绑定变量虽然有很多优点，但在某些情况下不可以使用绑定变量，
比如，对于被绑定的字段的值，如果对于不同的值，Oracle应该使用不同的查询计划，这种情况下就不应该使用绑定变量。
以下的例子说明了这个问题：
create table records_to_be_processedasselect decode( mod(rownum,100), 0, 'N', 'Y' ) processed, a.* from all_objects a;
create index processed_idx on records_to_be_processed(processed);
analyze table records_to_be_processed compute statisticsfor tablefor all indexesfor all indexed columns/
Alter system flush shared_pool;Alter session set sql_trace=true;variable processed varchar2(1);exec :processed := 'N';select * from records_to_be_processed initially_N where processed = :processed;
exec :processed := 'Y';select * from records_to_be_processed initially_N where processed = :processed;
exec :processed := 'Y';select * from records_to_be_processed initially_Y where processed = :processed;
exec :processed := 'N';select * from records_to_be_processed initially_Y where processed = :processed;set autotrace off
tkprof的结果摘录如下（设置aggregate=no）：
查询1：
select * from records_to_be_processed initially_N where processed = :processed
call count cpu elapsed disk query current rows------- ------ -------- ---------- ---------- ---------- ---------- ----------Parse 1 0.00 0.00 0 0 0 0Execute 1 0.00 0.00 0 0 0 0Fetch 35 0.00 0.00 0 532 0 496------- ------ -------- ---------- ---------- ---------- ---------- ----------total 37 0.00 0.01 0 532 0 496
Misses in library cache during parse: 1Optimizer mode: ALL_ROWSParsing user id: 5
Rows Row Source Operation------- --------------------------------------------------- 496 TABLE ACCESS BY INDEX ROWID RECORDS_TO_BE_PROCESSED (cr=532 pr=0 pw=0 time=9940 us) 496 INDEX RANGE SCAN PROCESSED_IDX (cr=36 pr=0 pw=0 time=4966 us)(object id 52912)
查询2：select * from records_to_be_processed initially_N where processed = :processed
call count cpu elapsed disk query current rows------- ------ -------- ---------- ---------- ---------- ---------- ----------Parse 1 0.00 0.00 0 0 0 0Execute 1 0.00 0.00 0 0 0 0Fetch 3279 0.20 0.27 0 7289 0 49157------- ------ -------- ---------- ---------- ---------- ---------- ----------total 3281 0.20 0.27 0 7289 0 49157
Misses in library cache during parse: 0Optimizer mode: ALL_ROWSParsing user id: 5
Rows Row Source Operation------- --------------------------------------------------- 49157 TABLE ACCESS BY INDEX ROWID RECORDS_TO_BE_PROCESSED (cr=7289 pr=0 pw=0 time=491723 us) 49157 INDEX RANGE SCAN PROCESSED_IDX (cr=3369 pr=0 pw=0 time=196654 us)(object id 52912)
查询3：
select * from records_to_be_processed initially_Y where processed = :processed
call count cpu elapsed disk query current rows------- ------ -------- ---------- ---------- ---------- ---------- ----------Parse 1 0.00 0.00 0 0 0 0Execute 1 0.00 0.00 0 0 0 0Fetch 3279 0.21 0.21 0 3925 0 49157------- ------ -------- ---------- ---------- ---------- ---------- ----------total 3281 0.21 0.21 0 3925 0 49157
Misses in library cache during parse: 1Optimizer mode: ALL_ROWSParsing user id: 5
Rows Row Source Operation------- --------------------------------------------------- 49157 TABLE ACCESS FULL RECORDS_TO_BE_PROCESSED (cr=3925 pr=0 pw=0 time=196768 us)
查询4：
select * from records_to_be_processed initially_Y where processed = :processed
call count cpu elapsed disk query current rows------- ------ -------- ---------- ---------- ---------- ---------- ----------Parse 1 0.00 0.00 0 0 0 0Execute 1 0.00 0.00 0 0 0 0Fetch 35 0.00 0.00 0 734 0 496------- ------ -------- ---------- ---------- ---------- ---------- ----------total 37 0.00 0.00 0 734 0 496
Misses in library cache during parse: 0Optimizer mode: ALL_ROWSParsing user id: 5
Rows Row Source Operation------- --------------------------------------------------- 496 TABLE ACCESS FULL RECORDS_TO_BE_PROCESSED (cr=734 pr=0 pw=0 time=6489 us)
这个结果说明，如果使用了绑定变量，Oracle在第一次查询时，会通过绑定变量窥视，确定使用何种查询计划，
以后就使用完全相同的查询计划，不再考虑因不同的输入值而更新查询计划，这样做在有些情况下会导致数据库
执行错误的查询计划。
因此，对于这种情况，正确的做法是不使用绑定变量。
