一段存储过程，请大家帮忙优化
本帖最后由 zhangenming20080324 于 2010-05-07 16:21:23 编辑

CREATE PROCEDURE "loan".getpayschedule(dat date)
    returning varchar(255);
    
    define returnstr varchar(20);
    define v_duebillno varchar(20);--借据号
    define v_maxcurrentno INTEGER;--最大期次
    define v_duebno varchar(20);
    define v_sq INTEGER;
    define v_paydate varchar(20);
    define v_maturedate varchar(10);
    define count1 int;
    define v_gitid int;--自动id
    define v_currentnosc int;--提前还款上一次期次
    define v_gitid1 int;
    define sql_err,isam_err int;--异常处理

    SET DEBUG FILE TO "gittestcom.txt";
    let sql_err = 0;
    TRACE ON;
    begin
        on exception set sql_err 
            delete from Acc_PayScheduleTmp;
            delete from acc_payschedule;
            trace "sql_err--->"||sql_err;
            return "0";
        end exception with resume ;

        delete from Acc_PayScheduleTmp;
        delete from acc_payschedule;
        --从核心中间表导入所需数据岛临时表
       -- insert into Acc_PayScheduleTmp select  ap.*,TO_CHAR(today,'%Y%m%d')||seqhkjh.NEXTVAL from Acc_PaySchedule1 ap,loanduebill ld where ap.duebillno=ld.duebillno;
        insert into Acc_PayScheduleTmp  select contractno,duebillno,currentno,maturedate,currentsum,currentinterest,currentnormalinterest,currentoverdueinterest,currentoffsheetinterest,payoption,paydate,paysum,payinterest,currency,hexinappid,TO_CHAR(today,'%m%d')||seqhkjh.NEXTVAL from Acc_PaySchedule1 where duebillno in('000001000096011001','000001000167011001');
        foreach select duebillno into v_duebillno from Acc_PayScheduleTmp
            --开始轮循单条借据
           let count1=0;
           foreach select duebillno,currentno,paydate,maturedate,gitid into v_duebno,v_sq,v_paydate,v_maturedate,v_gitid from Acc_PayScheduleTmp where duebillno=v_duebillno
            if(v_sq==2) then--提前还款从2开始
                   let count1=count1+1;
                   if(count1==2) then --如果第二次提前还款跟新状态
                        select max(currentno) into v_maxcurrentno from Acc_PayScheduleTmp where duebillno=v_duebno and currentno>=2 and gitid<v_gitid; 
 
                        
                        foreach select currentno,gitid into v_currentnosc,v_gitid1 from Acc_PayScheduletmp where duebillno=v_duebno and gitid >v_gitid-1
                            update Acc_PayScheduleTmp set currentno=v_maxcurrentno+v_currentnosc-1 where duebillno=v_duebillno and gitid=v_gitid1;
                        end foreach
                end if
            end if
            end foreach
      end foreach
      insert into acc_payschedule select contractno,duebillno,currentno,maturedate,currentsum,currentinterest,currentnormalinterest,currentoverdueinterest,currentoffsheetinterest,payoption,paydate,paysum,payinterest,currency,hexinappid from acc_payscheduleTmp;
      let returnstr="1";
    end
    TRACE OFF;
    return returnstr;
end procedure;

============================================
acc_payscheduletmp有6W左右的数据，按照现在的方式执行半小时都跑不出来，请大家帮忙优化下，不太会用存储过程，先谢了

数据结构：
duebillno currentno gitid
000001000096011001 15071
000001000096011001 25072
000001000096011001 35073
000001000096011001 45074
000001000096011001 55075
000001000096011001 65076
000001000096011001 75077
000001000096011001 85078
000001000096011001 25079
000001000096011001 350710
000001000096011001 450711
[解决办法]
主要是两个foreach影响效率。能不能改改逻辑，去掉一个甚至消灭循环？
[解决办法]
可以看看update语法：
update tab1
set (a,b) =
((select c,d from tab2
where ......))
where ......
这个应该可以用于两张表关联的更新。