd9c60bc60]
${SAGE_LOCAL}/lib/python/site-packages/numpy/core/multiarray.so(initmultiarray+0x2b)[0x2abdcd07186b]
${SAGE_LOCAL}/lib/libpython2.7.so.1.0(_PyImport_LoadDynamicModule+0x99)[0x2abd9c6073b9]
...
Segmentation fault

[jdemeyer]

Another Singular bug:

sage -t --long "devel/sage/sage/rings/function_field/function_field.py"
**********************************************************************
File "/release/merger/sage-5.2.rc1/devel/sage/sage/rings/function_field/function_field.py", line 1132:
    sage: f.factor()
Expected:
    (1/t) * (X + 3*t) * (X + 5*t) * (X + 6*t) * (X^2 + 1/t) * (X^2 + 6/t)
Got:
    (1/t) * (X + 3*t) * (X + 5*t) * (X + 6*t) * (X^4 + 6/t^2)
**********************************************************************

Reported upstream at ​http://www.singular.uni-kl.de:8002/trac/ticket/440.

[jdemeyer]

New spkg, adds fixes for Singular bugs 439 and 440.

[jdemeyer]

Reproducible doctest failure:

**********************************************************************
File "/release/merger/sage-5.3.beta0/devel/sage-main/sage/libs/singular/function.pyx", line 1569:
    sage: singular_list(2, 3, 6, ring=P)
Exception raised:
    Traceback (most recent call last):
      File "/release/merger/sage-5.3.beta0/local/bin/ncadoctest.py", line 1231, in run_one_test
        self.run_one_example(test, example, filename, compileflags)
      File "/release/merger/sage-5.3.beta0/local/bin/sagedoctest.py", line 38, in run_one_example
        OrigDocTestRunner.run_one_example(self, test, example, filename, compileflags)
      File "/release/merger/sage-5.3.beta0/local/bin/ncadoctest.py", line 1172, in run_one_example
        compileflags, 1) in test.globs
      File "<doctest __main__.example_72[16]>", line 1, in <module>
        singular_list(Integer(2), Integer(3), Integer(6), ring=P)###line 1569:
    sage: singular_list(2, 3, 6, ring=P)
      File "function.pyx", line 1248, in sage.libs.singular.function.SingularFunction.__call__ (sage/libs/singular/function.cpp:11875)
        return call_function(self, args, ring, interruptible, attributes)
      File "function.pyx", line 1438, in sage.libs.singular.function.call_function (sage/libs/singular/function.cpp:13425)
        raise RuntimeError("Error in Singular function call '%s':\n%s"%(self._name,
    RuntimeError: Error in Singular function call 'list':
     list(`int`,`int`,`int`) failed
**********************************************************************
File "/release/merger/sage-5.3.beta0/devel/sage-main/sage/libs/singular/function.pyx", line 1585:
    sage: singular_list((1,2,3),3,[1,2,3], ring=P)
Exception raised:
    Traceback (most recent call last):
      File "/release/merger/sage-5.3.beta0/local/bin/ncadoctest.py", line 1231, in run_one_test
        self.run_one_example(test, example, filename, compileflags)
      File "/release/merger/sage-5.3.beta0/local/bin/sagedoctest.py", line 38, in run_one_example
        OrigDocTestRunner.run_one_example(self, test, example, filename, compileflags)
      File "/release/merger/sage-5.3.beta0/local/bin/ncadoctest.py", line 1172, in run_one_example
        compileflags, 1) in test.globs
      File "<doctest __main__.example_72[26]>", line 1, in <module>
        singular_list((Integer(1),Integer(2),Integer(3)),Integer(3),[Integer(1),Integer(2),Integer(3)], ring=P)###line 1585:
    sage: singular_list((1,2,3),3,[1,2,3], ring=P)
      File "function.pyx", line 1248, in sage.libs.singular.function.SingularFunction.__call__ (sage/libs/singular/function.cpp:11875)
        return call_function(self, args, ring, interruptible, attributes)
      File "function.pyx", line 1438, in sage.libs.singular.function.call_function (sage/libs/singular/function.cpp:13425)
        raise RuntimeError("Error in Singular function call '%s':\n%s"%(self._name,
    RuntimeError: Error in Singular function call 'list':
     list(`intvec`,`int`,`list`) failed
**********************************************************************

Non-reproducible doctest failure:

**********************************************************************
File "/release/merger/sage-5.3.beta0/devel/sage-main/sage/rings/polynomial/multi_polynomial_libsingular.pyx", line 3921:
    sage: f.factor()
Expected:
    x * y^3 * (y^8 + (a)*y^7 + (a + 1)*x) * (x^7*y^3 + x*y^9 + (a)*x^8 + (a)*y^4)
Got:
    x * y^3 * (y^8 + (a)*y^7 + (a + 1)*x) * (x^7*y^3 + x*y^9 + (a + 1)*x^8 + (a + 1)*y^4)
**********************************************************************

[leif]

So far I only tested the files that previously failed for me (with the new spkg and the Sage library patch applied).

I get the same list() errors (although prepended by some debug output it seems, before the doctest error is printed), and sporadically the f.factor() error -- with varying results:

sage -t --long -force_lib "devel/sage/sage/rings/polynomial/multi_polynomial_libsingular.pyx"
**********************************************************************
File "${SAGE_ROOT}/devel/sage/sage/rings/polynomial/multi_polynomial_libsingular.pyx", line 3910:
    sage: f.factor()
Expected:
    x * y^3 * (y^8 + (a)*y^7 + (a + 1)*x) * (x^7*y^3 + x*y^9 + (a)*x^8 + (a)*y^4)
Got:
    x * y^3 * (y^8 + (a)*y^7 + (a + 1)*x) * (x^7*y^3 + x*y^9 + (a + 1)*x^8 + (a + 1)*y^4)
**********************************************************************
1 items had failures:
   1 of  86 in __main__.example_77
***Test Failed*** 1 failures.

sage -t --long -force_lib "devel/sage/sage/rings/polynomial/multi_polynomial_libsingular.pyx"
**********************************************************************
File "${SAGE_ROOT}/devel/sage/sage/rings/polynomial/multi_polynomial_libsingular.pyx", line 3910:
    sage: f.factor()
Expected:
    x * y^3 * (y^8 + (a)*y^7 + (a + 1)*x) * (x^7*y^3 + x*y^9 + (a)*x^8 + (a)*y^4)
Got:
    x * y^3 * (y^8 + (a)*y^7 + (a + 1)*x) * (x^7*y^3 + x*y^9)
**********************************************************************
1 items had failures:
   1 of  86 in __main__.example_77
***Test Failed*** 1 failures.

sage -t --long -force_lib "devel/sage/sage/rings/polynomial/multi_polynomial_libsingular.pyx"
**********************************************************************
File "${SAGE_ROOT}/devel/sage/sage/rings/polynomial/multi_polynomial_libsingular.pyx", line 3910:
    sage: f.factor()
Expected:
    x * y^3 * (y^8 + (a)*y^7 + (a + 1)*x) * (x^7*y^3 + x*y^9 + (a)*x^8 + (a)*y^4)
Got:
    x * y^3 * (y^8 + (a)*y^7 + (a + 1)*x) * (x^7*y^3 + x*y^9 + x^8 + y^4)
**********************************************************************
1 items had failures:
   1 of  86 in __main__.example_77
***Test Failed*** 1 failures.

(For me it's line 3910; presumably due to some other patch. The first variant seems to occur more often. All tests run sequentially, FWIW.)

P.S.: Run 100 times from the shell, the test failed 51 times.

[leif]

Replying to leif:

P.S.: Run 100 times from the shell, the test failed 51 times.

(Running sage -t ... in a for loop.)

25x
    x * y^3 * (y^8 + (a)*y^7 + (a + 1)*x) * (x^7*y^3 + x*y^9 + (a + 1)*x^8 + (a + 1)*y^4)

15x
    x * y^3 * (y^8 + (a)*y^7 + (a + 1)*x) * (x^7*y^3 + x*y^9 + x^8 + y^4)

11x
    x * y^3 * (y^8 + (a)*y^7 + (a + 1)*x) * (x^7*y^3 + x*y^9)

[jdemeyer]

Non-reproducible factoring bug reported upstream: ​http://www.singular.uni-kl.de:8002/trac/ticket/441

[jdemeyer]

Patch for the Sage library

[jdemeyer]

New patch fixes all doctest errors (except for the upstream factoring bug).

[jdemeyer]

I added a patch for Singular Trac 441 (which is in the git sources, even though Trac hasn't been updated yet).

Needs review (the spkg will be committed after positive_review if that's okay).

[AlexanderDreyer]

spkg and patch apply/install fine. Also the tests are fine. So I'm close to a positive review, just waiting for the patch from #12089.

[leif]

Just for the record: Did pass ptestlong (with a bit of luck, cf. the "random" factoring) with Sage 5.2[.rc0] (Ubuntu 10.04.4 LTS x86_64, FSF GCC 4.7.0, with -O3, see above).

Patch applies to Sage 5.3.beta0 as well (a few lines offset, never mind).

Only just noticed that old installations get removed before we even attempt to build the new one; this should in general only be done if the build succeeded.

(Haven't yet closely looked at [the changes to] spkg-install.)

[leif]

Replying to leif:

Patch applies to Sage 5.3.beta0 as well (a few lines offset, never mind).

ptestlong also passed without errors, but took longer than usual (despite the machine being otherwise idle).

[jpflori]

It would be nice to fix #13344 as well in this ticket. Hopefully the problem should be tracked down quite quickly.

[dimpase]

Replying to jpflori:

It would be nice to fix #13344 as well in this ticket. Hopefully the problem should be tracked down quite quickly.

yes, I can confirm that installing this spkg on MacOSX one gets Singular LIB files into SAGELOCAL/lib/, as well as in the right place, i.e. SAGELOCAL/share/singular/, which is certainly not what one would want.

[AlexanderDreyer]

Replying to dimpase:

yes, I can confirm that installing this spkg on MacOSX one gets Singular LIB files into SAGELOCAL/lib/, as well as in the right place, i.e. SAGELOCAL/share/singular/, which is certainly not what one would want.

It think the best solution would be to replace $MAKE by "$MAKE slibdir=$SAGE_LOCAL/share/singular" in the spkg.

[AlexanderDreyer]

Or maybe "$MAKE slibdir=\"$SAGE_LOCAL/share/singular\""?

[AlexanderDreyer]

@jdemeyer I could take care for the changes, if you don't mind?

[AlexanderDreyer]

Just in case (won't have time tomorrow), I was so free to propose a changed spkg here: ​http://boxen.math.washington.edu/home/dreyer/spkg/singular-3-1-5.p0.spkg

As before, the changes are not committed yet. Also I haven't changes the URL in the description, as Jeroen might continue with other fixes.

The changes are as follows (for review):

singular-3-1-5.p0.

@@ -9 +9 @@
 diff -r 1275ac543b53 SPKG.txt
 --- a/SPKG.txt  Sat Mar 17 11:18:23 2012 +0100
 +++ b/SPKG.txt  Sun Aug 05 22:36:36 2012 +0200
-@@ -30,14 +30,17 @@
+@@ -30,14 +30,19 @@
  See spkg-changes for how to delete unnecessary stuff under src/
  
  Several patches are applied. 
@@ -32 +32 @@
 +   http://www.singular.uni-kl.de:8002/trac/ticket/440
 + * singular_trac_441.patch: fix taken from upstream for
 +   http://www.singular.uni-kl.de:8002/trac/ticket/441
++ * sage_trac_12089.patch: fix (also reported upstream) from
++   https://github.com/alexanderdreyer/SingularSources/commit/7902222
  
  Other notes 
   * The option '--without-dynamic-kernel' is used on *all* 
-@@ -51,9 +54,51 @@
+@@ -51,9 +56,54 @@
     investigated as a possible cause later. 
   * Be sure that Singular/Singular.rc.in is present, as occasionally this
     was missing, which caused problems on Cygwin (e.g., see #10235)
@@ -49 +51 @@
  
  == ChangeLog ==
  
-+=== singular-3-1-5.p0 (Jeroen Demeyer, 2 August 2012) ===
++=== singular-3-1-5.p0 (Jeroen Demeyer, 5 August 2012) ===
 + * Trac #13237: Upgrade to version 3-1-5.
 + * Removed patches which are now upstreamed:
 +    - patches/Singular.Makefile.in.shared.patch
@@ -70 +72 @@
 +    - singular_trac_439.patch
 +    - singular_trac_440.patch
 +    - singular_trac_441.patch
++    - sage_trac_12089.patch
 + * In spkg-install, remove distclean() step; merge clean_headers() and
 +   part of the old distclean() into remove_old_version().
++ * In spkg-install, added slibdir="$SAGE_LOCAL/share/singular" to 
++   $MAKE install-libsingular .
 + * Only unset LD on Darwin.
 + * Remove the unsetting of TMPDIR (bug fixed upstream).
 + * Do not override user-set CFLAGS and CXXFLAGS.
@@ -6812 +6817 @@
 - #if defined(ix86Mac_darwin)
 - #define HAVE_ELF_SYSTEM
 - #endif
+diff -r 1275ac543b53 patches/sage_trac_12089.patch
+--- /dev/null   Thu Jan 01 00:00:00 1970 +0000
++++ b/patches/sage_trac_12089.patch     Sun Aug 05 22:36:36 2012 +0200
+@@ -0,0 +1,29 @@
++From 7902222c6e73d452b3d75576bd9bc0140f5e827a Mon Sep 17 00:00:00 2001
++From: Alexander Dreyer <alexander.dreyer@itwm.fraunhofer.de>
++Date: Sat, 4 Aug 2012 21:53:32 +0200
++Subject: [PATCH] fix: apply LDFLAGS fixes from svn#13210/git#b1dfafe to cygwin-specific flags (Sage's trac #12089; reported by Jean-Pierre Flori)
++
++---
++ Singular/Makefile.in |    6 +++---
++ 1 files changed, 3 insertions(+), 3 deletions(-)
++
++diff --git a/Singular/Makefile.in b/Singular/Makefile.in
++index c4a6e78..b907085 100644
++--- a/Singular/Makefile.in
+++++ b/Singular/Makefile.in
++@@ -90,9 +90,9 @@ LDFLAGSG     = -L@prefix@/kernel -L../kernel -lkernel_g @LD_DYN_FLAGS1@ @LDFLAGS@
++ LDFLAGSP      = -L@prefix@/kernel -L../kernel -lkernel_p @LD_DYN_FLAGS1@ @LDFLAGS@ 
++ else
++ ## -L/usr/local/lib -L@prefix@/modules/python --> PySingular
++-LDFLAGS               = -L@prefix@/kernel -L/bin -lkernel @LDFLAGS@ -L/usr/local/lib -L../modules/python 
++-LDFLAGSG      = -L@prefix@/kernel -L/bin -lkernel_g @LDFLAGS@ 
++-LDFLAGSP      = -L@prefix@/kernel -L/bin -lkernel_p @LDFLAGS@ 
+++LDFLAGS               = -L@prefix@/kernel -L../kernel -L/bin -lkernel @LDFLAGS@ -L/usr/local/lib -L../modules/python 
+++LDFLAGSG      = -L@prefix@/kernel -L../kernel -L/bin -lkernel_g @LDFLAGS@ 
+++LDFLAGSP      = -L@prefix@/kernel -L../kernel -L/bin -lkernel_p @LDFLAGS@ 
++ endif
++ LD_DYN_FLAGS1 = @LD_DYN_FLAGS1@
++ LD_DYN_FLAGS2 = @LD_DYN_FLAGS2@
++-- 
++1.6.0.2
++
 diff -r 1275ac543b53 patches/sing_win.cc.diff
 --- a/patches/sing_win.cc.diff  Sat Mar 17 11:18:23 2012 +0100
 +++ /dev/null   Thu Jan 01 00:00:00 1970 +0000
@@ -7330 +7368 @@
 +    config || return $?
  
      $MAKE clean
-     $MAKE install-libsingular
+-    $MAKE install-libsingular
++    $MAKE slibdir="$SAGE_LOCAL/share/singular" install-libsingular
  
      if [ $? -ne 0 ]; then
 -        echo "Unable to install libsingular."

[jdemeyer]

For the slibdir issue, I am patching Makefile.in instead, I think this is the best solution.

[jpflori]

Just a reminder for not forgetting to close #12089 and #13344. when this one is closed.

[AlexanderDreyer]

Just nitpicking: there a typo in my name.

[jdemeyer]

Replying to AlexanderDreyer:

Just nitpicking: there a typo in my name.

Apologies, fixed.

[jdemeyer]

Replying to leif:

Only just noticed that old installations get removed before we even attempt to build the new one; this should in general only be done if the build succeeded.

Unfortunately, Singular doesn't have separate build and install targets. So building and installing must happen together.

[dimpase]

Replying to jdemeyer: I can confirm that the latest spkg installs cleanly on MacOSX 10.6.8 and on Cygwin.

[jpflori]

Replying to dimpase:

Replying to jdemeyer: I can confirm that the latest spkg installs cleanly on MacOSX 10.6.8 and on Cygwin.

Same here on Cygwin and Linux (Ubuntu 12.04 x86_64). And I'm happy with the changes related to #13344. I've given it positive review.

[jdemeyer]

This doesn't build on Solaris: ​http://www.singular.uni-kl.de:8002/trac/ticket/443