Merge sage_brial into sagelib

[mkoeppe]

As discussed in #29369 and #28918.

[fbissey]

I propose to copy the current sage-brial code under sage/rings/polynomial. This is based on the fact all (but one) imports of brial happen there.

$ grep -r brial src/sage/*
src/sage/rings/polynomial/pbori.pyx:# distutils: libraries = brial brial_groebner M4RI_LIBRARIES LIBPNG_LIBRARIES
src/sage/rings/polynomial/pbori.pyx:    sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:        #from brial.interpolate import interpolate_smallest_lex
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import BooleanMonomialMonoid, BooleanMonomial
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid, BooleanMonomial
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleSet
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import BooleanPolynomial
src/sage/rings/polynomial/pbori.pyx:        from brial.parallel import _encode_polynomial
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleSet
src/sage/rings/polynomial/pbori.pyx:        from brial import red_tail
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:        from brial.gbcore import groebner_basis
src/sage/rings/polynomial/pbori.pyx:        from brial import red_tail
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import BooleSet
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleSet
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleSet
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import BooleanPolynomialVector
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanPolynomialVector
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanPolynomialVector
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanPolynomialVector
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanPolynomialVector
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanPolynomialVector
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanPolynomialVector
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import GroebnerStrategy
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import GroebnerStrategy
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import groebner_basis
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import GroebnerStrategy
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanPolynomialVector
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import GroebnerStrategy
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import groebner_basis
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import GroebnerStrategy
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import GroebnerStrategy
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import GroebnerStrategy
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleanMonomialMonoid
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import map_every_x_to_x_plus_one
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import zeros
src/sage/rings/polynomial/pbori.pyx:        sage: from brial.interpolate import *
src/sage/rings/polynomial/pbori.pyx:        sage: from brial.interpolate import *
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import ll_red_nf_redsb
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import ll_red_nf_noredsb
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import ll_red_nf_noredsb_single_recursive_call
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import if_then_else
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import top_index
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import substitute_variables
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import substitute_variables
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import random_set, set_random_seed
src/sage/rings/polynomial/pbori.pyx:        sage: from brial import random_set, set_random_seed
src/sage/rings/polynomial/pbori.pyx:# todo: merge with pickling from brial.parallel
src/sage/rings/polynomial/pbori.pyx:# todo: merge with pickling from brial.parallel
src/sage/rings/polynomial/pbori.pyx:    from brial.parallel import _decode_polynomial
src/sage/rings/polynomial/pbori.pyx:# todo: merge with pickling from brial.parallel
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleConstant
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleConstant
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleConstant
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleConstant
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleConstant
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleConstant
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleConstant
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import BooleConstant
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/pbori.pyx:            sage: from brial import *
src/sage/rings/polynomial/multi_polynomial_sequence.py:        from brial import gauss_on_polys
src/sage/rings/polynomial/multi_polynomial_sequence.py:        from brial.ll import eliminate,ll_encode,ll_red_nf_redsb
src/sage/rings/polynomial/multi_polynomial_sequence.py:            from brial.interred import interred as inter_red
src/sage/sat/converters/__init__.py:from brial.cnf import CNFEncoder as PolyBoRiCNFEncoder

but other suggestions based on best practices will be given the priority.

[mkoeppe]

So there's a Cython interface sage.libs.polybori, then a Python library brial, and also sage.rings.polynomial.pbori? What is depending on what?

[fbissey]

I thought that sage-brial depends on sage.libs.polybori but that may not be the case. It imports sage.all and specifically sage.rings.polynomial.pbori and that's the only sage specific bits it imports.

sage.rings.polynomial.pbori itself relies on sage.libs.polybori (through sage.libs.polybori.decl) and both need the C++ brial library.

Does that answer your question?

[mkoeppe]

Thanks. I agree that sage.rings.polynomial seems to be a good place for the Python code from sage-brial. Perhaps it would be best to create a package sage.rings.polynomial.brial and to move the current pbori.pyx inside it (leaving deprecated reimport behind). There are only a few places in Sage where things are imported from sage.rings.polynomial.pbori and they could be easily update. Or, if you don't mind keep the old name pbori, the same could be done with sage.rings.polynomial.pbori instead of ...brial.

One thing to keep in mind is the context of the planned modularization of sagelib. In Sage 9.3 I would hope to be able to create a separate distribution (distutils package) sage-brial, which would then package sage.rings.polynomial.{brial,pbori}, sage.libs.polybori, sage.libs.fes, sage.crypto.boolean_function -- everything that has a compile-time dependency on libbrial. This will make it possible to separately compile sagelib on systems without brial.

[fbissey]

On the legal side, the code is GPL2.0 or later. Although I will need to fix the license on github. It seems I changed it from 2+ to 2 carelessly back in 2017 (probably by copying a template). I obviously had no rights to do that.

The only thing I may need guidance with is the deprecated re-import. Otherwise moving stuff, including pbori.{pyx,pxd}, in sage.rings.polynomial.pbori seems to be a good idea and hopefully will help with modularization.

[mkoeppe]

Replying to fbissey:

On the legal side, the code is GPL2.0 or later. Although I will need to fix the license on github. It seems I changed it from 2+ to 2 carelessly back in 2017 (probably by copying a template). I obviously had no rights to do that.

I think GitHub takes the position that there is no such thing as "GPL 2 only": That the license includes "... or any later version" no matter whether you write these words in the files or not.

[fbissey]

Replying to mkoeppe:

Replying to fbissey:

On the legal side, the code is GPL2.0 or later. Although I will need to fix the license on github. It seems I changed it from 2+ to 2 carelessly back in 2017 (probably by copying a template). I obviously had no rights to do that.

I think GitHub takes the position that there is no such thing as "GPL 2 only": That the license includes "... or any later version" no matter whether you write these words in the files or not.

Very nice of them. But they are not lawyers and I think the "LICENSE" file should be explicit for people who look for these subtle clarification (distros). The GPL FAQ ​https://www.gnu.org/licenses/old-licenses/gpl-2.0-faq.html#VersionTwoOrLater seem to imply that you should explicitly state it.

[fbissey]

Tests included inside sage-brial will have to be moved to sage doctesting. More generally it will have to be documented properly as much as possible. This will be so much fun.

It looks like that will take much more than just drag and drop in place with a few cosmetic changes.

[mkoeppe]

I suppose you could prepare the change to using the sage doctester first, before merging into Sage.

[mkoeppe]

Given that sage_brial has no Cython bits, I think it could in fact be removed from the dependencies of sagelib, as it won't really be needed while installing sagelib, only later for docbuild and doctest.

[fbissey]

Replying to mkoeppe:

Given that sage_brial has no Cython bits, I think it could in fact be removed from the dependencies of sagelib, as it won't really be needed while installing sagelib, only later for docbuild and doctest.

Technically correct.

[fbissey]

This is not ready for review. I am just pushing my work in progress so it is not just on my workstation, it is visible and open to others contribution.

There is a lot to do:

1. update paths
2. convert docstrings and tests to sage framework

This may take some time and I doubt it will be ready in time for 9.2 but we can only try.

---

New commits:

​884ca66

WIP: move sage-brial files and pbori.{pxd,pyx} in sage/rings/polynomial/pbori/

[git]

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

​f947293	replace "from sage.rings.polynomial.pbori" with "from sage.rings.polynomial.pbori.pbori"
​e7653a3	Replace "from brial" with "from sage.rings.polynomial.pbori.brial"

[git]

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

​2226097

src/sage/rings/polynomial/pbori/__init__.py: Draft of lazy_import call

[git]

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

​d0bc786

transform python test into sage tests - step one.

[fbissey]

I am currently converting docstrings :( I am also doing some selective code deletion as I come along across it. There is a module for memory usage, it is imported once but never used and even less tested. So, I am removing it. It will be worth another look after that first pass.

[git]

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

​ae30cf3	Basic conversion to sage doctrings/doctesting. Remove original copyright/license functions.
​4c0d80f	Some basic docstring in general_boolean_polynomial.py - remove memusage.py and its only import.
​bdf1fec	First pass in converting to docstring and doctests. It is quite likely a lot of doctests are broken at this stage.

[git]

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

​2e94a77

Merge branch 'develop' into brial-merging

[fbissey]

I suspect the building of the html doc will break at this stage. If it doesn't the output may very well be useless. Once I figure that out, I'll do the doctesting to see what's broken. I am expecting a lot of breaking in the newly imported tests.

[git]

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

​4e3ed88

one too many pbori.

[git]

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

​563e959

add miising import of lazy_import

[git]

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

​7cb8ba1	remove excessive .brial from import. The initial replacement was wrong.
​408824d	Fix building of the pbori interface documentation. No "brial" documentation is buiilt as far as I can tell.

[git]

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

​443a3a8

Remove addition.py - nothing from this file is used and tests failures point to the file being obsolote.

[git]

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

​167ef09

migrate xrange (unsupported in python3) to range

[fbissey]

At this point documentation builds and is identical to the old one. doctesting gets a lot of failures. There is still quite a bit of cleaning to be done, including removing files that should have been trashed long ago.

[fbissey]

More joy! I found a subtle coding error while fixing the test suite. I guess it won't have an impact on the rest of sage because that code path just error-ed out when taken. So it is probably a dead code path anyway.

[git]

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

​afdbb9d	fix doctests in blocks.py
​ec4cddb	remove check_claims.py which appears to be a standalone script.
​d53714f	remove another script
​72f8a6a	Fix doctest and code in cnf.py
​e25dd81	Fix pbori related doctests and code in multi_polynomial_sequence

[git]

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

​2c9b36f	Remove unused files
​a0ccf77	removing unused files: were used by the previously removed files.

[git]

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

​b3c16a8	Revert "removing unused files: were used by the previously removed files." Turns out interred is used somewhere else.
​20d1f9e	Fix doctests in pbori.pyx

[git]

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

​bc1f9e2

fix more doctests

[git]

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

​b513981	more doctests fixes
​1cf36c7	Remove VariableFactory from the lazy_import list for now.
​64c57d2	Last bunch of doctest fixes

[fbissey]

This is ready for a first review. I am not expecting it to make it as is. But other people should comment now.

[git]

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

​878d941

Remove the sage_brial package since its meaningful content is merged.

[fbissey]

Last commit to burn the bridges :)

[mkoeppe]

BooleanPolynomialRing_constructor vs. BooleanPolynomialRing looks a bit confusing

[fbissey]

Replying to mkoeppe:

BooleanPolynomialRing_constructor vs. BooleanPolynomialRing looks a bit confusing

Not my code, I am only the messenger. I certainly agree that this largely decade old code would need a good fleeing - apart from the trimming I already did on import. I assumed we could do it after merging it.

[mkoeppe]

Sure, just trying to figure out what the high level interface is intended to be.

[mkoeppe]

build/pkgs/sagelib/dependencies still has sage_brial

[fbissey]

Right. I remember thinking about that last night. Forgot to remove it. Will do in a few hours if no one beats me to it.

[git]

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

​02a5af1

build/pkgs/sagelib/dependencies: Remove sage_brial

[tscrim]

Replying to fbissey:

Replying to mkoeppe:

BooleanPolynomialRing_constructor vs. BooleanPolynomialRing looks a bit confusing

Not my code, I am only the messenger. I certainly agree that this largely decade old code would need a good fleeing - apart from the trimming I already did on import. I assumed we could do it after merging it.

Indeed, this is a known problem that needs some attention: #21892, #24748, possibly #27019, #23310, #23312, possibly #23311, #15223.

My idea is to scrap BooleanPolynomialRing_constructor and just have BooleanPolynomialRing be a UniqueRepresentation object, possibly just being a subclass of the monoid algebra parent.

[fbissey]

My idea was to merge this in 9.2 and work on improving things in 9.3. However, if you know exactly what to do, we can work on this now.

[git]

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

​754dea9	Properly format tests in gbcore
​0c5fe64	Merge branch 'public/merge_sage_brial' of trac.sagemath.org:sage into brial-merging

[tscrim]

I think that is a better plan. I am too stretched for time right now to work on this.

[mkoeppe]

I agree with this plan - it will also be a good basis for modularization work in 9.3

[fbissey]

I am in the process of looking at adding some of the documentation from sage-brial to the sage documentation. At least to see how it looks like.

If you think that's not worth adding now, it is fine.

[git]

Branch pushed to git repo; I updated commit sha1 and set ticket back to needs_review. New commits:

​91fe5e1

Fix doctest and docstring formatting

[fbissey]

Just a touch up. One doctest was commented out so I decided to enable it. Building documentation revealed formatting issues in docstrings. All in the same file.

I built the doc with some added files from brial and we definitely should keep that for later work. Individual files need sensible title blocks for a start.