[
https://issues.apache.org/jira/browse/IOTDB-622?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17094083#comment-17094083
]
Xiangdong Huang commented on IOTDB-622:
---------------------------------------
Hi,
Maybe, we can support more...
Let's see an example,
suppose we have a table:
name: table1
| country | carID | s1| timestamp|
in which (country, carID, timestamp) is the primary key.
In IoTDB, we can define paths like root.table.some_country.some_carID.
We all know in RDB, we can write SQLs like:
* 1. select count (s1) from table1
* 2. select count (s1) from table1 group by country
* 3. select count (s1) from table1 group by carID
* 4. select count (s1) from table1 group by country,carID
If we have only "IN TOTAL" keyword, then it matches the RDB-SQL1.
If, we want to support RDB-SQL2 and 3, how about add some other keyword in
"group by" statement.
e.g.,
* SQL1 -> "select count (s1) from root group by (1m, [start time, end time)),
level N"
* SQL2 -> "select count (s1) from root group by (1m, [start time, end time)),
level 3"
* SQL3 -> "select count (s1) from root group by (1m, [start time, end time)),
level 4"
* SQL4 -> "select count (s1) from root group by (1m, [start time, end time)),
level 1"
So, the grammar is :
SELECT ...
FROM ...
GROUP BY time_interval | LEVEL n | time_interval, LEVEL n.
where "level n" means the N-th level of the time series path.
if there is no "level" specified, n =INT.MAX, which means group by each
timeseries;
if there is no "time_interval", which means do not downsampling data;
if n > the max level in a given path set, then n = the max level in the given
path set.
> count all records of the whole database
> ---------------------------------------
>
> Key: IOTDB-622
> URL: https://issues.apache.org/jira/browse/IOTDB-622
> Project: Apache IoTDB
> Issue Type: Improvement
> Components: Core/Engine
> Reporter: WangChao
> Assignee: WangChao
> Priority: Trivial
>
> Now the count function only count the number of every timeseries, it's useful
> to count the total record of the specific path.
>
> like:
>
> *_select count(*) from root.ln._** should return the total number of the
> records in storage group root.ln, but now it returns the number of each
> timeseries which is in root.ln.
>
> Please leave your opinion.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
