August 2010 - PPL-devel - BUGSENG Mailman

Problems with test suite on alpha
by Michael Tautschnig 09 Sep '10

09 Sep '10

Hi PPL developers, I've reported problems with some test cases because of suspected gcc bugs on alpha earlier already. Now it seems that even more test cases are broken, most probably because of the very same bug (I suspect it to be http://gcc.gnu.org/bugzilla/show_bug.cgi?id=8966). My question now is: How to proceed? Shall I specifically disable the failing test cases by surrounding them with an #ifndef __alpha__ or disable the entire test suite on alpha? Thanks, Michael

2 6

[GIT] ppl/ppl(master): Let the input routine for checked numbers accepts C99 hexadecimal float syntax .
by Enea Zaffanella 31 Aug '10

31 Aug '10

Module: ppl/ppl Branch: master Commit: c00c6ad02ef68fb19f2692ff0ff0d290d203673b URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=c00c6ad02ef6… Author: Enea Zaffanella <zaffanella(a)cs.unipr.it> Date: Tue Aug 31 22:20:12 2010 +0200 Let the input routine for checked numbers accepts C99 hexadecimal float syntax. --- src/Checked_Number.defs.hh | 6 +++--- src/checked.cc | 29 ++++++++++++++++++++++------- tests/Polyhedron/numberinput1.cc | 2 ++ 3 files changed, 27 insertions(+), 10 deletions(-) diff --git a/src/Checked_Number.defs.hh b/src/Checked_Number.defs.hh index 514732d..5bef261 100644 --- a/src/Checked_Number.defs.hh +++ b/src/Checked_Number.defs.hh @@ -960,9 +960,9 @@ num : unum | '+' | SIGN unum ; unum : unum1 EXP : 'e' - | HEX unum1 | '*^' - | base BASE unum1 ; - ; + | HEX unum1 | 'p' + | base BASE unum1 | '*^' + ; ; POINT : '.' unum1 : mantissa ; | mantissa EXP exponent diff --git a/src/checked.cc b/src/checked.cc index 9e32beb..c3ba12f 100644 --- a/src/checked.cc +++ b/src/checked.cc @@ -40,6 +40,7 @@ struct number_struct { bool neg_mantissa; bool neg_exponent; std::string mantissa; + unsigned int base_for_exponent; unsigned long exponent; }; @@ -100,7 +101,9 @@ parse_number_part(std::istream& is, number_struct& num) { bool empty_exponent = true; bool empty_mantissa = true; long exponent_offset = 0; + long exponent_offset_scale = 1; num.base = 10; + num.base_for_exponent = 10; num.neg_mantissa = false; num.neg_exponent = false; num.mantissa.erase(); @@ -144,6 +147,7 @@ parse_number_part(std::istream& is, number_struct& num) { int d = is.get(); if (d == 'x' || d == 'X') { num.base = 16; + num.base_for_exponent = 16; state = INTEGER; c = is.get(); } @@ -179,6 +183,7 @@ parse_number_part(std::istream& is, number_struct& num) { } if (num.base < 2) goto error; + num.base_for_exponent = num.base; num.mantissa.erase(); empty_mantissa = true; state = INTEGER; @@ -210,7 +215,16 @@ parse_number_part(std::istream& is, number_struct& num) { if (empty_mantissa) goto error; if (c == 'e' || c == 'E') - goto exp; + goto exp; + if (c == 'p' || c == 'P') { + if (num.base == 16) { + num.base_for_exponent = 2; + exponent_offset_scale = 4; + goto exp; + } + else + goto error; + } if (c == '*') { c = is.get(); if (c != '^') @@ -263,7 +277,7 @@ parse_number_part(std::istream& is, number_struct& num) { else neg = false; sum_sign(num.neg_exponent, num.exponent, - neg, exponent_offset); + neg, exponent_offset * exponent_offset_scale); return V_EQ; } @@ -292,7 +306,8 @@ parse_number(std::istream& is, number_struct& num, number_struct& den) { r = parse_number_part(is, den); if (r != V_EQ) return V_CVT_STR_UNK; - if (num.base == den.base) { + if (num.base == den.base + && num.base_for_exponent == den.base_for_exponent) { if (sum_sign(num.neg_exponent, num.exponent, !den.neg_exponent, den.exponent)) { if (num.neg_exponent) { @@ -333,14 +348,14 @@ input_mpq(mpq_class& to, std::istream& is) { mpz_t z; mpz_init(z); if (num_struct.exponent) { - mpz_ui_pow_ui(z, num_struct.base, num_struct.exponent); + mpz_ui_pow_ui(z, num_struct.base_for_exponent, num_struct.exponent); if (num_struct.neg_exponent) mpz_mul(den, den, z); else mpz_mul(num, num, z); } if (den_struct.exponent) { - mpz_ui_pow_ui(z, den_struct.base, den_struct.exponent); + mpz_ui_pow_ui(z, den_struct.base_for_exponent, den_struct.exponent); if (den_struct.neg_exponent) mpz_mul(num, num, z); else @@ -355,13 +370,13 @@ input_mpq(mpq_class& to, std::istream& is) { if (num_struct.exponent) { if (num_struct.neg_exponent) { // Add the negative exponent as a denominator. - mpz_ui_pow_ui(den, num_struct.base, num_struct.exponent); + mpz_ui_pow_ui(den, num_struct.base_for_exponent, num_struct.exponent); goto end; } // Multiply the exponent into the numerator. mpz_t z; mpz_init(z); - mpz_ui_pow_ui(z, num_struct.base, num_struct.exponent); + mpz_ui_pow_ui(z, num_struct.base_for_exponent, num_struct.exponent); mpz_mul(num, num, z); mpz_clear(z); } diff --git a/tests/Polyhedron/numberinput1.cc b/tests/Polyhedron/numberinput1.cc index e98a2f7..cdc43f1 100644 --- a/tests/Polyhedron/numberinput1.cc +++ b/tests/Polyhedron/numberinput1.cc @@ -192,6 +192,8 @@ test07() { && aux_test("0xx", "nan", "x", V_CVT_STR_UNK) && aux_test("0x0.f", "15/16", "", V_EQ) && aux_test("0x.f", "15/16", "", V_EQ) + && aux_test("0x.fp3", "15/2", "", V_EQ) + && aux_test("0x100p-9", "1/2", "", V_EQ) && aux_test("0x.f*^1", "15", "", V_EQ) && aux_test("0x-f", "nan", "-f", V_CVT_STR_UNK) && aux_test("0xfa", "250", "", V_EQ)

1 0

[GIT] ppl/ppl(sparse_matrices): Merge branch 'master' into sparse_matrices
by Roberto Bagnara 30 Aug '10

30 Aug '10

Module: ppl/ppl Branch: sparse_matrices Commit: 2383538528e5352f404923d2b919fe970658611b URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=2383538528e5… Author: Roberto Bagnara <bagnara(a)cs.unipr.it> Date: Mon Aug 30 09:24:00 2010 +0200 Merge branch 'master' into sparse_matrices ---

1 0

[GIT] ppl/ppl(sparse_matrices): Use a convenience library to confine the compilation of glpk_set_d_eps.c.
by Roberto Bagnara 30 Aug '10

30 Aug '10

Module: ppl/ppl Branch: sparse_matrices Commit: eac65c875791c7ad6065a918cada3783f487f706 URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=eac65c875791… Author: Roberto Bagnara <bagnara(a)cs.unipr.it> Date: Mon Aug 30 09:20:53 2010 +0200 Use a convenience library to confine the compilation of glpk_set_d_eps.c. --- demos/ppl_lpsol/Makefile.am | 20 ++++++++++++-------- 1 files changed, 12 insertions(+), 8 deletions(-) diff --git a/demos/ppl_lpsol/Makefile.am b/demos/ppl_lpsol/Makefile.am index a8c234f..1365fa2 100644 --- a/demos/ppl_lpsol/Makefile.am +++ b/demos/ppl_lpsol/Makefile.am @@ -42,27 +42,31 @@ AM_CPPFLAGS = \ -I$(top_builddir)/interfaces/C \ @extra_includes@ -if GCC - -# We cannot use -std=c89 because of the definition of sigaction in <signal.h>. -AM_CFLAGS = -pedantic -std=gnu89 -Werror - -endif GCC - bin_PROGRAMS = ppl_lpsol # dummy.cc serves the only purpose to force linking using the C++ compiler. ppl_lpsol_SOURCES = \ ppl_lpsol.c \ -glpk_set_d_eps.c \ dummy.cc +if GCC + +# We cannot use -std=c89 because of the definition of sigaction in <signal.h>. +ppl_lpsol_CFLAGS = -pedantic -std=gnu89 -Werror + +endif GCC + ppl_lpsol_LDADD = \ +libglpk_set_d_eps.a \ -lglpk \ $(top_builddir)/interfaces/C/libppl_c.la \ $(top_builddir)/src/libppl.la \ @extra_libraries@ +noinst_LIBRARIES = libglpk_set_d_eps.a +libglpk_set_d_eps_a_SOURCES = \ +glpk_set_d_eps.c + dist_man_MANS = \ ppl_lpsol.1

1 0

[GIT] ppl/ppl(master): Use a convenience library to confine the compilation of glpk_set_d_eps.c.
by Roberto Bagnara 30 Aug '10

30 Aug '10

Module: ppl/ppl Branch: master Commit: eac65c875791c7ad6065a918cada3783f487f706 URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=eac65c875791… Author: Roberto Bagnara <bagnara(a)cs.unipr.it> Date: Mon Aug 30 09:20:53 2010 +0200 Use a convenience library to confine the compilation of glpk_set_d_eps.c. --- demos/ppl_lpsol/Makefile.am | 20 ++++++++++++-------- 1 files changed, 12 insertions(+), 8 deletions(-) diff --git a/demos/ppl_lpsol/Makefile.am b/demos/ppl_lpsol/Makefile.am index a8c234f..1365fa2 100644 --- a/demos/ppl_lpsol/Makefile.am +++ b/demos/ppl_lpsol/Makefile.am @@ -42,27 +42,31 @@ AM_CPPFLAGS = \ -I$(top_builddir)/interfaces/C \ @extra_includes@ -if GCC - -# We cannot use -std=c89 because of the definition of sigaction in <signal.h>. -AM_CFLAGS = -pedantic -std=gnu89 -Werror - -endif GCC - bin_PROGRAMS = ppl_lpsol # dummy.cc serves the only purpose to force linking using the C++ compiler. ppl_lpsol_SOURCES = \ ppl_lpsol.c \ -glpk_set_d_eps.c \ dummy.cc +if GCC + +# We cannot use -std=c89 because of the definition of sigaction in <signal.h>. +ppl_lpsol_CFLAGS = -pedantic -std=gnu89 -Werror + +endif GCC + ppl_lpsol_LDADD = \ +libglpk_set_d_eps.a \ -lglpk \ $(top_builddir)/interfaces/C/libppl_c.la \ $(top_builddir)/src/libppl.la \ @extra_libraries@ +noinst_LIBRARIES = libglpk_set_d_eps.a +libglpk_set_d_eps_a_SOURCES = \ +glpk_set_d_eps.c + dist_man_MANS = \ ppl_lpsol.1

1 0

[GIT] ppl/ppl(sparse_matrices): Merge branch 'master' into sparse_matrices
by Roberto Bagnara 29 Aug '10

29 Aug '10

Module: ppl/ppl Branch: sparse_matrices Commit: 23a67cd07cf8ffa710f3c23ba6ee182aa84d7a57 URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=23a67cd07cf8… Author: Roberto Bagnara <bagnara(a)cs.unipr.it> Date: Sun Aug 29 14:32:58 2010 +0200 Merge branch 'master' into sparse_matrices ---

1 0

[GIT] ppl/ppl(sparse_matrices): Started revising the expected results.
by Roberto Bagnara 29 Aug '10

29 Aug '10

Module: ppl/ppl Branch: sparse_matrices Commit: f078b9a6bcc0c1fa82d474e676e82a8e5dd8d319 URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=f078b9a6bcc0… Author: Roberto Bagnara <bagnara(a)cs.unipr.it> Date: Sun Aug 29 14:32:07 2010 +0200 Started revising the expected results. --- demos/ppl_lpsol/.gitignore | 2 +- demos/ppl_lpsol/Makefile.am | 44 +- demos/ppl_lpsol/expected_mpz | 1852 +++++++++++++++++++++++++++++++--------- demos/ppl_lpsol/expected_mpz_a | 1796 +++++++++++++++++++++++++++++++-------- 4 files changed, 2912 insertions(+), 782 deletions(-) Diff: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commitdiff;h=f078b9a6…

1 0

[GIT] ppl/ppl(sparse_matrices): For checking purposes, use the exact LP solver of GLPK.
by Roberto Bagnara 29 Aug '10

29 Aug '10

Module: ppl/ppl Branch: sparse_matrices Commit: 108769b7f3ca869c9c852da8f0fdaed22d4e6d64 URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=108769b7f3ca… Author: Roberto Bagnara <bagnara(a)cs.unipr.it> Date: Sun Aug 29 14:10:03 2010 +0200 For checking purposes, use the exact LP solver of GLPK. --- demos/ppl_lpsol/ppl_lpsol.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/demos/ppl_lpsol/ppl_lpsol.c b/demos/ppl_lpsol/ppl_lpsol.c index 5b8abed..abf9fbd 100644 --- a/demos/ppl_lpsol/ppl_lpsol.c +++ b/demos/ppl_lpsol/ppl_lpsol.c @@ -589,7 +589,7 @@ maybe_check_results(const int ppl_status, const double ppl_optimum_value) { /* Set the problem class to LP: MIP problems are thus treated as LP ones. */ lpx_set_class(glpk_lp, LPX_LP); - lpx_simplex(glpk_lp); + lpx_exact(glpk_lp); glpk_status = lpx_get_status(glpk_lp); } else {

1 0

[GIT] ppl/ppl(sparse_matrices): When the `--check' option is used, input data is perturbed the same way as GLPK does.
by Roberto Bagnara 29 Aug '10

29 Aug '10

Module: ppl/ppl Branch: sparse_matrices Commit: 12f60158e2122303413a67c19425ba8714196f1e URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=12f60158e212… Author: Roberto Bagnara <bagnara(a)cs.unipr.it> Date: Sun Aug 29 12:10:11 2010 +0200 When the `--check' option is used, input data is perturbed the same way as GLPK does. --- README | 3 +- demos/ppl_lpsol/Makefile.am | 1 + demos/ppl_lpsol/glpk_set_d_eps.c | 285 ++++++++++++++++++++++++++++++++++++++ demos/ppl_lpsol/ppl_lpsol.1 | 7 +- demos/ppl_lpsol/ppl_lpsol.c | 41 ++++-- doc/ppl-config.1 | 2 +- 6 files changed, 319 insertions(+), 20 deletions(-) Diff: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commitdiff;h=12f60158…

1 0

[GIT] ppl/ppl(sparse_matrices): Do also read the inhomogeneous term of the objective function.
by Enea Zaffanella 29 Aug '10

29 Aug '10

Module: ppl/ppl Branch: sparse_matrices Commit: c82a27391846139072cadf0754f59f4805f4f99b URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=c82a27391846… Author: Enea Zaffanella <zaffanella(a)cs.unipr.it> Date: Sun Aug 29 11:01:58 2010 +0200 Do also read the inhomogeneous term of the objective function. This should allow for a (almost) meaningful comparison with the optimizied value computed by glpk when using command line option -c. Also corrected a minor bug in the output routine for the objective function (an open parenthesis was sometimes missing). --- demos/ppl_lpsol/ppl_lpsol.c | 9 ++++++++- 1 files changed, 8 insertions(+), 1 deletions(-) diff --git a/demos/ppl_lpsol/ppl_lpsol.c b/demos/ppl_lpsol/ppl_lpsol.c index 3828f6e..fb44b4a 100644 --- a/demos/ppl_lpsol/ppl_lpsol.c +++ b/demos/ppl_lpsol/ppl_lpsol.c @@ -1172,7 +1172,12 @@ solve(char* file_name) { /* Set the ppl_objective_le to be the objective function. */ ppl_new_Linear_Expression_with_dimension(&ppl_objective_le, dimension); - /* The inhomogeneous term is completely useless for our purpose. */ + /* Set value for objective function's inhomogeneous term. */ + mpz_mul(tmp_z, den_lcm, mpq_numref(objective[0])); + mpz_divexact(tmp_z, tmp_z, mpq_denref(objective[0])); + ppl_assign_Coefficient_from_mpz_t(ppl_coeff, tmp_z); + ppl_Linear_Expression_add_to_inhomogeneous(ppl_objective_le, ppl_coeff); + /* Set values for objective function's variable coefficients. */ for (i = 1; i <= dimension; ++i) { mpz_mul(tmp_z, den_lcm, mpq_numref(objective[i])); mpz_divexact(tmp_z, tmp_z, mpq_denref(objective[i])); @@ -1182,6 +1187,8 @@ solve(char* file_name) { if (verbosity >= 4) { fprintf(output_file, "Objective function:\n"); + if (mpz_cmp_si(den_lcm, 1) != 0) + fprintf(output_file, "("); ppl_io_fprint_Linear_Expression(output_file, ppl_objective_le); }

1 0