- PPL-devel - BUGSENG Mailman

[GIT] ppl/ppl(floating_point): Started rewriting tests on digital filters.
by Fabio Bossi 30 Jul '10

30 Jul '10

Module: ppl/ppl Branch: floating_point Commit: e85af354adc157725146aa506a6dc2f9507407e4 URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=e85af354adc1… Author: Fabio Bossi <bossi(a)cs.unipr.it> Date: Fri Jul 30 17:54:02 2010 +0200 Started rewriting tests on digital filters. --- tests/Concrete_Expression/Makefile.am | 2 +- tests/Concrete_Expression/digitalfilters1.cc | 13 ++++++++++--- 2 files changed, 11 insertions(+), 4 deletions(-) diff --git a/tests/Concrete_Expression/Makefile.am b/tests/Concrete_Expression/Makefile.am index 3c64827..dd5983f 100644 --- a/tests/Concrete_Expression/Makefile.am +++ b/tests/Concrete_Expression/Makefile.am @@ -52,11 +52,11 @@ $(top_builddir)/src/libppl.la \ ORIGINAL_TESTS = \ C_Expr \ +digitalfilters1 \ linearform1 \ linearize #bdshape1 \ #bdshape2 \ -#digitalfilters1 \ #floatingpointexpr1 \ #octagonalshape1 \ #octagonalshape2 \ diff --git a/tests/Concrete_Expression/digitalfilters1.cc b/tests/Concrete_Expression/digitalfilters1.cc index 41c6229..204ef54 100644 --- a/tests/Concrete_Expression/digitalfilters1.cc +++ b/tests/Concrete_Expression/digitalfilters1.cc @@ -25,6 +25,10 @@ site: http://www.cs.unipr.it/ppl/ . */ namespace { using namespace Parma_Polyhedra_Library::IO_Operators; +Concrete_Expression_Type FP_Type = + Concrete_Expression_Type::floating_point(ANALYZED_FP_FORMAT); + +typedef Integer_Interval_Type Int_Interval; /* This file tests a rate limiter that, given random input flows of floating @@ -105,7 +109,7 @@ test01() { // D = [0, 16]; tmp.lower() = 0; tmp.upper() = 16; - abstract_store.affine_form_image(X, FP_Linear_Form(tmp)); + abstract_store.affine_form_image(D, FP_Linear_Form(tmp)); // S = Y; abstract_store.affine_form_image(S, FP_Linear_Form(Y)); @@ -193,7 +197,7 @@ test02() { tmp.upper() = 16; abstract_store.affine_form_image(D, FP_Linear_Form(tmp)); - // S = tmp; + // S = Y; abstract_store.affine_form_image(S, FP_Linear_Form(Y)); // R = X - S; @@ -419,6 +423,7 @@ test04() { return (tmp.lower() == -128 && tmp.upper() == 128); } +/* // Tests rate limiter using bounded differences and linearization of // floating point expressions. // In order to improve the analysis, the interval domain is used @@ -976,6 +981,7 @@ test07() { nout << "*** Y in " << tmp << " ***" << endl; return (tmp.lower() == -144 && tmp.upper() == 144); } +*/ } // namespace @@ -984,7 +990,7 @@ BEGIN_MAIN DO_TEST_F8(test02); DO_TEST_F8(test03); DO_TEST_F64A(test04); - +/* #define COND_float PPL_CPP_EQ(PPL_CPP_FP_FORMAT(ANALYZER_FP_FORMAT), 1) #define COND_double PPL_CPP_EQ(PPL_CPP_FP_FORMAT(ANALYZER_FP_FORMAT), 2) #define COND_float_or_double PPL_CPP_OR(COND_float, COND_double) @@ -999,4 +1005,5 @@ BEGIN_MAIN COND_DO_TEST(PPL_CUSTOM_COND, test06); DO_TEST_F64(test07); +*/ END_MAIN

1 0

[GIT] ppl/ppl(floating_point): Do not add errors after rounding to a less precise format.
by Fabio Bossi 30 Jul '10

30 Jul '10

Module: ppl/ppl Branch: floating_point Commit: 8cfeaf81d2f501d2040e6af07815763173d7357c URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=8cfeaf81d2f5… Author: Fabio Bossi <bossi(a)cs.unipr.it> Date: Fri Jul 30 17:04:16 2010 +0200 Do not add errors after rounding to a less precise format. --- src/Float.defs.hh | 6 ++++++ src/linearize.hh | 5 +++-- 2 files changed, 9 insertions(+), 2 deletions(-) diff --git a/src/Float.defs.hh b/src/Float.defs.hh index 75d2c2b..219e7d3 100644 --- a/src/Float.defs.hh +++ b/src/Float.defs.hh @@ -58,6 +58,7 @@ struct float_ieee754_half { static const int EXPONENT_MIN = -EXPONENT_MAX + 1; static const int EXPONENT_MIN_DENORM = EXPONENT_MIN - static_cast<int>(MANTISSA_BITS); + static const Floating_Point_Format floating_point_format = IEEE754_HALF; int is_inf() const; int is_nan() const; int is_zero() const; @@ -90,6 +91,7 @@ struct float_ieee754_single { static const int EXPONENT_MIN = -EXPONENT_MAX + 1; static const int EXPONENT_MIN_DENORM = EXPONENT_MIN - static_cast<int>(MANTISSA_BITS); + static const Floating_Point_Format floating_point_format = IEEE754_SINGLE; int is_inf() const; int is_nan() const; int is_zero() const; @@ -135,6 +137,7 @@ struct float_ieee754_double { static const int EXPONENT_MIN = -EXPONENT_MAX + 1; static const int EXPONENT_MIN_DENORM = EXPONENT_MIN - static_cast<int>(MANTISSA_BITS); + static const Floating_Point_Format floating_point_format = IEEE754_DOUBLE; int is_inf() const; int is_nan() const; int is_zero() const; @@ -166,6 +169,7 @@ struct float_ibm_single { static const int EXPONENT_MIN = -EXPONENT_MAX + 1; static const int EXPONENT_MIN_DENORM = EXPONENT_MIN - static_cast<int>(MANTISSA_BITS); + static const Floating_Point_Format floating_point_format = IBM_SINGLE; int is_inf() const; int is_nan() const; int is_zero() const; @@ -217,6 +221,8 @@ struct float_intel_double_extended { static const int EXPONENT_MIN = -EXPONENT_MAX + 1; static const int EXPONENT_MIN_DENORM = EXPONENT_MIN - static_cast<int>(MANTISSA_BITS); + static const Floating_Point_Format floating_point_format = + INTEL_DOUBLE_EXTENDED; int is_inf() const; int is_nan() const; int is_zero() const; diff --git a/src/linearize.hh b/src/linearize.hh index 0093586..d21e5cb 100644 --- a/src/linearize.hh +++ b/src/linearize.hh @@ -629,8 +629,9 @@ cast_linearize(const Cast_Operator<Target>& cast_expr, } else { result = FP_Linear_Form(FP_Interval_Type(cast_arg->get_integer_interval())); - /* FIXME: we can avoid adding errors if FP_Interval_Type::boundary_type - is less precise than analyzed_format. */ + if (is_less_precise_than(Float<analyzer_format>::Binary::floating_point_format, analyzed_format)) + // We are rounding to a less precise format. Do not add errors. + return true; } FP_Linear_Form rel_error;

1 0

[GIT] ppl/ppl(master): Moved two methods in the right place.
by Abramo Bagnara 30 Jul '10

30 Jul '10

Module: ppl/ppl Branch: master Commit: afcafb8e2d3c9a8c1da3ac005d86e66819e22059 URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=afcafb8e2d3c… Author: Abramo Bagnara <abramo.bagnara(a)gmail.com> Date: Fri Jul 30 16:35:04 2010 +0200 Moved two methods in the right place. --- src/Interval.defs.hh | 46 +-------------------------------------- src/Interval.templates.hh | 52 +++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 54 insertions(+), 44 deletions(-) diff --git a/src/Interval.defs.hh b/src/Interval.defs.hh index ac6ce24..e751eb7 100644 --- a/src/Interval.defs.hh +++ b/src/Interval.defs.hh @@ -296,28 +296,7 @@ public: template <typename C> typename Enable_If<Is_Same_Or_Derived<I_Constraint_Base, C>::value, I_Result>::type - lower_extend(const C& c) { - PPL_ASSERT(OK()); - bool open; - switch (c.rel()) { - case V_LGE: - return lower_extend(); - case V_NAN: - return I_NOT_EMPTY | I_EXACT | I_UNCHANGED; - case V_GT: - open = true; - break; - case V_GE: - case V_EQ: - open = false; - break; - default: - PPL_ASSERT(false); - } - min_assign(LOWER, lower(), info(), LOWER, c.value(), f_info(c.value(), open)); - PPL_ASSERT(OK()); - return I_ANY; - } + lower_extend(const C& c); I_Result upper_extend() { info().clear_boundary_properties(UPPER); @@ -327,28 +306,7 @@ public: template <typename C> typename Enable_If<Is_Same_Or_Derived<I_Constraint_Base, C>::value, I_Result>::type - upper_extend(const C& c) { - PPL_ASSERT(OK()); - bool open; - switch (c.rel()) { - case V_LGE: - return lower_extend(); - case V_NAN: - return I_NOT_EMPTY | I_EXACT | I_UNCHANGED; - case V_LT: - open = true; - break; - case V_LE: - case V_EQ: - open = false; - break; - default: - PPL_ASSERT(false); - } - max_assign(UPPER, upper(), info(), UPPER, c.value(), f_info(c.value(), open)); - PPL_ASSERT(OK()); - return I_ANY; - } + upper_extend(const C& c); I_Result build() { return assign(UNIVERSE); diff --git a/src/Interval.templates.hh b/src/Interval.templates.hh index 0c17b64..456458e 100644 --- a/src/Interval.templates.hh +++ b/src/Interval.templates.hh @@ -28,6 +28,58 @@ site: http://www.cs.unipr.it/ppl/ . */ namespace Parma_Polyhedra_Library { template <typename Boundary, typename Info> +template <typename C> +typename Enable_If<Is_Same_Or_Derived<I_Constraint_Base, C>::value, I_Result>::type +Interval<Boundary, Info>::lower_extend(const C& c) { + PPL_ASSERT(OK()); + bool open; + switch (c.rel()) { + case V_LGE: + return lower_extend(); + case V_NAN: + return I_NOT_EMPTY | I_EXACT | I_UNCHANGED; + case V_GT: + open = true; + break; + case V_GE: + case V_EQ: + open = false; + break; + default: + PPL_ASSERT(false); + } + min_assign(LOWER, lower(), info(), LOWER, c.value(), f_info(c.value(), open)); + PPL_ASSERT(OK()); + return I_ANY; +} + +template <typename Boundary, typename Info> +template <typename C> +typename Enable_If<Is_Same_Or_Derived<I_Constraint_Base, C>::value, I_Result>::type +Interval<Boundary, Info>::upper_extend(const C& c) { + PPL_ASSERT(OK()); + bool open; + switch (c.rel()) { + case V_LGE: + return lower_extend(); + case V_NAN: + return I_NOT_EMPTY | I_EXACT | I_UNCHANGED; + case V_LT: + open = true; + break; + case V_LE: + case V_EQ: + open = false; + break; + default: + PPL_ASSERT(false); + } + max_assign(UPPER, upper(), info(), UPPER, c.value(), f_info(c.value(), open)); + PPL_ASSERT(OK()); + return I_ANY; +} + +template <typename Boundary, typename Info> template <typename From, typename Iterator> typename Enable_If<Is_Interval<From>::value, void>::type Interval<Boundary, Info>::CC76_widening_assign(const From& y,

1 0

[GIT] ppl/ppl(floating_point): Avoid adding errors when casting to a more precise
by Fabio Bossi 30 Jul '10

30 Jul '10

Module: ppl/ppl Branch: floating_point Commit: a69cc00cdd67f30135df670f9ebb4b2061c9fb35 URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=a69cc00cdd67… Author: Fabio Bossi <bossi(a)cs.unipr.it> Date: Fri Jul 30 16:04:19 2010 +0200 Avoid adding errors when casting to a more precise floating point format. --- src/Float.defs.hh | 6 ++++++ src/Float.inlines.hh | 6 +++++- src/linearize.hh | 13 ++++++++++--- 3 files changed, 21 insertions(+), 4 deletions(-) diff --git a/src/Float.defs.hh b/src/Float.defs.hh index e484458..75d2c2b 100644 --- a/src/Float.defs.hh +++ b/src/Float.defs.hh @@ -361,6 +361,12 @@ public: #endif /*! \brief + Returns <CODE>true</CODE> if and only if there is some floating point + number that is representable by \p f2 but not by \p f1. +*/ +bool is_less_precise_than(Floating_Point_Format f1, Floating_Point_Format f2); + +/*! \brief Computes the absolute error of floating point computations. \par Template type parameters diff --git a/src/Float.inlines.hh b/src/Float.inlines.hh index 16e0f21..13692b8 100644 --- a/src/Float.inlines.hh +++ b/src/Float.inlines.hh @@ -444,6 +444,11 @@ float_ieee754_quad::build(bool negative, mpz_t mantissa, int exponent) { << (MANTISSA_BITS - 64); } +inline bool +is_less_precise_than(Floating_Point_Format f1, Floating_Point_Format f2) { + return f1 < f2; +} + #if PPL_SUPPORTED_FLOAT inline Float<float>::Float() { @@ -492,7 +497,6 @@ Float<long double>::value() { } #endif - } // namespace Parma_Polyhedra_Library #endif // !defined(PPL_Float_inlines_hh) diff --git a/src/linearize.hh b/src/linearize.hh index a4f131d..0093586 100644 --- a/src/linearize.hh +++ b/src/linearize.hh @@ -616,16 +616,23 @@ cast_linearize(const Cast_Operator<Target>& cast_expr, typedef Box<FP_Interval_Type> FP_Interval_Abstract_Store; typedef std::map<dimension_type, FP_Linear_Form> FP_Linear_Form_Abstract_Store; + Floating_Point_Format analyzed_format = + cast_expr.type().floating_point_format(); const Concrete_Expression<Target>* cast_arg = cast_expr.argument(); if (cast_arg->type().is_floating_point()) { if (!linearize(*cast_arg, int_store, lf_store, result)) return false; + if (!is_less_precise_than(analyzed_format, + cast_arg->type().floating_point_format())) + // We are casting to a more precise format. Do not add errors. + return true; } - else + else { result = FP_Linear_Form(FP_Interval_Type(cast_arg->get_integer_interval())); + /* FIXME: we can avoid adding errors if FP_Interval_Type::boundary_type + is less precise than analyzed_format. */ + } - Floating_Point_Format analyzed_format = - cast_expr.type().floating_point_format(); FP_Linear_Form rel_error; result.relative_error(analyzed_format, rel_error); result += rel_error;

1 0

[GIT] ppl/ppl(floating_point): Removed a test file.
by Fabio Bossi 30 Jul '10

30 Jul '10

Module: ppl/ppl Branch: floating_point Commit: fcb542b0c0c0beca4216ece67163b0013ec2b61d URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=fcb542b0c0c0… Author: Fabio Bossi <bossi(a)cs.unipr.it> Date: Fri Jul 30 10:49:47 2010 +0200 Removed a test file. --- tests/Concrete_Expression/Makefile.am | 3 -- tests/Concrete_Expression/float.cc | 44 --------------------------------- 2 files changed, 0 insertions(+), 47 deletions(-) diff --git a/tests/Concrete_Expression/Makefile.am b/tests/Concrete_Expression/Makefile.am index ead3d19..3c64827 100644 --- a/tests/Concrete_Expression/Makefile.am +++ b/tests/Concrete_Expression/Makefile.am @@ -52,7 +52,6 @@ $(top_builddir)/src/libppl.la \ ORIGINAL_TESTS = \ C_Expr \ -float \ linearform1 \ linearize #bdshape1 \ @@ -140,8 +139,6 @@ bdshape2_SOURCES = bdshape2.cc digitalfilters1_SOURCES = digitalfilters1.cc -float_SOURCES = float.cc - floatingpointexpr1_SOURCES = floatingpointexpr1.cc linearform1_SOURCES = linearform1.cc diff --git a/tests/Concrete_Expression/float.cc b/tests/Concrete_Expression/float.cc deleted file mode 100644 index 7506993..0000000 --- a/tests/Concrete_Expression/float.cc +++ /dev/null @@ -1,44 +0,0 @@ -/* Test function compute_absolute_error. - Copyright (C) 2001-2010 Roberto Bagnara <bagnara(a)cs.unipr.it> - -This file is part of the Parma Polyhedra Library (PPL). - -The PPL is free software; you can redistribute it and/or modify it -under the terms of the GNU General Public License as published by the -Free Software Foundation; either version 3 of the License, or (at your -option) any later version. - -The PPL is distributed in the hope that it will be useful, but WITHOUT -ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or -FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License -for more details. - -You should have received a copy of the GNU General Public License -along with this program; if not, write to the Free Software Foundation, -Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111-1307, USA. - -For the most up-to-date information see the Parma Polyhedra Library -site: http://www.cs.unipr.it/ppl/ . */ - -#include "ppl_test.hh" - -namespace { - -using namespace Parma_Polyhedra_Library::IO_Operators; - -bool -test01() { - std::cout << compute_absolute_error<FP_Interval>(ANALYZED_FP_FORMAT) << - std::endl; - - std::cout << compute_absolute_error<FP_Interval>(ANALYZED_FP_FORMAT) << - std::endl; - - return 1; -} - -} - -BEGIN_MAIN - DO_TEST(test01); -END_MAIN

1 0

[GIT] ppl/ppl(floating_point): Added linearization of cast expressions along with a
by Fabio Bossi 29 Jul '10

29 Jul '10

Module: ppl/ppl Branch: floating_point Commit: e7b16c31a4cbd4f654222054b031ab3ec437bd68 URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=e7b16c31a4cb… Author: Fabio Bossi <bossi(a)cs.unipr.it> Date: Thu Jul 29 16:48:03 2010 +0200 Added linearization of cast expressions along with a simple test. Consistently use typedefs. --- src/Concrete_Expression.defs.hh | 4 +- src/linearize.hh | 77 +++++++++++++++++++++++++-- tests/Concrete_Expression/C_Expr.defs.hh | 22 ++++---- tests/Concrete_Expression/C_Expr.inlines.hh | 16 +++--- tests/Concrete_Expression/linearize.cc | 27 +++++++++- 5 files changed, 120 insertions(+), 26 deletions(-) Diff: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commitdiff;h=e7b16c31…

1 0

[GIT] ppl/ppl(floating_point): Corrected the policy values.
by Fabio Bossi 29 Jul '10

29 Jul '10

Module: ppl/ppl Branch: floating_point Commit: 253cc5c29243856d96d5d73b9d5acfc18e4d5a39 URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=253cc5c29243… Author: Fabio Bossi <bossi(a)cs.unipr.it> Date: Thu Jul 29 15:34:08 2010 +0200 Corrected the policy values. Make check now works again. --- src/Concrete_Expression.defs.hh | 6 +++--- 1 files changed, 3 insertions(+), 3 deletions(-) diff --git a/src/Concrete_Expression.defs.hh b/src/Concrete_Expression.defs.hh index dc937ad..7aa09e7 100644 --- a/src/Concrete_Expression.defs.hh +++ b/src/Concrete_Expression.defs.hh @@ -117,13 +117,13 @@ private: }; struct Integer_Interval_Info_Policy { - const_bool_nodef(store_special, false); - const_bool_nodef(store_open, true); + const_bool_nodef(store_special, true); + const_bool_nodef(store_open, false); const_bool_nodef(cache_empty, true); const_bool_nodef(cache_singleton, true); const_bool_nodef(cache_normalized, false); const_int_nodef(next_bit, 0); - const_bool_nodef(may_be_empty, false); + const_bool_nodef(may_be_empty, true); const_bool_nodef(may_contain_infinity, false); const_bool_nodef(check_empty_result, false); const_bool_nodef(check_inexact, false);

1 0

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

29 Jul '10

Module: ppl/ppl Branch: floating_point Commit: 3e682219b5f19c1ce1cd7a849c7d55d2524fedc7 URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=3e682219b5f1… Author: Roberto Bagnara <bagnara(a)cs.unipr.it> Date: Thu Jul 29 14:15:40 2010 +0200 Merge branch 'master' into floating_point ---

1 0

[GIT] ppl/ppl(floating_point): Unwanted assertions removed.
by Roberto Bagnara 29 Jul '10

29 Jul '10

Module: ppl/ppl Branch: floating_point Commit: ab1d650d0dcb0d122abf3d965c16ebdf94ac3325 URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=ab1d650d0dcb… Author: Roberto Bagnara <bagnara(a)cs.unipr.it> Date: Thu Jul 29 14:15:10 2010 +0200 Unwanted assertions removed. --- src/DB_Row.templates.hh | 2 -- 1 files changed, 0 insertions(+), 2 deletions(-) diff --git a/src/DB_Row.templates.hh b/src/DB_Row.templates.hh index 8684c04..765e719 100644 --- a/src/DB_Row.templates.hh +++ b/src/DB_Row.templates.hh @@ -39,7 +39,6 @@ DB_Row_Impl_Handler<T>::Impl::construct_upward_approximation(const U& y) { bump_size(); } #else // PPL_CXX_SUPPORTS_FLEXIBLE_ARRAYS - PPL_ASSERT(y_size > 0); if (y_size > 0) { assign_r(vec_[0], y[0], ROUND_UP); bump_size(); @@ -100,7 +99,6 @@ DB_Row_Impl_Handler<T>::Impl::copy_construct_coefficients(const Impl& y) { bump_size(); } #else // PPL_CXX_SUPPORTS_FLEXIBLE_ARRAYS - PPL_ASSERT(y_size > 0); if (y_size > 0) { vec_[0] = y.vec_[0]; bump_size();

1 0

[GIT] ppl/ppl(floating_point): Fixed a bug in DB_Row_Impl_Handler<T>::Impl: :expand_within_capacity(dimension_type).
by Roberto Bagnara 29 Jul '10

29 Jul '10

Module: ppl/ppl Branch: floating_point Commit: 4a26e69ad62c5c2b9d256f20392289d3161fc7bb URL: http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=4a26e69ad62c… Author: Roberto Bagnara <bagnara(a)cs.unipr.it> Date: Thu Jul 29 11:24:08 2010 +0200 Fixed a bug in DB_Row_Impl_Handler<T>::Impl::expand_within_capacity(dimension_type). This affected only compilers not supporting flexible arrays. --- src/DB_Row.templates.hh | 6 ++++-- 1 files changed, 4 insertions(+), 2 deletions(-) diff --git a/src/DB_Row.templates.hh b/src/DB_Row.templates.hh index e886b61..8684c04 100644 --- a/src/DB_Row.templates.hh +++ b/src/DB_Row.templates.hh @@ -58,9 +58,11 @@ DB_Row_Impl_Handler<T>:: Impl::expand_within_capacity(const dimension_type new_size) { PPL_ASSERT(size() <= new_size && new_size <= max_size()); #if !PPL_CXX_SUPPORTS_FLEXIBLE_ARRAYS - // vec_[0] is already constructed. - if (size() == 0 && new_size > 0) + if (size() == 0 && new_size > 0) { + // vec_[0] is already constructed: we just need to assign +infinity. + assign_r(vec_[0], PLUS_INFINITY, ROUND_NOT_NEEDED); bump_size(); + } #endif // Construct in direct order: will destroy in reverse order. for (dimension_type i = size(); i < new_size; ++i) {

1 0