Retired Note one #mysql = Lambda Sql:sql + & #39; Source Code 4 Explain Plan & #39;

The Mysql query run process is broadly divided into 4 stages:

Grammar analysis (sql_parse.cc< lexical analysis, grammar analysis, semantic check >)

>>sql_resolver.cc # Join.prepare

Generate logical Query plan (sql_optimizer.cc)

>># join.optimize

Generate physical query plan (sql_planner.cc)

Run the explain plan (sql_executor.cc) join.exec

Join.prepare ():

Redundant clause elimination for subqueries

In type subquery optimization Resolve_subquery ()

To optimize a simple subquery by converting a all/any type subquery to a MIN/MAX operation

Join.optimize ():

Subquery Pull-up flatten_subqueries ()

Optimize the outer connection for internal connection simplify_joins ()

Eliminate nested connections

WHERE clause, join/on clause, having clause simplification, expression simplification with constants, equation merging Optimize_cond () Opt_sum_query ()

Optimization does not have a GROUP BY clause under the COUNT (*) min () man ()

Determine multi-table connection path Make_join_statistice ()

Optimize distinct

Create temporary table storage temporary results optimized grouping sort Choose_table_order ()

In fact, logical and physical optimization, a bit confusing, not understood,

Is that after the physical query optimization, some logical optimizations are continued.

There are only two types of single-table scans:

Index Scan (others), table Scan (Jt_all)

Opt_range.h > class quick_select_i > Using index to scan total drop interface

See source code for several sub-categories;

Basic classes and structs:

Query Tree Class St_select_lex

Index struct St_key_create_information class Key_part_spec

Connection Table St_join_table

Connection Class Join:public Sql_alloc

Condition Cond_equal

Position struct St_position

Cost Estimation cost_estimate

SELECT now (), Curdate (), Curtime ()