你好, 
我有一个sql, 
select marketCode as market_code,
       CURRENT_TIMESTAMP as process_time,
       sum((CASE
               WHEN msgType = 'FUTURE' THEN matchedPrice
               WHEN msgType = 'SPOT' THEN matchedPrice
               ELSE 0.5 * (leg1Price + leg2Price)
           END) * matchedQty) as volume,
       sum(matchedQty) as currency_volume,
       first_value(matchedPrice) over (partition by marketCode ORDER BY
transTime) as vopen,
       max(matchedPrice) as vhigh,
       min(matchedPrice) as vlow,
       last_value(matchedPrice) over (partition by marketCode ORDER BY
transTime) as vclose
  from TickData
 where action = 'OrderMatched' and side = 'BUY'
 group by marketCode, HOP(transTime, INTERVAL '1' SECOND, INTERVAL '24'
hour)

执行的时候报, matchedPrice not being grouped, 我的理解matchedPrice也是group function里的,
为什么还要在group里, 如果加了这个group column,我的结果会不对。

还是说我用的first/last 聚合函数错了?  我想要的是这个窗口的最早和最晚的那个值。
谢谢!



--
Sent from: http://apache-flink.147419.n8.nabble.com/

Reply via email to