rust/greptimedb
1
0
Fork 0
mirror of https://github.com/GreptimeTeam/greptimedb.git synced 2026-01-08 22:32:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

modify docs (#213)

Browse Source
This commit is contained in:
egg
2022-08-30 18:56:21 +08:00
committed by GitHub
parent 787aab9c00
commit 38d5febafe
1 changed files with 4 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
docs/how-to/how-to-write-aggregate-function.md
View File

@@ -59,9 +59,11 @@ Once you know the meaning of each method, you can easily write your accumulator.
# 3. Register your aggregate function to our query engine.
You can call `register_aggregate_function` method in query engine to register your aggregate function. To do that, you have to new an instance of struct `AggregateFunctionMeta`. The struct has two fields, first is the name of your aggregate function's name. The function name is case-sensitive due to DataFusion's restriction. We strongly recommend using lowercase for your name. If you have to use uppercase name, wrap your aggregate function with quotation marks. For example, if you define an aggregate function named "my_aggr", you can use "`SELECT MY_AGGR(x)`"; if you define "my_AGGR", you have to use "`SELECT "my_AGGR"(x)`".
You can call `register_aggregate_function` method in query engine to register your aggregate function. To do that, you have to new an instance of struct `AggregateFunctionMeta`. The struct has three fields, first is the name of your aggregate function's name. The function name is case-sensitive due to DataFusion's restriction. We strongly recommend using lowercase for your name. If you have to use uppercase name, wrap your aggregate function with quotation marks. For example, if you define an aggregate function named "my_aggr", you can use "`SELECT MY_AGGR(x)`"; if you define "my_AGGR", you have to use "`SELECT "my_AGGR"(x)`".
The second field is a function about how to create your accumulator creator that you defined in step 1 above. Create creator, that's a bit intertwined, but it is how we make DataFusion use a newly created aggregate function each time it executes a SQL, preventing the stored input types from affecting each other. The key detail can be starting looking at our `DfContextProviderAdapter` struct's `get_aggregate_meta` method.
The second field is arg_counts ,the count of the arguments. Like accumulator `percentile`, caculating the p_number of the column. We need to input the value of column and the value of p to cacalate, and so the count of the arguments is two.
The third field is a function about how to create your accumulator creator that you defined in step 1 above. Create creator, that's a bit intertwined, but it is how we make DataFusion use a newly created aggregate function each time it executes a SQL, preventing the stored input types from affecting each other. The key detail can be starting looking at our `DfContextProviderAdapter` struct's `get_aggregate_meta` method.
# (Optional) 4. Make your aggregate function automatically registered.