search

sagemath / sage

Main repository of SageMath
https://www.sagemath.org
Other
1.35k stars 462 forks source link

Upgrade to cypari2-2.0.3 #26442

Closed jdemeyer closed 5 years ago

jdemeyer commented 5 years ago

Tarball: https://files.pythonhosted.org/packages/64/bd/93fd5d8e7515a87be0f74958d0e1dd1b67ac0d009c0ca4d36be891e72c49/cypari2-2.0.3.tar.gz

Depends on #27060

CC: @videlec @defeo @kiwifb

Component: packages: standard

Author: Jeroen Demeyer

Branch/Commit: cd62d45

Reviewer: Vincent Delecroix, Timo Kaufmann

Issue created by migration from https://trac.sagemath.org/ticket/26442

videlec commented 5 years ago

Attachment: vincent-cypari-2.0.1.log

videlec commented 5 years ago
comment:46

For the full log of the failures mentioned in [comment:42] and [comment:43] is attachment: vincent-cypari-2.0.1.log. Let me mention that it is a Python 3 build of Sage.

jdemeyer commented 5 years ago
comment:47

I solved the segmentation fault with SAGE_DEBUG=yes.

The underlying problem was exceptions containing pointers to PARI objects, where the exception was being kept alive after the PARI object was deallocated. This invalid exception objects then crashed Python. I solved the problem in cypari2 upstream by always making copies of PARI objects in exceptions (assuming that exceptions are exceptional, this should not affect performance). I don't know why the problem occurs only (or at least with much higher probability) with SAGE_DEBUG=yes.

I will do some more testing and then release cypari2-2.0.2

jdemeyer commented 5 years ago
comment:48

Replying to @videlec:

Let me mention that it is a Python 3 build of Sage.

Indeed. Confirmed with Python 3 but only inside Sage.

jdemeyer commented 5 years ago
comment:49

OK, the problem with Python 3 is simple: the test suite is always being run with Python 2 which obviously doesn't have cypari2 installed.

jdemeyer commented 5 years ago

Description changed:

--- 
+++ 
@@ -1 +1 @@
-**Tarball**: https://files.pythonhosted.org/packages/a5/8a/c0cd4ace712745f6d40f35feeb6795812ea78a5ccb30cc176a57ef83ec34/cypari2-2.0.1.tar.gz
+**Tarball**: https://files.pythonhosted.org/packages/b4/0d/817d280300e525a879b1d68ae31a4500f29b82ee4a603be5b0ff9572976e/cypari2-2.0.2.tar.gz
7ed8c4ca-6d56-4ae9-953a-41e42b4ed313 commented 5 years ago

Changed commit from e949ce9 to d754ee9

7ed8c4ca-6d56-4ae9-953a-41e42b4ed313 commented 5 years ago

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

d754ee9Upgrade to cypari2 version 2.0.2
videlec commented 5 years ago
comment:52

Replying to @jdemeyer:

OK, the problem with Python 3 is simple: the test suite is always being run with Python 2 which obviously doesn't have cypari2 installed.

Why on earth python defaults to python2 on py3 build!?

videlec commented 5 years ago
comment:54

sage-python23 is indeed a solution. I confirm that tests now pass with py3 build.

videlec commented 5 years ago
comment:55

The following line in the installation log looks suspicious

[cypari-2.0.2] No record that 'cypari' was ever installed; skipping uninstall

Any idea where it comes from? Is it not detected because of the discrepency cypari vs cypari2?

jhpalmieri commented 5 years ago
comment:56

Replying to @videlec:

Replying to @jdemeyer:

OK, the problem with Python 3 is simple: the test suite is always being run with Python 2 which obviously doesn't have cypari2 installed.

Why on earth python defaults to python2 on py3 build!?

python is always supposed to default to python2: that's a Python convention.

videlec commented 5 years ago
comment:57

Replying to @jhpalmieri:

Replying to @videlec:

Replying to @jdemeyer:

OK, the problem with Python 3 is simple: the test suite is always being run with Python 2 which obviously doesn't have cypari2 installed.

Why on earth python defaults to python2 on py3 build!?

python is always supposed to default to python2: that's a Python convention.

Really? At least on archlinux python points to Python 3

$ ls -l /usr/bin/python
lrwxrwxrwx 1 root root 7 22 oct.  12:41 /usr/bin/python -> python3
jhpalmieri commented 5 years ago
comment:58

See https://www.python.org/dev/peps/pep-0394/ (which mentions Arch Linux specifically).

videlec commented 5 years ago
comment:59

Let's give it another try.

jdemeyer commented 5 years ago
comment:60

Replying to @videlec:

The following line in the installation log looks suspicious

[cypari-2.0.2] No record that 'cypari' was ever installed; skipping uninstall

Any idea where it comes from?

When you run ./sage -f cypari, it first uninstalls the package. So by the time you are installing again, there is "No record that 'cypari' was ever installed". I agree that the message is confusing, but it's completely normal.

timokau commented 5 years ago
comment:61

I tried to update the nix package to 2.0.2 and got a segfault during ptestlong (not reproducible in isolation):

sage -t --long /nix/store/rppc0jilfm84ydj405mafd8kjzqr97rp-sage-src-8.5/src/sage/groups/additive_abelian/all.py
    [0 tests, 0.00 s]
doctest:warning
  [...]
  File "/nix/store/351z9795scybnc45vxydpcsp3x9bc88g-python-2.7.15-env/lib/python2.7/multiprocessing/queues.py", line 268, in _feed
    send(obj)
:
RuntimeWarning: cypari2 leaked 140736969821680 bytes on the PARI stack
------------------------------------------------------------------------
/nix/store/351z9795scybnc45vxydpcsp3x9bc88g-python-2.7.15-env/lib/python2.7/site-packages/cysignals/signals.so(+0x7bfb)[0x7ffff1a2dbfb]
[...]
/nix/store/p8fyjvz3djclmndwxjfcj7zkmx89wniw-glibc-2.27/lib/libc.so.6(clone+0x3f)[0x7ffff6b231ff]
------------------------------------------------------------------------
Attaching gdb to process id 10236.
Cannot find gdb installed
GDB is not installed.
Install gdb for enhanced tracebacks.
------------------------------------------------------------------------
Unhandled SIGSEGV: A segmentation fault occurred.
This probably occurred because a *compiled* module has a bug
in it and is not properly wrapped with sig_on(), sig_off().
Python will now terminate.
------------------------------------------------------------------------
sage -t --long /nix/store/rppc0jilfm84ydj405mafd8kjzqr97rp-sage-src-8.5/src/sage/groups/additive_abelian/additive_abelian_wrapper.py
    Killed due to segmentation fault
**********************************************************************
Tests run before process (pid=10236) failed:
sage: E = EllipticCurve('30a2') ## line 14 ##
[...]
sage: G([1,1]) ## line 270 ##
doctest:warning
  [...]
  File "/nix/store/351z9795scybnc45vxydpcsp3x9bc88g-python-2.7.15-env/lib/python2.7/site-packages/sage/misc/superseded.py", line 148, in warning
    warn(message, warning_class, stacklevel)
:
DeprecationWarning: The default behaviour changed! If you *really* want a linear combination of smith generators, use .linear_combination_of_smith_form_gens.
See http://trac.sagemath.org/16261 for details.
(6, 2)
sage: G(G([1,1])) ## line 276 ##
(6, 2)
sage: sig_on_count() # check sig_on/off pairings (virtual doctest) ## line 278 ##
0

Full log: https://gist.github.com/timokau/62edfecf23ed96b3a5284c5d913a1643

That was with sage 8.5 with this ticket applied and pari 2.11.1.

Haven't had any time to investigate much further yet. Although I don't really know what to do besides checking how reliable this reproduces.

jdemeyer commented 5 years ago
comment:62

Replying to @timokau:

Although I don't really know what to do besides checking how reliable this reproduces.

Running the tests with gdb installed might help.

jdemeyer commented 5 years ago
comment:63

Replying to @timokau:

not reproducible in isolation

Is it reproducible when running make ptestlong?

timokau commented 5 years ago
comment:64

Replying to @jdemeyer:

Replying to @timokau:

not reproducible in isolation

Is it reproducible when running make ptestlong?

On the second try I got a different failure:

sage -t --long /nix/store/rppc0jilfm84ydj405mafd8kjzqr97rp-sage-src-8.5/src/sage/doctest/test.py
**********************************************************************
File "/nix/store/rppc0jilfm84ydj405mafd8kjzqr97rp-sage-src-8.5/src/sage/doctest/test.py", line 220, in sage.doctest.test
Failed example:
    subprocess.call(["sage", "-tp", "1000000", "--timeout=120",  # long time
         "--warn-long", "0", "99seconds.rst", "interrupt_diehard.rst"], **kwds2)
Expected:
    Running doctests...
    Doctesting 2 files using 1000000 threads.
    Killing test 99seconds.rst
    Killing test interrupt_diehard.rst
    ----------------------------------------------------------------------
    Doctests interrupted: 0/2 files tested
    ----------------------------------------------------------------------
    ...
    128
Got:
    Running doctests with ID 2019-01-12-08-52-31-cbdf1975.
    Using --optional=dochtml,gmp,memlimit,nose,sage
    Doctesting 2 files using 1000000 threads.
    Killing test 99seconds.rst
    Killing test interrupt_diehard.rst
    ------------------------------------------------------------------------
    /nix/store/351z9795scybnc45vxydpcsp3x9bc88g-python-2.7.15-env/lib/python2.7/site-packages/cysignals/signals.so(+0x7bfb)[0x7ffff1a2dbfb]
    [...]
    /nix/store/46kfixv5s95klr3wmnydcp8pc6cfgr08-python-2.7.15/lib/libpython2.7.so.1.0(Py_Main+0xb4a)[0x7ffff78f4aaa]
    /nix/store/p8fyjvz3djclmndwxjfcj7zkmx89wniw-glibc-2.27/lib/libc.so.6(__libc_start_main+0xee)[0x7ffff6a52b5e]
    /nix/store/46kfixv5s95klr3wmnydcp8pc6cfgr08-python-2.7.15/bin/python(_start+0x2a)[0x4007ba]
    ------------------------------------------------------------------------
    ----------------------------------------------------------------------
    Doctests interrupted: 0/2 files tested
    ----------------------------------------------------------------------
    Total time for all tests: 0.1 seconds
        cpu time: 0.0 seconds
        cumulative wall time: 0.0 seconds
    Attaching gdb to process id 6906.
    Cannot find gdb installed
    GDB is not installed.
    Install gdb for enhanced tracebacks.
    ------------------------------------------------------------------------
    128
**********************************************************************
1 item had failures:
   1 of  24 in sage.doctest.test
    [54 tests, 1 failure, 50.91 s]
Killing test /nix/store/rppc0jilfm84ydj405mafd8kjzqr97rp-sage-src-8.5/src/sage/doctest/sources.py
Killing test /nix/store/rppc0jilfm84ydj405mafd8kjzqr97rp-sage-src-8.5/src/sage/dynamics/arithmetic_dynamics/projective_ds.py
Killing test /nix/store/rppc0jilfm84ydj405mafd8kjzqr97rp-sage-src-8.5/src/sage/dynamics/flat_surfaces/all.py
------------------------------------------------------------------------
/nix/store/351z9795scybnc45vxydpcsp3x9bc88g-python-2.7.15-env/lib/python2.7/site-packages/cysignals/signals.so(+0x7bfb)[0x7ffff1a2dbfb]
[...]
/nix/store/46kfixv5s95klr3wmnydcp8pc6cfgr08-python-2.7.15/bin/python(_start+0x2a)[0x4007ba]
------------------------------------------------------------------------
----------------------------------------------------------------------
sage -t --long /nix/store/rppc0jilfm84ydj405mafd8kjzqr97rp-sage-src-8.5/src/sage/doctest/test.py  # 1 doctest failed
Doctests interrupted: 1384/3738 files tested

Only difference to the first run was that I passed --exitfirst. Full log: https://gist.github.com/timokau/84fa6a991119087661e3491181612727 sta To make sure this is really caused by the cypari2 update, I then reverted the update and re-ran the testsuite again. This time it succeeded.

timokau commented 5 years ago
comment:65

I'll try to reproduce with sage-the-distro on 8.6.rc0, but that will take a while.

jdemeyer commented 5 years ago
comment:66

Replying to @timokau:

On the second try I got a different failure:

You already reported that as #26912. This is certainly unrelated to cypari2.

timokau commented 5 years ago
comment:67

Right, I forgot.

jdemeyer commented 5 years ago
comment:68

So, how should we proceed? If I have no way to reproduce the problem, I cannot fix it.

timokau commented 5 years ago
comment:69

I'll try to narrow it down a bit. If all else fails, bisect cypari2 and find the commit that causes this.

That takes time, especially considering that each test requires a ptestlong. Again, is there any hurry for this? Presumably this will only make it into 8.7 either way right?

jdemeyer commented 5 years ago
comment:70

Replying to @timokau:

If all else fails, bisect cypari2 and find the commit that causes this.

It's almost certainly https://github.com/defeo/cypari2/commit/551a5277456a3170dd690773e22f319cc1ad795a

jdemeyer commented 5 years ago
comment:71

(in the unlikely case that it's not caused by that commit, that would be very useful information)

timokau commented 5 years ago
comment:72

I'm currently trying to get vanilla sage to bulid+test to try to reproduce it there (unrelated issues with linking and python).

For the nix package it definitely reproduces very reliably (3 times in a row) so after that I'll check if the commit is responsible.

jdemeyer commented 5 years ago
comment:73

For the nix package it definitely reproduces very reliably (3 times in a row)

If possible, please keep all logs of all make ptestlong runs. For example, I want to know if the failures are literally identical or whether there is some random variation.

timokau commented 5 years ago
comment:74

Alright, I'll keep them. Kept the first 3 already. Currently running with the cypari checkout you pointed out.

Unrelated but do you have any idea why the python patched could fail to apply? I'm getting errors from the safe-directory test and the regex group index issue when testing vanilla sage. It does use sages python (the system python even has the regex thing patched). The build log does say that the patches were applied. The log also says "checking for sys/types.h... yes" which seems relevant for the sys_path_security patch.

jdemeyer commented 5 years ago
comment:75

Replying to @timokau:

It does use sages python

Are you sure about that? Is there by chance more than 1 Python installed?

timokau commented 5 years ago
comment:76

Yes a system python is installed. But I did export PATH=$PWD/local/bin:$PATH, verified with which python and also tried the safe directory example manually (./local/bin/python did not complain).

jdemeyer commented 5 years ago
comment:77

I pushed two more relevant commits to https://github.com/defeo/cypari2/commits/master

They clean up some of the code and might by chance fix your problem.

jdemeyer commented 5 years ago
comment:78

Replying to @timokau:

Yes a system python is installed. But I did export PATH=$PWD/local/bin:$PATH, verified with which python and also tried the safe directory example manually (./local/bin/python did not complain).

Could it be a Python 2 versus Python 3 problem? Just guessing...

timokau commented 5 years ago
comment:79

Thanks, I'll do a test with latest master too.

timokau commented 5 years ago
comment:80

Replying to @jdemeyer:

Replying to @timokau:

Yes a system python is installed. But I did export PATH=$PWD/local/bin:$PATH, verified with which python and also tried the safe directory example manually (./local/bin/python did not complain).

Could it be a Python 2 versus Python 3 problem? Just guessing...

I don't think so:

sage-build-env-chrootenv:timo@pad:~/repos/sage$ which python
/home/timo/repos/sage/local/bin/python
sage-build-env-chrootenv:timo@pad:~/repos/sage$ python --version
Python 2.7.14
sage-build-env-chrootenv:timo@pad:~/repos/sage$ mkdir /tmp/dir
sage-build-env-chrootenv:timo@pad:~/repos/sage$ chmod 777 /tmp/dir
sage-build-env-chrootenv:timo@pad:~/repos/sage$ touch /tmp/dir/file
sage-build-env-chrootenv:timo@pad:~/repos/sage$ python /tmp/dir/file
sage-build-env-chrootenv:timo@pad:~/repos/sage$ grep Applying logs/pkgs/python2-2.7.15.p0.log 
Applying patches from ../patches...
Applying ../patches/2.6.5-FD_SETSIZE.patch
Applying ../patches/cygwin64.patch
Applying ../patches/descr_ref-issue_25750.patch
Applying ../patches/getcallargs-issue_20108.patch
Applying ../patches/hashlibfallbacks-issue_18000.patch
Applying ../patches/linux_linking_issue_25229.patch
Applying ../patches/ncurses-issue_14438.patch
Applying ../patches/ncurses-issue_9665.patch
Applying ../patches/no_strict_proto-issue_5755.patch
Applying ../patches/permissions.patch
Applying ../patches/re_match_index-issue_27177.patch
Applying ../patches/socket.patch
Applying ../patches/sys_path_security-issue_16202.patch
Applying ../patches/tinfo.patch
Applying ../patches/uuid-issue_11063.patch
sage-build-env-chrootenv:timo@pad:~/repos/sage$

Very weird...

jdemeyer commented 5 years ago
comment:81

A library problem? Note that the python executable is really just a wrapper around a library libpython2.7.so

timokau commented 5 years ago
comment:82

551a5277456a3170dd690773e22f319cc1ad795a actually doesn't show this problem:

sage -t --long /nix/store/rppc0jilfm84ydj405mafd8kjzqr97rp-sage-src-8.5/src/sage/groups/additive_abelian/additive_abelian_group.py
    [76 tests, 1.50 s]

Although I didn't adjust the sage patch, so there are some other failures like this:

Failed example:
    ModularSymbols(11, 4, base_ring=QuadraticField(-7,'b'),sign=1).cuspidal_submodule()._q_expansion_module(prec=5, algorithm="eigen") # indirect doctest
Expected:
    Vector space of degree 5 and dimension 2 over Number Field in b with defining polynomial x^2 + 7
    Basis matrix:
    [ 0  1  0  3 -6]
    [ 0  0  1 -4  2]
Got:
    Vector space of degree 5 and dimension 2 over Number Field in b with defining polynomial x^2 + 7
    Basis matrix:
    [ 0  1  1  0 -6]
    [ 0  0  0  1 -2]

Possible that there is some interaction.

Didn't test master yet.

timokau commented 5 years ago
comment:83

Replying to @jdemeyer:

A library problem? Note that the python executable is really just a wrapper around a library libpython2.7.so

I think that is it, at least LD_DEBUG=libs python shows 

calling init: /usr/lib/libpython2.7.so.1.0
jdemeyer commented 5 years ago
comment:84

In the mean time, I found an unrelated serious bug in cypari2-2.0.2 so I'll need to make a new version anyway.

jdemeyer commented 5 years ago
comment:85

After a few make ptestlong runs with this ticket, I got

sage -t --long src/sage/modular/abvar/torsion_subgroup.py
**********************************************************************
File "src/sage/modular/abvar/torsion_subgroup.py", line 46, in sage.modular.abvar.torsion_subgroup
Failed example:
    for N in range(1,38):
       for A in J0(N).new_subvariety().decomposition():
           T = A.rational_torsion_subgroup()
           print('%-5s%-5s%-5s%-5s'%(N, A.dimension(), T.divisor_of_order(), T.multiple_of_order()))
Expected:
    11   1    5    5
    14   1    6    6
    15   1    8    8
    17   1    4    4
    19   1    3    3
    20   1    6    6
    21   1    8    8
    23   2    11   11
    24   1    8    8
    26   1    3    3
    26   1    7    7
    27   1    3    3
    29   2    7    7
    30   1    6    6
    31   2    5    5
    32   1    4    4
    33   1    4    4
    34   1    6    6
    35   1    3    3
    35   2    16   16
    36   1    6    6
    37   1    1    1
    37   1    3    3
Got:
    11   1    5    5    
    14   1    6    6    
    15   1    8    8    
    17   1    4    4    
    19   1    3    3    
    20   1    6    6    
    21   1    8    8    
    23   2    11   11   
    24   1    8    8    
    26   1    3    3    
    26   1    7    7    
    27   1    3    3    
    doctest:warning
      File "/home/jdemeyer/sage-test/src/bin/sage-runtests", line 163, in <module>
        err = DC.run()
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/doctest/control.py", line 1225, in run
        self.run_doctests()
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/doctest/control.py", line 929, in run_doctests
        self.dispatcher.dispatch()
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/doctest/forker.py", line 1991, in dispatch
        self.parallel_dispatch()
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/doctest/forker.py", line 1888, in parallel_dispatch
        w.start()  # This might take some time
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/doctest/forker.py", line 2174, in start
        super(DocTestWorker, self).start()
      File "/home/jdemeyer/sage-test/local/lib/python2.7/multiprocessing/process.py", line 130, in start
        self._popen = Popen(self)
      File "/home/jdemeyer/sage-test/local/lib/python2.7/multiprocessing/forking.py", line 126, in __init__
        code = process_obj._bootstrap()
      File "/home/jdemeyer/sage-test/local/lib/python2.7/multiprocessing/process.py", line 267, in _bootstrap
        self.run()
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/doctest/forker.py", line 2130, in run
        task(self.options, self.outtmpfile, msgpipe, self.result_queue)
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/doctest/forker.py", line 2468, in __call__
        doctests, extras = self._run(runner, options, results)
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/doctest/forker.py", line 2515, in _run
        result = runner.run(test)
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/doctest/forker.py", line 860, in run
        return self._run(test, compileflags, out)
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/doctest/forker.py", line 671, in _run
        self.compile_and_execute(example, compiler, test.globs)
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/doctest/forker.py", line 1086, in compile_and_execute
        exec(compiled, globs)
      File "<doctest sage.modular.abvar.torsion_subgroup[9]>", line 2, in <module>
        for A in J0(N).new_subvariety().decomposition():
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/modular/abvar/abvar.py", line 4521, in decomposition
        S = ModularAbelianVariety_abstract.decomposition(self, simple=simple, bound=bound)
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/modular/abvar/abvar.py", line 3398, in decomposition
        I_F, I_E, X = self._classify_ambient_factors(simple=simple, bound=bound)
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/modular/abvar/abvar.py", line 3531, in _classify_ambient_factors
        S   = self.vector_space()
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/modular/abvar/abvar.py", line 288, in vector_space
        self.__vector_space = self.lattice().change_ring(QQ)
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/modular/abvar/abvar.py", line 4122, in lattice
        K0 = M.integral_structure()
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/modular/modsym/space.py", line 1584, in integral_structure
        I = A.integral_structure()
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/modular/modsym/ambient.py", line 2313, in integral_structure
        B = B.echelon_form(algorithm='pari', include_zero_rows=False)
      File "/home/jdemeyer/sage-test/local/lib/python2.7/site-packages/sage/matrix/matrix_space.py", line 817, in __call__
        return self.element_class(self, entries, copy, coerce)
    :
    RuntimeWarning: cypari2 leaked 11040 bytes on the PARI stack
    29   2    7    7    
    30   1    6    6    
    31   2    5    5    
    32   1    4    4    
    33   1    4    4    
    34   1    6    6    
    35   1    3    3    
    35   2    16   16   
    36   1    6    6    
    37   1    1    1    
    37   1    3    3    
**********************************************************************
timokau commented 5 years ago
comment:86

I haven't seen that particular failure yet.

jdemeyer commented 5 years ago

Dependencies: #27060

7ed8c4ca-6d56-4ae9-953a-41e42b4ed313 commented 5 years ago

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

9fffd00Fix various invalid uses of sig_on()
68a7babUpgrade to cypari2 version 2.0.2
7ed8c4ca-6d56-4ae9-953a-41e42b4ed313 commented 5 years ago

Changed commit from d754ee9 to 68a7bab

timokau commented 5 years ago
comment:89

git bisect actually points me to this commit: https://github.com/defeo/cypari2/commit/b980e648648

And with that reverted on 2.0.2, the tests pass.

jdemeyer commented 5 years ago
comment:90

Replying to @timokau:

git bisect actually points me to this commit: https://github.com/defeo/cypari2/commit/b980e648648

Maybe, but that's surely not the real reason for the failing tests.

timokau commented 5 years ago
comment:91

Why not? What does that do?

7ed8c4ca-6d56-4ae9-953a-41e42b4ed313 commented 5 years ago

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

9e1fd1eClean up HNF computation using PARI
a185f05Upgrade to cypari2 version 2.0.2
7ed8c4ca-6d56-4ae9-953a-41e42b4ed313 commented 5 years ago

Changed commit from 68a7bab to a185f05