BooleanMonomial.index() SEGFAULT on constant monomials

[Bouillaguet]

sage: R.<x,y> = BooleanPolynomialRing()
sage: R(1).lm().index()

crashes miserably (from inside polybori, as I understand) with:

/usr/local/sage/local/lib/libcsage.so(print_backtrace+0x3b)[0x1cd9d3]
/usr/local/sage/local/lib/libcsage.so(sigdie+0x17)[0x1cda13]
/usr/local/sage/local/lib/libcsage.so(sage_signal_handler+0x20d)[0x1cd598]
[0xc27400]
/usr/local/sage/local/lib/python2.6/site-packages/sage/rings/polynomial/pbori.so(+0x13d90)[0x5235d90]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalFrameEx+0x482e)[0x2a55ee]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalCodeEx+0x7d1)[0x2a6fa1]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalCode+0x63)[0x2a70c3]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalFrameEx+0x50c9)[0x2a5e89]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalCodeEx+0x7d1)[0x2a6fa1]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalFrameEx+0x4370)[0x2a5130]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalCodeEx+0x7d1)[0x2a6fa1]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalFrameEx+0x4370)[0x2a5130]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalFrameEx+0x59e1)[0x2a67a1]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalCodeEx+0x7d1)[0x2a6fa1]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalFrameEx+0x4370)[0x2a5130]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalCodeEx+0x7d1)[0x2a6fa1]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalFrameEx+0x4370)[0x2a5130]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalCodeEx+0x7d1)[0x2a6fa1]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalFrameEx+0x4370)[0x2a5130]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalCodeEx+0x7d1)[0x2a6fa1]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyEval_EvalCode+0x63)[0x2a70c3]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyRun_FileExFlags+0xb1)[0x2c6f91]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyRun_SimpleFileExFlags+0xdc)[0x2c71cc]
/usr/local/sage/local/lib/libpython2.6.so.1.0(PyRun_AnyFileExFlags+0x7a)[0x2c75fa]
/usr/local/sage/local/lib/libpython2.6.so.1.0(Py_Main+0xc7d)[0x2d3c4d]
python(main+0x27)[0x8048657]
/lib/i386-linux-gnu/libc.so.6(__libc_start_main+0xf3)[0x405113]

------------------------------------------------------------------------
Unhandled SIGSEGV: A segmentation fault occurred in Sage.
This probably occurred because a *compiled* component of Sage has a bug
in it and is not properly wrapped with sig_on(), sig_off(). You might
want to run Sage under gdb with 'sage -gdb' to debug this.
Sage will now terminate.
------------------------------------------------------------------------

This seems to be a bug in PolyBori?. The patch I propose just checks that the monomial is not constant before calling the polybori function. This is a workaround though

[AlexanderDreyer]

Confirmed, this is a bug of both Sage's interface (it should not call firstIndex() in this case) as well as upstream PolyBoRi (it should not crash on inconsistent user commands). The patch solved the problem very well, but you should use is_one() instead of degree()==0. This performs better (degree() needs to count first).

[Bouillaguet]

I expected the bug to also occur with the zero monomial, and there is no is_constant() method that I could have used. However, the bug does not occur with zero, because "zero" is not a monomial :

sage: R(0).lm().parent()
Boolean PolynomialRing in x, y
sage: R(1).lm().parent()
MonomialMonoid of Boolean PolynomialRing in x, y

I wonder if this behavior is the best possible, but it does make sense. Patch updated

[Bouillaguet]

patch fixing the problem (with doctest)

[AlexanderDreyer]

Replying to Bouillaguet:

I expected the bug to also occur with the zero monomial, and there is no is_constant() method that I could have used. However, the bug does not occur with zero, because "zero" is not a monomial :

This is exactly the reason why there is no is_constant() routine.

sage: R(0).lm().parent()
Boolean PolynomialRing in x, y
sage: R(1).lm().parent()
MonomialMonoid of Boolean PolynomialRing in x, y

I wonder if this behavior is the best possible, but it does make sense. Patch updated

This is the best compromise: PolyBoRi's original interface throws a ValueError? when accessing zero polynomial leading monomials. (As you pointed out, there are no zero monomials.) But this would not be consistent with other polynomial types in Sage (e. g. Singular polynomials used for QQ[]). But these return polynomials (not monomials) anyway, so zero is not an issue there.

The patch is fine, I'll give it a positive review and update the pending spkg for PolyBoRi 0.8.2 at #13124 soon.

[AlexanderDreyer]

I also fixed #13124 accordingly. Without your patch, we'd have this then:

sage: R.<x,y> = BooleanPolynomialRing()
sage: R(1).lm().index()
2147483647

(final line platform dependent)