apache / apache/iotdb

In select into statement, default encoding type may cause compatibility problems

Open
#6,953 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Java
Stars
6.4k
Forks
1.2k
Avg merge
1d 23h
Merged PRs (30d)
115

Description

**Describe the bug**
1. In the documentation, GORILLA encoding type has a restriction. However, some select into statements will automatic create a timeseries with GORILLA encoding type(while the original timeseries is not a GORILLA type), which may be conflict with GORILLA's restriction.
If we can't change the default encoding type in select...into... clause, this problem may exist. Perhaps there will be other conficts in this select...into... clause.
2. I try to insert a -Double.MAX_VALUE point into the database, with other points all positive, which means the sum will not over Double; but it still returns -Infinity while I use a avg agggregation statement with group by.

**To Reproduce**
1. running the following iotdb-sql in iotdb-client:
SET STORAGE GROUP TO root.sg0;
CREATE TIMESERIES root.sg0.t1 WITH DATATYPE=DOUBLE, ENCODING=PLAIN;
INSERT INTO root.sg0(timestamp, t1) VALUES (9603, 0.732498412173562);
INSERT INTO root.sg0(timestamp, t1) VALUES (12492, 0.885314087043015);
INSERT INTO root.sg0(timestamp, t1) VALUES (39503, 0.08816170378111365);
INSERT INTO root.sg0(timestamp, t1) VALUES (25965, -1.7976931348623157E308);
INSERT INTO root.sg0(timestamp, t1) VALUES (78893, 0.47220064168506504);
select t1 into t0 from root.sg0
select avg(t1),avg(t0) from root.* group by([1000, 100000), 40000ms)

**Expected behavior**
1. There will be a -Double.MAX_VALUE in a GORILLA type timeseries(root.sg0.t0), which might be invalid? (I think there will also be a restriction in GORILLA while using Double datatype, but the document doesn't mention it)
2. There will be no Infinity exists in the result, since all single points are in the range of double, (though there is a point has a value of -Double.MAX_VALUE)

**Screenshots**
This is the result of the last query, which has -Infinity.
![image](https://user-images.githubusercontent.com/25195216/184050086-21002450-f58e-4d0c-838d-79e3617fc0a7.png)

This avg aggregation return -Infinity, while some other range of group by / sum aggregation will not return infinity.
![1660181660436](https://user-images.githubusercontent.com/25195216/184050285-88fe2821-6ad7-416b-8b3e-3a6c7d80b5b9.png)
![image](https://user-images.githubusercontent.com/25195216/184050315-16a0cdca-ef23-4c21-8b48-29936adac9df.png)

The encoding of t0 becomes GORILLA
![image](https://user-images.githubusercontent.com/25195216/184052028-8d9185c3-1a0f-4bee-9a32-cfa29003b1aa.png)

- OS: Windows 10
- Version: iotdb 0.13.1

Contributor guide

Open the contributing guide

Research direction

Start by reproducing the SQL sequence in iotdb-client, focusing on the select...into statement, GORILLA encoding, and avg with group by. Trace the paths handling automatic timeseries creation and aggregation to determine why the encoding restriction and -Infinity result occur. Done means the reproduced cases no longer create an incompatible encoding or return -Infinity for finite double values.

Written by the indexing model from the issue text.

Assessment

Tech stack
java
Domain
databases
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
30/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.