InnoDB Plugin  1.0
Data Fields
plan_t Struct Reference

#include <row0sel.h>

Collaboration diagram for plan_t:
Collaboration graph

Data Fields

btr_pcur_t pcur
ibool asc
ibool pcur_is_open
ibool cursor_at_end
ibool stored_cursor_rec_processed
que_node_t ** tuple_exps
ulint mode
ulint n_exact_match
ibool unique_search
ulint n_rows_fetched
ulint n_rows_prefetched
ulint first_prefetched
ibool no_prefetch
sym_node_list_t columns
ut_list_base< func_node_tend_conds
ut_list_base< func_node_tother_conds
ibool must_get_clust
ulint * clust_map
btr_pcur_t clust_pcur

Detailed Description

Query plan

Field Documentation

ibool plan_t::asc

TRUE if cursor traveling upwards

ulint* plan_t::clust_map

map telling how clust_ref is built from the fields of a non-clustered record

btr_pcur_t plan_t::clust_pcur

if index is non-clustered, we use this pcur to search the clustered index

dtuple_t* plan_t::clust_ref

the reference to the clustered index entry is built here if index is a non-clustered index

sym_node_list_t plan_t::columns

symbol table nodes for the columns to retrieve from the table

ibool plan_t::cursor_at_end

TRUE if the cursor is open but we know that there are no more qualifying rows left to retrieve from the index tree; NOTE though, that there may still be unprocessed rows in the prefetch stack; always FALSE when pcur_is_open is FALSE

ut_list_base< func_node_t > plan_t::end_conds

conditions which determine the fetch limit of the index segment we have to look at: when one of these fails, the result set has been exhausted for the cursor in this index; these conditions are normalized so that in a comparison the column for this table is the first argument

ulint plan_t::first_prefetched

index of the first cached row in select buffer arrays for each column

dict_index_t* plan_t::index

table index used in the search

ulint plan_t::mode

search mode: PAGE_CUR_G, ...

ibool plan_t::must_get_clust

TRUE if index is a non-clustered index and we must also fetch the clustered index record; this is the case if the non-clustered record does not contain all the needed columns, or if this is a single-table explicit cursor, or a searched update or delete

ulint plan_t::n_exact_match

number of first fields in the search tuple which must be exactly matched

ulint plan_t::n_rows_fetched

number of rows fetched using pcur after it was opened

ulint plan_t::n_rows_prefetched

number of prefetched rows cached for fetch: fetching several rows in the same mtr saves CPU time

ibool plan_t::no_prefetch

no prefetch for this table

mem_heap_t* plan_t::old_vers_heap

memory heap used in building an old version of a row, or NULL

ut_list_base< func_node_t > plan_t::other_conds

the rest of search conditions we can test at this table in a join

btr_pcur_t plan_t::pcur

persistent cursor used to search the index

ibool plan_t::pcur_is_open

TRUE if pcur has been positioned and we can try to fetch new rows

ibool plan_t::stored_cursor_rec_processed

TRUE if the pcur position has been stored and the record it is positioned on has already been processed

dict_table_t* plan_t::table

table struct in the dictionary cache

dtuple_t* plan_t::tuple

search tuple

que_node_t** plan_t::tuple_exps

array of expressions which are used to calculate the field values in the search tuple: there is one expression for each field in the search tuple

ibool plan_t::unique_search

TRUE if we are searching an index record with a unique key

The documentation for this struct was generated from the following file: