FPLLL 5.4.0 and FPyLLL 0.5.4 #31025

malb · 2020-12-07T11:16:36Z

Tarball URLs - see checksums.ini

CC: @kiwifb

Component: packages: standard

Keywords: sd111

Author: Martin Albrecht

Branch: 29fcb34

Reviewer: Matthias Koeppe

Issue created by migration from https://trac.sagemath.org/ticket/31025

The text was updated successfully, but these errors were encountered:

malb · 2020-12-07T11:17:06Z

Branch: u/malb/fplll_5_4_0_and_fpylll_0_5_3_dev0

malb · 2020-12-07T11:18:54Z

Commit: 8faa924

malb · 2020-12-07T11:18:54Z

Author: Martin Albrecht

malb · 2020-12-07T11:18:54Z

New commits:

`8faa924`	`FPLLL 5.4.0 and FPyLLL 0.5.3.dev0`

mkoeppe · 2020-12-08T01:31:54Z

comment:4

build/pkgs/fpylll/checksums.ini needs upstream_url - see other python packages for the preferred URL format

sagetrac-git · 2020-12-08T10:34:11Z

Changed commit from 8faa924 to 0ab1d0e

sagetrac-git · 2020-12-08T10:34:11Z

Branch pushed to git repo; I updated commit sha1. New commits:

`0ab1d0e`	`add upstream_url`

malb · 2020-12-08T10:34:43Z

comment:6

Ta. Fixed.

mkoeppe · 2020-12-10T18:43:53Z

Changed keywords from none to sd111

mkoeppe · 2020-12-10T18:48:12Z

Reviewer: https://github.com/mkoeppe/sage/actions/runs/413724088

mkoeppe · 2020-12-10T22:57:57Z

Changed reviewer from https://github.com/mkoeppe/sage/actions/runs/413724088 to https://github.com/mkoeppe/sage/actions/runs/414158703

mkoeppe · 2020-12-11T06:00:51Z

comment:11

On local-conda-forge-ubuntu-standard (https://github.com/mkoeppe/sage/runs/1533944840)

  build/src/fpylll/fplll/enumeration.cpp: In function 'PyObject* __pyx_pf_6fpylll_5fplll_11enumeration_11Enumeration_6get_nodes(__pyx_obj_6fpylll_5fplll_11enumeration_Enumeration*, PyObject*)':
  build/src/fpylll/fplll/enumeration.cpp:7430:99: error: no matching function for call to 'fplll::Enumeration<fplll::Z_NR<__mpz_struct [1]>, fplll::FP_NR<double> >::get_nodes(int&)'
       __pyx_t_3 = __Pyx_PyInt_From_unsigned_long(__pyx_v_self->_core.mpz_d->get_nodes(__pyx_v__level)); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 576, __pyx_L1_error)

kiwifb · 2020-12-11T06:59:20Z

comment:12

Works fine in Gentoo. I must I never know where to find the appropriate logs in your runs. Sometimes it is easy and sometimes I just get lost. I found the good stuff in the raw logs.

2020-12-11T03:40:06.7154190Z   building 'fpylll.fplll.enumeration' extension
2020-12-11T03:40:06.7164990Z   /home/runner/work/sage/sage/.tox/local-conda-forge-ubuntu-standard/conda/bin/x86_64-conda-linux-gnu-cc -Wno-unused-result -Wsign-compare -DNDEBUG -fwrapv -O2 -Wall -Wstrict-prototypes -march=nocona -mtune=haswell -ftree-vectorize -fPIC -fstack-protector-strong -fno-plt -O2 -pipe -march=nocona -mtune=haswell -ftree-vectorize -fPIC -fstack-protector-strong -fno-plt -O2 -pipe -march=nocona -mtune=haswell -ftree-vectorize -fPIC -fstack-protector-strong -fno-plt -O2 -ffunction-sections -pipe -isystem /home/runner/work/sage/sage/.tox/local-conda-forge-ubuntu-standard/conda/include -DNDEBUG -D_FORTIFY_SOURCE=2 -O2 -isystem /home/runner/work/sage/sage/.tox/local-conda-forge-ubuntu-standard/conda/include -fPIC -Isrc/fpylll/fplll -I/home/runner/work/sage/sage/.tox/local-conda-forge-ubuntu-standard/local/lib/python3.8/site-packages/cysignals -I/home/runner/work/sage/sage/.tox/local-conda-forge-ubuntu-standard/local/include -I/home/runner/work/sage/sage/.tox/local-conda-forge-ubuntu-standard/local/lib/python3.8/site-packages/numpy/core/include -I/home/runner/work/sage/sage/.tox/local-conda-forge-ubuntu-standard/local/include -I/home/runner/work/sage/sage/.tox/local-conda-forge-ubuntu-standard/conda/include/python3.8 -c build/src/fpylll/fplll/enumeration.cpp -o build/temp.linux-x86_64-3.8/build/src/fpylll/fplll/enumeration.o -std=c++11 -fvisibility-inlines-hidden -std=c++17 -fmessage-length=0 -march=nocona -mtune=haswell -ftree-vectorize -fPIC -fstack-protector-strong -fno-plt -O2 -ffunction-sections -pipe -isystem /home/runner/work/sage/sage/.tox/local-conda-forge-ubuntu-standard/conda/include
2020-12-11T03:40:06.7174509Z   cc1plus: warning: command line option '-Wstrict-prototypes' is valid for C/ObjC but not for C++
2020-12-11T03:40:06.7176260Z   build/src/fpylll/fplll/enumeration.cpp: In function 'PyObject* __pyx_pf_6fpylll_5fplll_11enumeration_11Enumeration_6get_nodes(__pyx_obj_6fpylll_5fplll_11enumeration_Enumeration*, PyObject*)':
2020-12-11T03:40:06.7178575Z   build/src/fpylll/fplll/enumeration.cpp:7430:99: error: no matching function for call to 'fplll::Enumeration<fplll::Z_NR<__mpz_struct [1]>, fplll::FP_NR<double> >::get_nodes(int&)'
2020-12-11T03:40:06.7180107Z        __pyx_t_3 = __Pyx_PyInt_From_unsigned_long(__pyx_v_self->_core.mpz_d->get_nodes(__pyx_v__level)); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 576, __pyx_L1_error)

I have a feeling -std=c++17 is responsible for this, I compile with c++11 in Gentoo (probably a default somewhere).

malb · 2020-12-11T09:03:03Z

comment:13

Is there perhaps a race condition that FPyLLL is built before the new FPLLL?

kiwifb · 2020-12-11T09:06:01Z

comment:14

Replying to @malb:

Is there perhaps a race condition that FPyLLL is built before the new FPLLL?

Looking at the log again, I'd say it looks like it!

mkoeppe · 2020-12-11T16:17:33Z

comment:15

This build is accepting a system package

Checking whether SageMath should install SPKG fplll...
checking whether any of mpfr is installed as or will be installed as SPKG... no
checking for fplll >= 5.3... yes
configure: will use system package and not install SPKG fplll

If the new fpylll has other version requirements, build/pkgs/fplll/spkg-configure.m4 needs to be changed

mkoeppe · 2020-12-11T16:24:49Z

comment:16

Same failure also on fedora-32-standard https://github.com/mkoeppe/sage/runs/1533944016 and other builds

mkoeppe · 2020-12-11T16:26:11Z

comment:17

If practical at all, I would suggest to upstream to allow the Python bindings to accept a range of fplll versions...

sagetrac-git · 2020-12-11T17:23:54Z

Branch pushed to git repo; I updated commit sha1. New commits:

`9eb14e4`	`update required FPLLL version`

sagetrac-git · 2020-12-11T17:23:54Z

Changed commit from 0ab1d0e to 9eb14e4

malb · 2020-12-11T17:24:56Z

comment:19

Updated accordingly, I don't think we can commit to supporting a range of FPLLL versions in FPyLLL

mkoeppe · 2020-12-11T18:01:13Z

comment:20

Perhaps then an upper bound for the version needs to be set as well?

mkoeppe · 2020-12-14T17:45:39Z

comment:21

Likewise, should we set a range of acceptable fpylll versions (in the install-requires of sagelib)?

mkoeppe · 2020-12-15T21:29:36Z

comment:30

Based on your explanations, I think fpylll's install-requires.txt should also fix the version.

mkoeppe · 2020-12-16T22:50:16Z

Changed reviewer from https://github.com/mkoeppe/sage/actions/runs/414158703 to https://github.com/mkoeppe/sage/actions/runs/426748847

malb · 2020-12-17T09:06:08Z

comment:32

This doesn't look like an FPLLL issue: https://github.com/mkoeppe/sage/runs/1566871595?check_suite_focus=true#step:7:65

mkoeppe · 2020-12-17T17:16:00Z

comment:33

No, that's not related to this ticket

mkoeppe · 2020-12-17T17:21:48Z

comment:34

Replying to @mkoeppe:

Based on your explanations, I think fpylll's install-requires.txt should also fix the version.

As in:

diff --git a/build/pkgs/fpylll/install-requires.txt b/build/pkgs/fpylll/install-requires.txt
index 6e7d17221c..df75cd9fa7 100644
--- a/build/pkgs/fpylll/install-requires.txt
+++ b/build/pkgs/fpylll/install-requires.txt
@@ -1 +1 @@
-fpylll >=0.5.4
+fpylll ==0.5.4

mkoeppe · 2020-12-17T17:25:03Z

comment:35

I would also suggest to add some comments:

diff --git a/build/pkgs/fplll/spkg-configure.m4 b/build/pkgs/fplll/spkg-configure.m4
index 6da0045fe0..b98a71938e 100644
--- a/build/pkgs/fplll/spkg-configure.m4
+++ b/build/pkgs/fplll/spkg-configure.m4
@@ -4,6 +4,9 @@ SAGE_SPKG_CONFIGURE([fplll], [
     dnl if there's a usable system copy of fplll. Unless there's
     dnl a system that ships fplll without fplll.pc file, falling
     dnl back to a manual header/library search is pointless.
+    dnl
+    dnl Trac #31025: FPLLL/FPyLLL make no guarantee regarding compatibility
+    dnl other than "whatever versions were released at the same time should work together"
     PKG_CHECK_MODULES([FPLLL],
                       [fplll == 5.4],
                       [sage_spkg_install_fplll=no],
diff --git a/build/pkgs/fpylll/install-requires.txt b/build/pkgs/fpylll/install-requires.txt
index 6e7d17221c..4b425c186e 100644
--- a/build/pkgs/fpylll/install-requires.txt
+++ b/build/pkgs/fpylll/install-requires.txt
@@ -1 +1,3 @@
-fpylll >=0.5.4
+# Trac #31025: FPLLL/FPyLLL make no guarantee regarding compatibility
+# other than "whatever versions were released at the same time should work together"
+fpylll ==0.5.4

sagetrac-git · 2020-12-22T15:34:14Z

Branch pushed to git repo; I updated commit sha1. New commits:

`29fcb34`	`implement suggested changes by Matthias`

sagetrac-git · 2020-12-22T15:34:14Z

Changed commit from 26032b1 to 29fcb34

mkoeppe · 2020-12-22T15:36:29Z

Changed reviewer from https://github.com/mkoeppe/sage/actions/runs/426748847 to Matthias Koeppe

mkoeppe · 2020-12-22T15:38:20Z

comment:39

Marking it as critical because some rolling distributions are already picking up 5.4.0

vbraun · 2020-12-27T23:25:46Z

Changed branch from u/malb/fplll_5_4_0_and_fpylll_0_5_3_dev0 to 29fcb34

dcoudert · 2020-12-28T20:14:50Z

comment:41

For some reason, even after make distclean, configure is unable to find homebrew install of fplll 5.4.0 (on macOS). Then, the compilation fails on fpylll (may be a mix of fplll installations).

After modifying spkg-configure.m4 to change fplll == 5.4.0 to fplll >= 5.4.0 and adding file distros/homebrew.txt with fplll, configure finds the local install and everything goes well.

Should we open a ticket for that ?

dcoudert · 2020-12-28T20:14:50Z

Changed commit from 29fcb34 to none

mkoeppe · 2020-12-28T21:00:39Z

comment:42

Replying to @dcoudert:

Should we open a ticket for that ?

Yes please. What version of fplll does pkg-config report?

dcoudert · 2020-12-29T00:00:29Z

comment:43

not sure how to get that, but inside config.log, I see:

## ------------------------------------------------------ ##
## Checking whether SageMath should install SPKG fplll... ##
## ------------------------------------------------------ ##
configure:19538: checking whether any of mpfr is installed as or will be installed as SPKG
configure:19547: result: no
configure:19552: checking for fplll >= 5.4
configure:19559: $PKG_CONFIG --exists --print-errors "fplll >= 5.4"
configure:19562: $? = 0
configure:19576: $PKG_CONFIG --exists --print-errors "fplll >= 5.4"
configure:19579: $? = 0
configure:19617: result: yes
configure:19629: will use system package and not install SPKG fplll

and I have fplll 5.4.0 installed by homebrew.

Follow-up on #31127

mkoeppe · 2021-01-01T21:55:59Z

comment:44

Another follow up in #31146: cygwin-standard: fpylll build fails

malb added this to the sage-9.3 milestone Dec 7, 2020

malb added the p: major / 3 label Dec 7, 2020

This comment has been minimized.

Sign in to view

malb added s: needs review labels Dec 7, 2020

This comment has been minimized.

Sign in to view

mkoeppe added s: needs work and removed s: needs info labels Dec 16, 2020

malb added s: needs review and removed s: needs work labels Dec 22, 2020

mkoeppe added s: positive review and removed s: needs review labels Dec 22, 2020

mkoeppe added p: critical / 2 and removed p: major / 3 labels Dec 22, 2020

vbraun removed the s: positive review label Dec 27, 2020

vbraun closed this as completed in 0d79fd6 Dec 27, 2020

This was referenced Jan 30, 2023

Meta-ticket: Modularize sagelib into separate distributions (pip-installable packages) sagemath-... for Sage 10.x #29705

Open

Meta-ticket: Fix dependencies of upstream Python packages #30922

Open

malb mentioned this issue Dec 26, 2020

FPLLL and FPyLLL #31024

Closed

dcoudert mentioned this issue Jan 3, 2021

FPLLL 5.4.0 on macOS with homebrew #31127

Closed

FPLLL 5.4.0 and FPyLLL 0.5.4 #31025

FPLLL 5.4.0 and FPyLLL 0.5.4 #31025

Comments

malb commented Dec 7, 2020

malb commented Dec 7, 2020

malb commented Dec 7, 2020

malb commented Dec 7, 2020

malb commented Dec 7, 2020

This comment has been minimized.

mkoeppe commented Dec 8, 2020

sagetrac-git mannequin commented Dec 8, 2020

sagetrac-git mannequin commented Dec 8, 2020

malb commented Dec 8, 2020

mkoeppe commented Dec 10, 2020

This comment has been minimized.

mkoeppe commented Dec 10, 2020

mkoeppe commented Dec 10, 2020

mkoeppe commented Dec 11, 2020

kiwifb commented Dec 11, 2020

malb commented Dec 11, 2020

kiwifb commented Dec 11, 2020

mkoeppe commented Dec 11, 2020

mkoeppe commented Dec 11, 2020

mkoeppe commented Dec 11, 2020

sagetrac-git mannequin commented Dec 11, 2020

sagetrac-git mannequin commented Dec 11, 2020

malb commented Dec 11, 2020

mkoeppe commented Dec 11, 2020

mkoeppe commented Dec 14, 2020

mkoeppe commented Dec 15, 2020

mkoeppe commented Dec 16, 2020

malb commented Dec 17, 2020

mkoeppe commented Dec 17, 2020

mkoeppe commented Dec 17, 2020

mkoeppe commented Dec 17, 2020

sagetrac-git mannequin commented Dec 22, 2020

sagetrac-git mannequin commented Dec 22, 2020

mkoeppe commented Dec 22, 2020

mkoeppe commented Dec 22, 2020

vbraun commented Dec 27, 2020

dcoudert commented Dec 28, 2020

dcoudert commented Dec 28, 2020

mkoeppe commented Dec 28, 2020

dcoudert commented Dec 29, 2020

mkoeppe commented Jan 1, 2021