Member fiber_bundle::array_sec_vd_dof_map::get_dof (pod_index_type xdisc_id, pod_index_type xfiber_dof_id, void *xdof, size_type xdof_size) const

new_ prefix added to temporarily prevent ambiguity with old signature get_dof(const scoped_index&, bool, void*, size_t); remove when old signature removed.

Member fiber_bundle::array_section_dof_map::get_dof (pod_index_type xdisc_id, pod_index_type xfiber_dof_id, void *xdof, size_type xdof_size) const

new_ prefix added to temporarily prevent ambiguity with old signature get_dof(const scoped_index&, bool, void*, size_t); remove when old signature removed.

Member fiber_bundle::at0::invariant () const

Guard invariants on p, d, dof_ct until poset member restriction implemented.

Member fiber_bundle::at0_lite::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::at1::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::at1_lite::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::at2::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::at2_lite::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::at3::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::at3_lite::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::base_space_poset::is_block_member (pod_index_type xmbr_id) const

product subposet hack, ver 2. Needed to enforce that a section space schema member with an implicit base space schema member is not supported by the i/o system.

Member fiber_bundle::base_space_poset::is_explicit_member (pod_index_type xmbr_id) const

product subposet hack, ver 2. Needed to enforce that a section space schema member with an implicit base space schema member is not supported by the i/o system.

Member fiber_bundle::binary_section_space_schema_member::get_decomposition (pod_index_type xindex) const

the product subposet hack, version 2. This member exists only to encapsulate the hack.

Member fiber_bundle::binary_section_space_schema_member::get_ext_ids (const std::string &xid_space_name, pod_index_type &xschema_poset_ext_id, pod_index_type &xschema_member_ext_id, pod_index_type &xbase_space_ext_id, pod_index_type &xfiber_schema_ext_id, pod_index_type &xlocal_schema_ext_id) const

the product subposet hack, version 2. This member exists only to encapsulate the hack.

Member fiber_bundle::binary_section_space_schema_member::get_int_id (pod_index_type xext_id, const std::string &xid_space_name) const

the product subposet hack, version 2. This member exists only to encapsulate the hack.

product subposet hack, ver 2. Section space schema member external ids are really base space external ids.

Member fiber_bundle::binary_section_space_schema_poset::get_index_from_components (pod_index_type xbase_space_id, pod_index_type xfiber_schema_id) const

Cartesian product subspace hack. Should be using product id space.

Member fiber_bundle::binary_section_space_schema_poset::new_schema_handle (const std::string &xid_space_name, pod_index_type xschema_member_ext_id, pod_index_type xbase_space_ext_id, pod_index_type xfiber_schema_ext_id, pod_index_type xlocal_schema_ext_id)

the product subposet hack, version 2. / This member exists only to encapsulate the hack. /

Member fiber_bundle::constant_fcn_space::basis_derivs_at_coord (const dof_type xlocal_coords[], size_type xlocal_coords_ub)

For point, db = 0 and there really aren't any derivatives, but we don't gaurd against that exceptional case, so just set db = 1 here; see also constant_point.h

Member fiber_bundle::constant_fcn_space::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::constant_hex::domain_constants

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::constant_point::_basis_deriv_value_buffer [DL]

For point, db = 0 and there really aren't any derivatives, but we don't gaurd against that exceptional case, so just allocate a single value; see also constant_fcn_space::basis_derivs_at_coord.

Member fiber_bundle::constant_point::domain_constants

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::constant_quad::domain_constants

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::constant_segment::domain_constants

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::constant_tet::domain_constants

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::constant_triangle::domain_constants

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::deep_size (const base_space_poset &xp, bool xinclude_shallow=true, size_t xresults[4]=0)

Should have an deep size policy for computing the size of the block without calling deep size on the pointers.

Member fiber_bundle::discretization_context::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::discretization_iterator::initialize_has_visited (const abstract_poset_member &xanchor)

remove depth_first_iterator::has_viist()/put_has_visited(zn_to_bool*) when no longer need here. _descending_has_visited = _has_visited; _ascending_has_visited = new zn_to_bool(*_has_visited);

Member fiber_bundle::discretization_iterator::next (bool xtruncate)

remove depth_first_iterator::has_viist()/put_has_visited(zn_to_bool*) when no longer need here.

remove depth_first_iterator::has_viist()/put_has_visited(zn_to_bool*) when no longer need here. _has_visited = _descending_has_visited;

Member fiber_bundle::discretization_iterator::~discretization_iterator ()

remove depth_first_iterator::has_viist()/put_has_visited(zn_to_bool*) when no longer need here.

Member fiber_bundle::eval_iterator::next (bool xtruncate)

shouldn't have these hard-wired, should look them up when iterator created.

Member fiber_bundle::fiber_bundles_namespace::make_fiber_space_schema_poset ()

assume any type with row dofs is concrete. this is currently true, but there's no reason it has to be.

Member fiber_bundle::fiber_bundles_namespace::make_sec_rep_descriptor_poset ()

"pointer bug"; don't detach or delete the host.

Member fiber_bundle::fiber_bundles_namespace::make_sec_rep_descriptor_schema_poset ()

"pointer bug"; don't detach or delete the host.

Member fiber_bundle::fiber_bundles_namespace::make_section_space_schema_schema_poset ()

"pointer bug"; don't detach or delete lschema.

Member fiber_bundle::field_dof_map::get_dofs (const client_index *xclient_ids, int xclient_id_ct, void *xbuf, size_t xbuf_len) const

Following 2 redefinitions provided only to silence warnings about hiding virtual function.

Member fiber_bundle::gl3::vector_space_type

e3 for the vector space type is too specific; we should layer this like t2, t2_e3, etc.

Member fiber_bundle::gln::d () const

can't call invariant because it's not const

Member fiber_bundle::gln::d (bool xauto_access) const

can't call invariant because it's not const

Member fiber_bundle::hex_connectivity::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::line_connectivity::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::linear_3d::coord_at_value (const dof_type xdofs[], size_type xdofs_ub, const dof_type xglobal_coords[], size_type xglobal_coord_ub, dof_type xlocal_coords[], size_type xlocal_coords_ub) const

commented out to suppress compiler warnings about declared but not referenced.

Member fiber_bundle::point_block_1d::new_state (poset *xhost, const size_type &xi_size, bool xauto_access=true)

a step in the direction of redeclaring host to base_space_poset

Member fiber_bundle::point_block_1d::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::point_block_2d::new_state (poset *xhost, const size_type &xi_size, const size_type &xj_size, bool xauto_access=true)

a step in the direction of redeclaring host to base_space_poset

Member fiber_bundle::point_block_2d::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::point_block_3d::new_state (poset *xhost, const size_type &xi_size, const size_type &xj_size, const size_type &xk_size, bool xauto_access=true)

a step in the direction of redeclaring host to base_space_poset

Member fiber_bundle::point_block_3d::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::point_connectivity::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::product_section_space_schema_crg_range::finalize (poset_state_handle &xhost)

this same call is made in poset_crg_state::new_member_range, where it fails silectly because the vertex range is empty. This protocol is way too arcane and we need to find some better way.

Member fiber_bundle::product_section_space_schema_member::get_decomposition (pod_index_type xindex) const

the product subposet hack, version 2. This member exists only to encapsulate the hack.

Member fiber_bundle::product_section_space_schema_member::get_ext_ids (const std::string &xid_space_name, pod_index_type &xschema_poset_ext_id, pod_index_type &xschema_member_ext_id, pod_index_type &xbase_space_ext_id, pod_index_type &xfiber_schema_ext_id, pod_index_type &xlocal_schema_ext_id) const

the product subposet hack, version 2. This member exists only to encapsulate the hack.

Member fiber_bundle::product_section_space_schema_poset::new_schema_handle (const std::string &xid_space_name, pod_index_type xschema_member_ext_id, pod_index_type xbase_space_ext_id, pod_index_type xfiber_schema_ext_id, pod_index_type xlocal_schema_ext_id)

the product subposet hack, version 2. / This member exists only to encapsulate the hack. /

Member fiber_bundle::product_section_space_schema_poset::new_state (namespace_poset &xhost, const std::string &xname, const arg_list &xargs, const poset_path &xschema_path, bool xauto_access)

the following is necessary because the i/o subsystem can't handle poset with same name as namespace.

unexecutable because may not have access.

Member fiber_bundle::quad_connectivity::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::quadratic_2d::coord_at_value (const dof_type xdofs[], size_type xdofs_ub, const dof_type xglobal_coords[], size_type xglobal_coord_ub, dof_type xlocal_coords[], size_type xlocal_coords_ub) const

commented out to prevent KCC warnings about declared but not referenced.

Member fiber_bundle::quadratic_3d::coord_at_value (const dof_type xdofs[], size_type xdofs_ub, const dof_type xglobal_coords[], size_type xglobal_coord_ub, dof_type xlocal_coords[], size_type xlocal_coords_ub) const

commented out to suppress compiler warnings about declared but not referenced.

Member fiber_bundle::sec_rep_descriptor::evaluators () const

return type any because can't return type eval_family / until we've finished refactoring sheaves and fiber_bundles. / Client should down cast to eval_family*. /

Member fiber_bundle::sec_rep_descriptor::make_standard_schema (namespace_poset &xns)

the following realy should be a bool, but we don't support bool as a primitive.

Member fiber_bundle::sec_rep_space::multisection_id_space (bool xauto_access) const

map should be created in sec rep space constructor.

Member fiber_bundle::sec_rep_space::multisection_id_space (bool xauto_access)

map should be created in sec rep space constructor.

Member fiber_bundle::sec_rep_space_member::db () const

virtual so it can be overridden in sec_ed_invertible as workaround for bug in schema().db(). Remove when bug fixed.

Member fiber_bundle::sec_rep_space_member::same_rep (const sec_rep_space_member *xother) const

this definition used for backwards compatibility / with previous implementation. /

this definition used for backwards compatibility with previous implementation.

Member fiber_bundle::sec_rep_space_member::standard_schema_poset_name (const std::string &xschema_member_name)

the "fiber space hack". Currently each section_space_schema_member / used to create a section space must have its own schema poset because the / schema poset owns the fiber space. See additional comments in / section_space_schema_poset. /

Member fiber_bundle::sec_vd::force_fiber (pod_index_type xdisc_id, const vd_lite &xfiber)

: need virtual size_type any_lite::row_dofs_size().

Member fiber_bundle::sec_vd::force_fiber (const scoped_index &xdisc_id, const vd_lite &xfiber, bool xauto_access)

: need virtual size_type any_lite::row_dofs_size().

Member fiber_bundle::sec_vd::get_fiber (const scoped_index &xdisc_id, vd_lite &xfiber, bool xauto_access) const

: need virtual size_type any_lite::row_dofs_size().

Member fiber_bundle::sec_vd::get_fiber (pod_index_type xdisc_id, vd_lite &xfiber) const

: need virtual size_type any_lite::row_dofs_size().

Member fiber_bundle::sec_vd::new_comp ()

this causes trouble in fields.t: ensure(invariant());

Member fiber_bundle::sec_vd::put_fiber (pod_index_type xdisc_id, const vd_lite &xfiber)

: need virtual size_type any_lite::row_dofs_size().

Member fiber_bundle::sec_vd::put_fiber (const scoped_index &xdisc_id, const vd_lite &xfiber, bool xauto_access)

: need virtual size_type any_lite::row_dofs_size().

Member fiber_bundle::sec_vd::value_at_point_ua (const chart_point &xpt, value_type *xresult, size_type xresult_ub, bool xauto_access=true) const

the preceding needs to be replaced with the following. require(unexecutable(xpt.chart_id() is mappable to a member of schema().evaluation()));

assume discretization members are lower cover.

Member fiber_bundle::sec_vd_algebra::extrema (const sec_vd &x0, sec_vd_value_type xmin[], size_type xmin_ub, sec_vd_value_type xmax[], size_type xmax_ub)

rep doesn't create the evaluator family, so we have to create an instance here.

Member fiber_bundle::section_dof_map::get_dof (pod_index_type xdisc_id, pod_index_type xfiber_dof_id, void *xdof, size_type xdof_size)

: get_dof(pod_index_type, pod_index_type, void*, size_type) const has an implicit 0-th argument const section_dof_map*. When called on a non-const section_dof_map, a conversion is required on the implicit first argument. This triggers the C++ argument matching rules, which conclude there is an ambiguity with the old signature get_dof(const scoped_index&, bool, void*, size_t), even though the explicit arguments in the call appear to be an exact match. The following signature can be removed when the conflicting old signature is removed.

Member fiber_bundle::section_dof_map::get_dof (const scoped_index &xdisc_id, const scoped_index &xfiber_dof_id, void *xdof, size_type xdof_size)

: get_dof(const scoped_index&, const scoped_index&, void*, size_type) const has an implicit 0-th argument const section_dof_map*. When called on a non-const section_dof_map, a conversion is required on the implicit first argument. This triggers the C++ argument matching rules, which conclude there is an ambiguity with the old signature get_dof(const scoped_index&, bool, void*, size_t), even though the explicit arguments in the call appear to be an exact match. This signature can be removed when the conflicting old signature is removed.

Member fiber_bundle::section_dof_map::supports_xfr_opt () const

dof tuple transfer hack. /

dof tuple transfer hack.

Member fiber_bundle::section_iteration_state::gather_dofs (const sec_vd &xsec)

uniform rep hack. treat uniform disc as if same as eval.

Member fiber_bundle::section_iteration_state::gather_dofs (const sec_vd &xsec, const block< discretization_context > &xother_pts)

required because discretization_context::values capacity is fixed at 3.

Member fiber_bundle::section_iteration_state::initialize (const section_space_schema_member &xschema)

uniform rep hack. treat uniform disc as if same as eval.

Member fiber_bundle::section_iteration_state::is_uniform

uniform rep hack; the following member is needed only to support the hack.

Class fiber_bundle::section_space_schema_member

temporarily redefine all new state functions as unimplemented to trap calls to them.

Member fiber_bundle::section_space_schema_member::attach_handle_data_members ()

section_space_schema version hack: version of this is same as version of base space.

Member fiber_bundle::section_space_schema_member::contains_dof (const schema_poset_member &xother, bool xin_table_dofs) const

using declaration broken, must explicitly propagate the next six declarations.

Member fiber_bundle::section_space_schema_member::fiber_schema_conforms_to (const namespace_poset &xns, const poset_path &xpath, const poset_path &xfiber_schema_path, bool xauto_access=true)

only binary schema supported.

Member fiber_bundle::section_space_schema_member::get_ext_id (const std::string &xid_space_name) const

the product subposet hack, version 2. / This member exists only to encapsulate the hack. /

product subposet hack, ver 2. Section space schema member external ids are really base space external ids.

the product subposet hack, version 2. / This member exists only to encapsulate the hack. /

product subposet hack, ver 2. Section space schema member external ids are really base space external ids.

the product subposet hack, version 2. / This member exists only to encapsulate the hack. /

Member fiber_bundle::section_space_schema_member::put_version (int xversion, bool xunalias=false)

section_space_schema version hack: version of this is same as version of base space.

Member fiber_bundle::section_space_schema_member::update_table_cache () const

Cartesian product subspace hack. Should be using product id space.

Member fiber_bundle::section_space_schema_poset::_fiber_space

the "fiber space hack", the schema poset should not own the fiber space. The fiber space is instantiated on a specific fiber schema member and hence should belong either to that member or to a section space instantiated on that member. But we have no way currently to implement either of these associations.

Member fiber_bundle::section_space_schema_poset::df () const

The fibers don't really have a dimension at this level of abstraction, dimesion is a vector space notion. The closest we can come is the number of columns in the representation, that is, the row dof ct.

Member fiber_bundle::section_space_schema_poset::get_decomposition (pod_index_type xindex) const

the product subposet hack, version 2. This member exists only to encapsulate the hack. Row decomposition i/o is not supported for section space schema in PSH2; this routine is not implemented and will issue a fatal error if called.

the product subposet hack, version 2. This member exists only to encapsulate the hack.

the product subposet hack, version 2. This member exists only to encapsulate the hack. Row decomposition i/o is not supported for section space schema in PSH2; this routine is not implemented and will issue a fatal error if called.

the product subposet hack, version 2. This member exists only to encapsulate the hack.

the product subposet hack, version 2. This member exists only to encapsulate the hack. Row decomposition i/o is not supported for section space schema in PSH2; this routine is not implemented and will issue a fatal error if called.

Member fiber_bundle::section_space_schema_poset::get_fiber_schema_id_from_index (pod_index_type xmbr_hub_id) const

Cartesian product subspace hack. Should be using product id space.

Member fiber_bundle::section_space_schema_poset::initialize_standard_members ()

Cartesian product subspace hack. Should be using product id space. There are no entries in the hub id space for the members of the Cartesian subspace of the general tensor product.Add an interval term to the hub id space which creates a general index space. Really should be a product index space and the factoring of the indices should use the tuple and ordinal functions of the index space. But since this class hierarchy is currently set up to do it's own factors with get_base_space_id_from_index, etc, we'll continue to do it that way, temporarily. The id space allocated here is in the interval [0, base_space().member_index_ub().pod()*base_space_stride()), which includes the ids reserved for bootom and top. This means bottom is the same as (base bottom, fiber_schema bottom), which is correct, and top is the same as (base bottom, fiber schema top), which is not correct. We can get away with it because we never refer to the (base bottom, fiber schema top).

Member fiber_bundle::section_space_schema_poset::invariant () const

Cartesian product subspace hack. Should be using product id space.

Member fiber_bundle::section_space_schema_poset::is_new_member_available () const

This function is the precondition for the new_member functions and only exists because the homogeneous range needs to be dynamic. The number of members created can not exceed the size of the reserved id space.

Member fiber_bundle::section_space_schema_poset::new_member_range (const std::string &xrange_type, size_type xsize, const block< scoped_index > &xdof_tuple_ids, const block< pod_index_type > &xdata)

This function can not be available to the client.

Member fiber_bundle::section_space_schema_poset::new_standard_member_hack (standard_member_index xindex)

Cartesian product subspace hack; intended only for use by section_space_schema_poset constructor.

Member fiber_bundle::section_space_schema_poset::version () const

section_space_schema version hack: version of this is same as version of base space.

Member fiber_bundle::section_space_schema_table_dof_crg_range::finalize (poset_state_handle &xhost)

this same call is made in poset_crg_state::new_member_range, where it fails silectly because the vertex range is empty. This protocol is way too arcane and we need to find some better way.

Member fiber_bundle::sparse_field_dof_map::get_dof (const scoped_index &xid, bool xis_poset_id, void *xbuf, size_t xbuf_len) const

We're assuming multiplicity == 1 below.

assume multiplicity = 1.

Member fiber_bundle::sparse_field_dof_map::put_dof (const scoped_index &xid, bool xis_poset_id, const void *xbuf, size_t xbuf_len)

We're assuming multiplicity == 1 below.

assume multiplicity = 1.

Member fiber_bundle::sparse_field_dof_map::put_dof_tuple (const void *xbuf, size_t xbuflen)

cast away constness because put_dof doesn't support it.

Member fiber_bundle::sparse_field_dof_map::supports_xfr_opt () const

dof tuple transfer hack. /

dof tuple transfer hack.

Member fiber_bundle::sparse_section_dof_map::get_dof (pod_index_type xdisc_id, pod_index_type xfiber_dof_id, void *xdof, size_type xdof_size) const

new_ prefix added to temporarily prevent ambiguity with old signature get_dof(const scoped_index&, bool, void*, size_t); remove when old signature removed.

Member fiber_bundle::sparse_section_dof_map::put_dof_tuple (const void *xbuf, size_t xbuflen)

cast away constness because put_dof doesn't support it.

Member fiber_bundle::sparse_section_dof_map::supports_xfr_opt () const

dof tuple transfer hack. /

dof tuple transfer hack.

Member fiber_bundle::st2::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::st2_lite::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::st3::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::st3_lite::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::st4::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::st4_lite::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::structured_block_1d::new_state (poset *xhost, const size_type &xi_size, bool xauto_access=true)

a step in the direction of redeclaring host to base_space_poset

Member fiber_bundle::structured_block_1d::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::structured_block_2d::new_state (poset *xhost, const size_type &xi_size, const size_type &xj_size, bool xauto_access=true)

a step in the direction of redeclaring host to base_space_poset

Member fiber_bundle::structured_block_2d::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::structured_block_3d::new_state (poset *xhost, const size_type &xi_size, const size_type &xj_size, const size_type &xk_size, bool xauto_access=true)

a step in the direction of redeclaring host to base_space_poset

Member fiber_bundle::structured_block_3d::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::t2::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::t2_lite::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::t3::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::t3_lite::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::t4::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::t4_lite::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::tetra_connectivity::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::tp::dd (bool xauto_access) const

can't call invariant because it's not const

Member fiber_bundle::tp::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::tp_lite::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::triangle_connectivity::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::tuple::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::uniform_1d::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::unstructured_block::new_state (poset *xhost, const base_space_member &xlocal_cell_prototype, const scoped_index *ids, int xub, bool xauto_access)

assume if block has maximal dimension, we don't need to compute the upper cover.

Member fiber_bundle::unstructured_block::new_state (poset *xhost, const base_space_member &xlocal_cell_prototype, const pod_index_type *ids, int xub, bool xauto_access)

assume if block has maximal dimension, we don't need to compute the upper cover.

Member fiber_bundle::unstructured_block::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fiber_bundle::unstructured_block_builder::build_block_pa (const base_space_member *xtemplate, const int *xglue, size_type xglue_ub, unstructured_block *result, bool xcompute_upper_cover, bool xauto_access)

make sure the base space contains the blocks and d_cells subposets. The following can be removed when base_space_member::host() is base_space_poset.

instead of resolving the above issue, just insist that we have a jim and atom and we are not in jim_edit_mode. This forces end_jim_edit_mode for each block, but also resolves the conflict with making the boundary.

Member fiber_bundle::vd::d () const

can't call invariant because it's not const

Member fiber_bundle::vd::d (bool xauto_access) const

can't call invariant because it's not const

Member fiber_bundle::vd_algebra::transform_basis_by (st2_e3_lite &xv, const gl3_lite &xtransform, bool is_contravariant)

Need to move this to vd.h.

Member fiber_bundle::zone_nodes_block::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member fields::body_pusher::body_pusher (const geometry::sec_ed_invertible &xdomain_coord, const sec_ed &xrange_coord, bool xauto_access)

iterator looks for evaluation subposet by name and we don't currently support multiple names for a subposet. So make sure name is "__vertices".

Member fields::edge_centered_polygon_refiner::refine_coordinates (field_refinement_buffer &xbuffer)

Need abstract base class for polygon section evaluators.

Class fields::edge_centered_refiner_family

basically barycentric_refiner_family with / barycentric_triangle_refiner replaced with edge_centered_triangle_refiner. /

Member fields::field_eval_iterator::repeat (bool xwas_eval_member)

this really should be done by modifying the state transition function in DEPTH_FIRST_ITERATOR.

Member fields::field_eval_iterator::type_name () const

this signature and implementation forces an unnecessary copy (or three).

Member fields::field_vd::_property

_coordinates and _property are pointers rather than values in order to make the accessor functions above const without fixing the constness of the entire sheaves and fiber bundles components.

Member fields::field_vd::db () const

trap for possible error in db due to base space rep incompatibility.

Member fields::field_vd::embed_property (const std::string &xresult_name, name_multimap &xresult_member_names, bool xauto_access)

This function belongs in sec_at0, but is put here temporarily to avoid problems merging with branch refactor-8.

Member fields::field_vd::field_vd (const sec_ed &xcoordinates, const sec_vd &xproperty, bool xauto_access)

schema().base_space() isn't a base_space_member, so we can't clone it.

Member fields::field_vd::field_vd (namespace_poset &xns, const poset_path &xcoordinates_path, const poset_path &xproperty_path, bool xauto_access)

schema().base_space() isn't a base_space_member, so we can't clone it.

Member fields::property_disc_iterator::get_prop_disc_values (const sec_vd &xsec)=0

required because discretization_context::values capacity is fixed at 3.

Member fields::property_disc_iterator::type_name () const

this signature and implementation forces an unnecessary copy (or three).

Member fields::property_disc_iterator_1_1::get_prop_disc_values (const sec_vd &xsec)

required because discretization_context::values capacity is fixed at 3.

Member fields::property_disc_iterator_1_2::get_prop_disc_values (const sec_vd &xsec)

required because discretization_context::values capacity is fixed at 3.

Member fields::property_disc_iterator_1_3::get_prop_disc_values (const sec_vd &xsec)

required because discretization_context::values capacity is fixed at 3.

Member fields::property_disc_iterator_4_2::get_prop_disc_values (const sec_vd &xsec)

required because discretization_context::values capacity is fixed at 3.

Member fields::property_disc_iterator_4_3::get_prop_disc_values (const sec_vd &xsec)

required because discretization_context::values capacity is fixed at 3.

Member fields::refining_section_pusher::push (const field_vd &xsource, field_vd &xtarget, bool xauto_access)

have to create separate handle because base_space_member::host() is type poset.

Member fields::section_pusher::invariant () const

unexecutable because we may not have required access.

Member fields::section_pusher::push (const sec_vd &xinput, bool xauto_access)

the "pointer bug", see bleow.

the "pointer bug". We need to make a new state for the result, but we must ensure that the host the state is made with is persistent, to avoid the "pointer bug", otherwise we will get a segv if the host gets deleted. _range is not persisent, so get the corresponding pointer from the name space.

Member fields::section_pusher::section_pusher (const sec_rep_space &xdomain, const sec_rep_space &xrange, const sec_ed &xdomain_coord, const sec_ed &xrange_coord, bool xauto_access)

Had to comment out this constructor due to COM-466.

Member geometry::cylindrical_point_locator::cylindrical_point_locator (sec_ed &xcoords)

a new eval_family is created every time put_schema_anchor(schema) is called in eval_itr, including initially from the eval_itr constructor. But we have to reanchor the eval_itr to each branch of a multisection in update(). We can do this only with put_schema_anchor(index), which does not recreate the eval family, if we initialize eval_itr here using its constructor, then force it to be done, ready for reanchoring.

Member geometry::cylindrical_point_locator::update ()

the multisection hack.

Member geometry::d_bin_point_locator< DC, DB >::d_bin_point_locator (sec_ed &xcoords)

a new eval_family is created every time put_schema_anchor(schema) is called in eval_itr, including initially from the eval_itr constructor. But we have to reanchor the eval_itr to each branch of a multisection in update(). We can do this only with put_schema_anchor(index), which does not recreate the eval family, if we initialize eval_itr here using its constructor, then force it to be done, ready for reanchoring.

Member geometry::d_bin_point_locator< DC, DB >::update (bool xpopulate, size_type xeval_capacity)

the multisection hack.

Member geometry::db0_point_locator< DC >::update ()

the multisection hack.

Class geometry::line_surface_intersecter

Currently implemented only for lines aligned with the z axis and for triangle meshes.

Member geometry::line_surface_intersecter::_evaluator

Only z-axis aligned lines and triangles

Member geometry::line_surface_intersecter::intersect (const e3_lite &xp0, const e3_lite &xp1, intersection_set_type &xresult) const

Only z-axis aligned lines.

Member geometry::point_locator::update_domain ()

the multisection hack. Iterate over the branches.

Member sheaf::abstract_poset_member::delete_down (bool xdelete_exterior=false, bool xenter_jim_edit_mode=true)

just link it to the bottom.

Member sheaf::abstract_poset_member::invariant () const

assertion should be true with or without read access, but dof_map requires read access and this doesn't have it in bootstrap.

Member sheaf::abstract_poset_member::is_valid_index (const poset_state_handle *xhost, pod_index_type xhub_id, int xversion=CURRENT_HOST_VERSION) const

swig doesn't like using poset_component::is_valid_index

Member sheaf::abstract_poset_member::l_join_pa (abstract_poset_member *other, abstract_poset_member *result, bool xnew_jem=true)

join operator should support tern argument.

Member sheaf::abstract_poset_member::maximal_jim_ct () const

implementation of abstract_poset_member::maximal_jim_ct is inefficient.

Member sheaf::abstract_poset_member::name () const

swig doesn't like using poset_component::name

Member sheaf::abstract_poset_member::table_dof_ptr (bool xrequire_write_access=false) const

can't call invariant because not const

Member sheaf::attributes_record::transfer_external_buffer_to_HDF (hdf_buf_type *xbuf)

the hdf type H5T_C_S1 with len set to H5T_VARIABLE hides the hdf variable length data type hvl_t, so this class doesn't quite fit the structure provided by base class variable_length_record. Make it work anyway. See also attributes_record_set::create_dataset.

Member sheaf::attributes_record::transfer_HDF_to_external_buffer (hdf_buf_type *xbuf)

the hdf type H5T_C_S1 with len set to H5T_VARIABLE hides the hdf variable length data type hvl_t, so this class doesn't quite fit the structure provided by base class variable_length_record. Make it work anyway. See also attributes_record_set::create_dataset.

Member sheaf::attributes_record_set::create_int_data_type ()

the hdf type H5T_C_S1 with len set to H5T_VARIABLE hides the hdf variable length data type hvl_t, so this class doesn't quite fit the structure provided by base class variable_length_record. Make it work anyway.

Member sheaf::deep_size (const ragged_array_index_space_interval &xn, bool xinclude_shallow=true)

The hub ids should only be added if the interval owns the data. Since this class is only used by zone_nodes_block_crg_interval we add the contributions of the hub id array here so we do not have to specialize deep_size functions for crg_interval.

Member sheaf::deep_size (const explicit_index_space_interval &xn, bool xinclude_shallow=true)

Since block calls the namespace level function, we must call deep size on all the states in the block directly.

Member sheaf::deep_size (const array_index_space_interval &xn, bool xinclude_shallow=true)

The hub ids should only be added if the interval owns the data. Since this class is only used by zone_nodes_block_crg_interval we add the contributions of the hub id array here so we do not have to specialize deep_size functions for crg_interval.

Member sheaf::depth_first_iterator::has_visited () const

temporary support for obsolete "multivalued" / feature of discretization_iterator; will be removed when / discretization_iterator updated. /

Member sheaf::depth_first_iterator::put_has_visited (zn_to_bool *xhas_visited)

temporary support for obsolete "multivalued" / feature of discretization_iterator; will be removed when / discretization_iterator updated. /

Class sheaf::depth_first_itr< T >

Included set and unordered_set as a workaround until we know how to get forward declarations for same.

Member sheaf::dof_tuple_col_bounds_record::transfer_poset_to_internal_buffer ()

product subposet hack, version 2. Poset has to tell us which map to use to translate subposet id.

Class sheaf::dof_tuple_record_set

the transparency hack. / The io system was originally designed / and implemented to provide record selection, restriction, and data / type conversion with as little dependence on HDF as possible. / A side effect of this was that HDF did not / know what the data types in the dof tuple dataset were and the results / of h5dump were unreadable. We subsequently hacked the io system to / tell HDF what the types were, thus achieving both dump transparency and / portability. This left the most of the record and conversion machinery / in place, even though it's no longer needed. The io subsystem was / hard to understand and maintain before this hack, now it needs to be / completely rewritten. /

Member sheaf::dof_tuple_record_set::_hdf_int_selection )[1]

Must use literal 1 in the next 2 declarations instead of static const data member to avoid bug in MS C++ which interprets initialization as definition and leads to violation of one-definition-rule at link time. Must be same value as DATASPACE_RANK in .cc

Member sheaf::dof_tuple_record_set::compute_ext_id (const scoped_index &xtuple_id)

Product subposet hack, version 2. Have to have a schema member to get bound iterator; use the external schema, it isn't modfied by previous bounds iterations and should always have the right fiber component for section space schema.

product subposet hack, ver 2. Not all posets have external ids; let schema give us a pseudo id.

Member sheaf::dof_tuple_record_set::create_dataset ()

need sequence ids, not client ids. The intent is for the external data space to be compact, that is, sequential starting at 0. We are currently using client ids as external dof ids and client ids are not in general compact. So also the comment in externalize.

if dof_ct == 0, we should just not write the dataset. But it turns out to be quite complicated to do and it doesn't occur often enough to be worth the trouble, so just force an extent of 1.

Member sheaf::dof_tuple_record_set::externalize ()

Product subposet hack, version 2. Have to have a schema member to get bound iterator; use the external schema, it isn't modfied by previous bounds iterations and should always have the right fiber component for section space schema.

product subposet hack, ver 2. Not all posets support external ids directly; let the schema give us a suitable key.

Member sheaf::dof_tuple_record_set::externalize_domain (const scoped_index &xtuple_ext_id, const scoped_index &xschema_id)

product subposet hack, ver 2. Not all posets support external ids directly; let the schema give us a suitable key.

Member sheaf::dof_tuple_record_set::externalize_one_domain (const scoped_index &xtuple_id, const scoped_index &xschema_id)

the product subposet hack, ver 2. Not all posets have external ids; let the schema give us a pseudo-id.

Member sheaf::dof_tuple_record_set::populate_internal_dof_map (const poset_dof_map &xext_dof_map, poset_dof_map &xint_dof_map)

sparse_field_dof_map doesn't currently support dof_tuple function.

Member sheaf::dof_tuple_record_set::read_records ()

Product subposet hack, version 2. We need to get the col bounds iterator from a schema member, not just a poset, because for section space schema we need to know the fiber id.

the product subposet hack, ver 2. Make the schema get us a suitable key for the offset map.

Member sheaf::dof_tuple_type_from_name (const std::string &xname)

The following if block is crude, but there doesn't seem to be any substantially better way to do it. The language doesn't support iteration over enums and integral values, except compile-time constants, can not be converted to enums, so searching an array of names is as at least as clumsy as this construction. STL maps would be clumsy to initialize.

Member sheaf::error_message::post (bool xforce_exit=false) const

just print the message and quit if the error is severe enough.

Member sheaf::forwarding_index_space_handle::id_spaces ()

The host should be const but a mutable interface for the family is required by put_name(). Do a const cast for now and review in the future.

Member sheaf::id_space_names_record::transfer_internal_buffer_to_poset ()

Require that the id space is mutable. See COM-475.

Member sheaf::id_space_names_record::transfer_poset_to_internal_buffer ()

Require that the id space is mutable. See COM-475.

Member sheaf::index_equivalence_iterator::invariant () const

section_space_schema_poset internal_id hack. !rep_id().is_scoped should go away once the section_space_schema_poset has a scoped internal_id.

Member sheaf::index_iterator::index_iterator (const zn_to_bool *xhost, const index_space_handle &xid_space, bool xdelete_host=false)

host should be a reference counted pointer; see destructor.

Member sheaf::index_iterator::reset (const zn_to_bool *xhost, const index_space_handle &xid_space, bool xdelete_host=false)

host should be a reference counted pointer; see destructor.

Member sheaf::index_iterator::~index_iterator ()

host should be a reference counted pointer.

Member sheaf::index_space_family::extend_last_term (size_type xct)

A temporary method to support extension of section space schema when the underlying base space is extended.

Member sheaf::index_space_family::new_state (const std::string &xname, const std::string &xstate_class_name)

See COM-475. Remove this precondition once COM-475 is fixed.

See COM-475. Remove this postcondition once COM-475 is fixed.

Member sheaf::index_space_iterator::invariant () const

The following invariance should be true except when iterating over reserved terms.

Member sheaf::member_record_set::read_record (const record_index &xext_id)

crudely adapted from read_records to read a single record at a time.

Member sheaf::member_record_set::~member_record_set ()

storage_agent::begin_read_transation() releases access to the poset. When the member_record_set local variable goes out of scope it no longer has access. Read access is only needed because writing a poset forces delete_member_id_space to only require read access when it should require read-write access.

Member sheaf::namespace_poset::attach_handle_data_members ()

Invariant requires all three entities below to be attached, but invariant gets called within scope of attach_to_state, so have to disable the invariant.

Member sheaf::namespace_poset::insert_poset (const poset_state_handle &xposet, const std::string &xposet_name, bool xauto_link)

previously this routine entered and exited jim_edit_mode without exporting any precondition. As a result, numerous new_state routines do not properly provide access before calling this routine. Continue the old practice until we get a chance to clean the mess up.

Member sheaf::namespace_poset_dof_map::extend_to_top ()

this function exists only to patch up / a problem in member_record::new_row_dof_tuple. /

Member sheaf::namespace_poset_schema::initialize_standard_members ()

This isn't strictly necessary, because the dof id map is supposed to map whatever order the lower cover is created in to the fixed order of the top id space, which is consistent with namespace_poset_dof_map, see abstract_poset_member::make_schema_info. But update_cache erroneously generates a different order for the dof descriptors. So until we've fixed that, the order of the lower cover is critical.

Member sheaf::operator<< (std::ostream &os, const abstract_poset_member *p)

when we've found the calls to this routine, remove it.

Member sheaf::poset::new_state (const poset_path &xpath, const schema_poset_member &xschema, array_poset_dof_map &xdof_map)

version subposet membership management. We can resolve the above version subposet membership management issue by setting the current version to CCR. This makes the refineable posets behave like non-version posets, at the cost of inserting every member into the CCR subposet twice, once for the CCR and once for the current version.

Member sheaf::poset_bounds::poset_bounds ()

the interface to this class has become a kluge due to two influences: (i) the product subposet hack, and (ii) an unsuccessful attempt to deal efficiently with singleton subposets.

Member sheaf::poset_bounds::ub_is_decomposition () const

avoid expense of computing join by assuming if bound is subposet then join is external schema.

Member sheaf::poset_crg_state::append_upper_cover_of_bottom (pod_index_type xmbr_begin, pod_index_type xmbr_end)

This function is only a temporary fix to avoid a problem size id space for the upper cover of bottom and will be replaced by a templatized get_cover_id_space in the future.

Member sheaf::poset_crg_state::initialize_member_covers (pod_index_type xid, bool xstandard_member_hack=false)

Cartesian product subspace hack; xstandard_member_hack argument is intended only for use by section_space_schema_poset constructor If xstandard_member_hack is true, only support covers for top and bottom.

Cartesian product subspace hack; xstandard_member_hack argument is intended only for use by section_space_schema_poset constructor.

Member sheaf::poset_crg_state::new_standard_member_hack (standard_member_index xid)

Cartesian product subspace hack; intended only for use by section_space_schema_poset constructor.

Member sheaf::poset_dof_map::_ref_ct

need to fully implement reference counting of maps /

Member sheaf::poset_dof_map::invariant () const

unexecutable because schema().version(true) requires read access.

Member sheaf::poset_dof_map::supports_xfr_opt () const

dof tuple transfer hack. /

dof tuple transfer hack.

Member sheaf::poset_general_record::transfer_poset_to_internal_buffer ()

the product subposet hack, ver 2.

Member sheaf::poset_scaffold::dof_tuple_schema_int_id (pod_index_type xdof_tuple_ext_id) const

the product suposet hack, ver 2. Not all posets have external ids, let the poset translate the pseudo-id.

Member sheaf::poset_scaffold::translate_dof_tuple_col_bounds ()

product subposet hack, version 2. Poset has to tell us which map to use to translate subposet id.

Member sheaf::poset_state_handle::attach_handle_data_members ()

Invariant requires all three entities below to be attached, but invariant gets called within scope of attach_to_state, so have to disable the invariant.

Member sheaf::poset_state_handle::bound_contains_member (const poset_bounds &xbounds, bool xis_ub, const scoped_index &xindex) const

the product subposet hack, version 2. This member exists only to encapsulate the hack.

Member sheaf::poset_state_handle::bound_iterator (const poset_bounds &xbnd_id, bool xis_ub) const

the product subposet hack, version 2. This member exists only to encapsulate the hack.

Member sheaf::poset_state_handle::end_jim_edit_mode (bool xensure_lattice_invariant=true, bool xauto_access=true)

ensure_lattice_invariant should be converted to a combination of client callable routines and pre/postconditions. Then xensure_lattice_invariant should be removed.

Member sheaf::poset_state_handle::extend_last_member_term (size_type xct, bool xauto_access)

A temporary method to support extension of section space schema when the underlying base space is extended.

Member sheaf::poset_state_handle::get_decomposition (pod_index_type xindex) const

the product subposet hack, version 2. / This member exists only to encapsulate the hack. /

the product subposet hack, version 2. This member exists only to encapsulate the hack.

Member sheaf::poset_state_handle::initialize_namespace (namespace_poset &xns, const std::string &xposet_name, bool xauto_link=true)

previously insert_poset called from this routine entered and exited jim_edit_mode without exporting any precondition. As a result, numerous new_state routines do not properly provide access before calling this routine. Continue the old practice until we get a chance to clean the mess up.

Member sheaf::poset_state_handle::le (pod_index_type xgreater, pod_index_type xlesser) const

should use upset traverser in poset_state_handle::le

Member sheaf::poset_state_handle::new_link (pod_index_type xgreater, pod_index_type xlesser)

the following precondition is unexecutable because contains_cover_link must do a linear search of the lower cover of xgreater. The lower cover of a block is O(size of block), which makes assertions non-linear.

Member sheaf::poset_state_handle::new_link (const scoped_index &xgreater, const scoped_index &xlesser)

the following precondition is unexecutable because cointains_cover_link must do a linear search of the lower cover of xgreater. The lower cover of a block is O(size of block), which makes assertions non-linear.

Member sheaf::poset_state_handle::new_schema_handle (const std::string &xid_space_name, pod_index_type xschema_member_ext_id, pod_index_type xbase_space_ext_id, pod_index_type xfiber_schema_ext_id, pod_index_type xlocal_schema_ext_id)

the product subposet hack, version 2. / This member exists only to encapsulate the hack. /

this function should be const, but descendants call mmber functions that are not currently const.

Member sheaf::poset_state_handle::version_ct () const

this implementation avoids the need to modify poset_state or introduce table dofs, but it is slow.

Member sheaf::poset_table_state::put_row_dof_tuple (poset_dof_map *xdof_tuple)

should set xdof_tuple->host() as well, but can't in this context.

Member sheaf::primitives_poset_dof_map::init_row_dof_map (const poset_state_handle *xhost, pod_index_type xschema_mbr_id, int xschema_version)

schema id and version ignored for this class.

Member sheaf::record_map< internal_index_type, external_index_type >::operator== (const record_map &xother) const

On AIX, xlC uses unordered_map rather than unordered_map. unordered_map contains no operator==. Rather than define operator== for unordered_map, we comment out the following two lines which test unordered_maps for equality. These are only used in the post condition of record_map's copy contructor, which has now been rendered "unexecutable."

Member sheaf::refinable_poset::new_member (bool xis_jim, pod_index_type xdof_tuple_id)

Using declaration confuses swig into trying to wrap protected new_member. Can't seem to get swig to ignore protected member, so don't use using declaration, override public new_member as well, but just forward to inherited version.

Member sheaf::refinable_poset::new_member_interval (const std::string &xinterval_type, size_type xsize, const block< pod_index_type > &xdof_tuple_ids, const block< pod_index_type > &xdata)

Using declaration confuses swig into trying to wrap protected new_member_interval. Can't seem to get swig to ignore protected member, so don't use using declaration, override public new_member_interval as well, but just forward to inherited version.

Member sheaf::schema_poset_member::bound_iterator (const poset_bounds &xbnd_id, bool xis_ub) const

the product subposet hack, version 2. This member exists only to encapsulate the hack.

Member sheaf::schema_poset_member::get_decomposition (pod_index_type xindex) const

the product subposet hack, version 2. This member exists only to encapsulate the hack.

Member sheaf::schema_poset_member::get_ext_id (const std::string &xid_space_name) const

the product subposet hack, version 2. This member exists only to encapsulate the hack.

Member sheaf::schema_poset_member::get_ext_ids (const std::string &xid_space_name, pod_index_type &xschema_poset_ext_id, pod_index_type &xschema_member_ext_id, pod_index_type &xbase_space_ext_id, pod_index_type &xfiber_schema_ext_id, pod_index_type &xlocal_schema_ext_id) const

the product subposet hack, version 2. This member exists only to encapsulate the hack.

Member sheaf::schema_poset_member::get_int_id (pod_index_type xext_id, const std::string &xid_space_name) const

the product subposet hack, version 2. This member exists only to encapsulate the hack.

Member sheaf::schema_poset_member::new_jim_state (const namespace_poset &xns, const std::string &xname, const poset_path &xparent_path, const wsv_block< schema_descriptor > &xdof_specs, bool xauto_access)

have to create dof map with some primitive type.

Class sheaf::singly_linked_list< T, A >

This wrapper is a temporary fix until gcc-4.2.2 is no longer supported.

Class sheaf::standard_member_hack_crg_interval

Cartesian product subspace hack;

Member sheaf::storage_agent::begin_read_transaction (namespace_poset &xns)

all the non-standard members are restricted to schema member io; see member_record::new_row_dof_tuple. Patch this up by extending the schema to top.

Member sheaf::storage_agent::state_is_consistent (const poset_state_handle &xposet) const

use of member index space as test for external state is a temporary implementation. Need more formal support for this test. In order for this hack to work, we clear the index spaces in write(namespace_poset&). Furthermore, since this routine is typically called before any attempt to write a poset, the call to member_id_space below actually creates the index space.

Member sheaf::storage_agent::write (poset_state_handle &xposet, const poset_bounds_descriptor &xrow_bounds=poset_bounds_descriptor::BOTTOM_TOP, const poset_bounds_descriptor &xcol_bounds=poset_bounds_descriptor::BOTTOM_TOP, bool xretain_read_access=false)

the product subposet hack, version 2. See poset_scaffold::dof_tuple_schema_int_id for the origin of the following requirement.

Member sheaf::storage_agent::write_col_decomposition (poset_state_handle &xposet, const scoped_index &xrow_id, const scoped_index &xdecomp_id)

the product subposet hack. Following precondition can not be enforced because xdecomp_id may be subposet of base space, not schema itself

Member sheaf::storage_agent::write_dof_tuple (const abstract_poset_member &xmbr, const scoped_index &xcol_ub_id, const scoped_index &xdecomp_id)

the product subposet hack. Following precondition can not be enforced because xdecomp_id may be subposet of base space, not schema itself

Member sheaf::subposet::is_valid_index (const poset_state_handle *xhost, pod_index_type xhub_id, int xversion=CURRENT_HOST_VERSION) const

swig doesn't like using poset_component::is_valid_index

Member sheaf::subposet::l_join (bool xnew_jem=true)

the join of the empty subset is well defined (== bottom) but this implmentation doesn't support it.

Member sheaf::subposet::name () const

swig doesn't like using poset_component::name

Member sheaf::subposet::new_id_space (const std::string &xstate_class_name)

Using the name of the id space state is a temporary fix until subposets become id spaces. See COM-183.

Member sheaf::tern::tern ()

Something in the AIX xlC environment is defining TRUE and FALSE. Until we find a workaround, undefine them.

Member sheaf::zn_to_bool::static_const_int

Inline initialization of static const integral data members causes link error in MSC++; use enum instead.

Member sheaf::zn_to_bool::zn_to_bool (int xub, unsigned *values)

an implicit assumption that values points to xub unsigned ints

Member tool::visualization_iterator::force_is_done ()

we're using coord_state.down_set for unmarking. Better not clear it here.

Member tool::visualization_iterator::initialize_states (const section_space_schema_member &xcoordinates_schema, const section_space_schema_member &xproperty_schema)

Had to use the initialization method that takes a schema_poset_member instead of a base_space_poset to avoid the assumption that the evaluation subposet is the elements subposet. This needs explored before merging with the head.

Member tool::visualization_iterator::initialize_states (const section_space_schema_member &xcoordinates_schema)

Had to use the initialization method that takes a schema_poset_member instead of a base_space_poset to avoid the assumption that the evaluation subposet is the elements subposet. This needs explored before merging with the head.

Member tool::visualization_iterator::set_vtk_data (const sec_vd &xcoords, const sec_vd &xprop, vtkDoubleArray *xvtk_pt_coords, vtkDoubleArray *xvtk_prop, int xprop_tensor_rank, block< int > &xvtk_cell_types, block< vtkIdType > &xvtk_cell_connectivity, block< int > &xvtk_cell_type_map, vtkIdTypeArray *xvtk_pt_label_ids, vtkIdTypeArray *xvtk_cell_label_ids)

required because discretization_context::values capacity is fixed at 3.

Member tool::visualization_iterator::set_vtk_data (const sec_vd &xcoords, vtkDoubleArray *xvtk_pt_coords, block< int > &xvtk_cell_types, block< vtkIdType > &xvtk_cell_connectivity, block< int > &xvtk_cell_type_map, vtkIdTypeArray *xvtk_pt_label_ids, vtkIdTypeArray *xvtk_cell_label_ids)

required because discretization_context::values capacity is fixed at 3.

Member tool::visualization_iterator_1_1::set_vtk_data (const sec_vd &xcoords, vtkDoubleArray *xvtk_pt_coords, block< int > &xvtk_cell_types, block< vtkIdType > &xvtk_cell_connectivity, block< int > &xvtk_cell_type_map, vtkIdTypeArray *xvtk_pt_label_ids, vtkIdTypeArray *xvtk_cell_label_ids)

required because discretization_context::values capacity is fixed at 3.

Member tool::visualization_iterator_1_1::set_vtk_data (const sec_vd &xcoords, const sec_vd &xprop, vtkDoubleArray *xvtk_pt_coords, vtkDoubleArray *xvtk_prop, int xprop_tensor_rank, block< int > &xvtk_cell_types, block< vtkIdType > &xvtk_cell_connectivity, block< int > &xvtk_cell_type_map, vtkIdTypeArray *xvtk_pt_label_ids, vtkIdTypeArray *xvtk_cell_label_ids)

required because discretization_context::values capacity is fixed at 3.

Member tool::visualization_iterator_1_3::set_vtk_data (const sec_vd &xcoords, vtkDoubleArray *xvtk_pt_coords, block< int > &xvtk_cell_types, block< vtkIdType > &xvtk_cell_connectivity, block< int > &xvtk_cell_type_map, vtkIdTypeArray *xvtk_pt_label_ids, vtkIdTypeArray *xvtk_cell_label_ids)

required because discretization_context::values capacity is fixed at 3.

Member tool::visualization_iterator_1_3::set_vtk_data (const sec_vd &xcoords, const sec_vd &xprop, vtkDoubleArray *xvtk_pt_coords, vtkDoubleArray *xvtk_prop, int xprop_tensor_rank, block< int > &xvtk_cell_types, block< vtkIdType > &xvtk_cell_connectivity, block< int > &xvtk_cell_type_map, vtkIdTypeArray *xvtk_pt_label_ids, vtkIdTypeArray *xvtk_cell_label_ids)

required because discretization_context::values capacity is fixed at 3.

Member tool::visualization_iterator_4_2::next (bool xtruncate)

_coord_state down_set is convenient place to store all members at or above the elements that have been visited for efficient unmarking in visualization_iterator::reset. Note that since _bove.vis_eval has not yet been evaluated for this iteration, it is still true if we are at an element.

Member tool::visualization_iterator_4_2::set_vtk_data (const sec_vd &xcoords, vtkDoubleArray *xvtk_pt_coords, block< int > &xvtk_cell_types, block< vtkIdType > &xvtk_cell_connectivity, block< int > &xvtk_cell_type_map, vtkIdTypeArray *xvtk_pt_label_ids, vtkIdTypeArray *xvtk_cell_label_ids)

required because discretization_context::values capacity is fixed at 3.

Member tool::visualization_iterator_4_2::set_vtk_data (const sec_vd &xcoords, const sec_vd &xprop, vtkDoubleArray *xvtk_pt_coords, vtkDoubleArray *xvtk_prop, int xprop_tensor_rank, block< int > &xvtk_cell_types, block< vtkIdType > &xvtk_cell_connectivity, block< int > &xvtk_cell_type_map, vtkIdTypeArray *xvtk_pt_label_ids, vtkIdTypeArray *xvtk_cell_label_ids)

required because discretization_context::values capacity is fixed at 3.

Member tool::visualization_iterator_4_3::next (bool xtruncate)

_coord_state down_set is convenient place to store all members at or above the elements that have been visited for efficient unmarking in visualization_iterator::reset. Note that since _bove.vis_eval has not yet been evaluated for this iteration, it is still true if we are at an element.

Member tool::visualization_iterator_4_3::set_vtk_data (const sec_vd &xcoords, vtkDoubleArray *xvtk_pt_coords, block< int > &xvtk_cell_types, block< vtkIdType > &xvtk_cell_connectivity, block< int > &xvtk_cell_type_map, vtkIdTypeArray *xvtk_pt_label_ids, vtkIdTypeArray *xvtk_cell_label_ids)

required because discretization_context::values capacity is fixed at 3.

Member tool::visualization_iterator_4_3::set_vtk_data (const sec_vd &xcoords, const sec_vd &xprop, vtkDoubleArray *xvtk_pt_coords, vtkDoubleArray *xvtk_prop, int xprop_tensor_rank, block< int > &xvtk_cell_types, block< vtkIdType > &xvtk_cell_connectivity, block< int > &xvtk_cell_type_map, vtkIdTypeArray *xvtk_pt_label_ids, vtkIdTypeArray *xvtk_cell_label_ids)

required because discretization_context::values capacity is fixed at 3.

Member tool::vtk_unstructured_grid_builder::build_pa (const sec_vd &xcoords, const sec_vd &xproperty1, const sec_vd &xproperty2, vtkUnstructuredGrid &xresult)

a temporary restriction, until if below generalized.

if the coordinates are a multi-section and the property is defined on some smaller chunk than the coordinates, we currently have no way to compute the vertex count for the restriction of the coordinates to the base space of the property. So:

Member tool::vtk_unstructured_grid_builder::build_pa (const sec_vd &xcoords, const block< sec_vd *> &xproperties, vtkUnstructuredGrid &xresult)

a temporary restriction, until if below generalized.

if the coordinates are a multi-section and the property is defined on some smaller chunk than the coordinates, we currently have no way to compute the vertex count for the restriction of the coordinates to the base space of the property. So:

we don't need to iterate over the coordinates now, but the iterators don't know how to avoid it.

Member tool::vtk_unstructured_grid_builder::build_pa (const sec_vd &xcoords, const sec_vd &xproperty, vtkUnstructuredGrid &xresult)

a temporary restriction, until if below generalized.

if the coordinates are a multi-section and the property is defined on some smaller chunk than the coordinates, we currently have no way to compute the vertex count for the restriction of the coordinates to the base space of the property. So: