A multi-modal vector database that supports upserts and vector queries using unified SQL (MySQL-Compatible) on structured and unstructured data, while meeting the requirements of high concurrency and ultra-low latency.
Home Page: https://www.dingodb.com
License: Apache License 2.0
Null values are not allowed in dingo-expr operations. If the value of operands is null, exeptions may be thrown.
Shoud implemente the null value operation according to SQL standards.
Now, literal 'null' is parsed as a variable and compiled to a RtNull, but no further processing in operators is implemented.
Summary:
add datetime function:
Now(), curDate(), Current_Date, Current_Time, CurTime(), Current_Timestamp(),From_UnixTime(), Unix_TimeStamp(),Date_Format().
As says in title. Now only COUNT, SUM is supported.
There are two aggregation function SUM and SUM0 in Calcite logical plan. Sometimes SUM is transformed to CASE(COUNT(x) = 0 , null , SUM0(x)). Without SUM0 and CASE supporting, these quries would fail.
Currently only inner equal join is supported. Need to support left, right and full join.
Currently in DingoDB, there is only one schema named 'DINGO', which is statically bound to a single meta service.
If there are serveral meta services provided, some other schema names are needed.
A "schema name --- meta service provider" register mechanism is good for maximize flexibility.
In meta data methods like getSchemas, getTables, getColumns. Now the input strings are treated as the name of corresponding element. Mostly it is enough for use, but the JDBC requires that they are patterns containing % and _.
Locate throw Exception when encounter \ characterLocate Function cannot be eval when it appeared in where causeLocate Function return index from 1, return 0 when substring not existv0.3.0
No response
No response
To support equal inner join by hash join operation, spread the data to available computing nodes and run in parallel.
Other types of join need to be supported in future.
Especially in self join.
0.3.0-SNAPSHOT
No response
No response
Now, the information about the heartbeat from Executor to Coordinator cannot be used for region management and scheduler.
change some item names:
| old | new |
|---|---|
| default.replicas | store.default.replicas |
| default.partitions | store.default.partitions |
| kv.cluster.id | store.cluster.id |
| kv.cluster.name | store.cluster.name |
| kv.initialServerList | store.initialServerList |
| kv.failoverRetries | store.failoverRetries |
| kv.futureTimeoutMillis | store.futureTimeoutMillis |
| placement.driver.options.fake | coordinator.options.fake |
| placement.driver.options.pdGroupId | coordinator.options.raftGroupId |
| placement.driver.options.initialPdServerList | coordinator.options.initCoordinatorSrvList |
| placement.driver.options.cliOptions.timeoutMs | coordinator.options.cliOptions.timeoutMs |
| placement.driver.options.cliOptions.maxRetry | coordinator.options.cliOptions.maxRetry |
| placement.driver.options.cliOptions.rpcProcessorThreadPoolSize | coordinator.options.cliOptions.rpcProcessorThreadPoolSize |
Insert a record into a table, in which there is a column with constraint 'NOT NULL DEFAULT ...', the if the value of the column is not provided, the running of query will not use the default value, but throw exceptions.
0.3.0
No response
No response
ArrayIndexOutOfBoundException is thrown when query with join of two tables.
0.3.0-SNAPSHOT
No response
No response
When do 'INSERT ...' via thin driver, no data returns.
After the tag modification, the status of IdealState and ExternalView is inconsistent.
As these types can be compared, and should be able to do max, min aggreagation.
Now DingoDB supports only equi-join, need to support non-equi join.
As says in title. Not all fields of TableDefinition are mapped into meta info of coordinator.
0.4.0-SNAPSHOT
No response
No response
Support date, timestamp, datetime to dingo.
The DATE type is used for values with a date part but no time part. Dingo retrieves and displays DATE values in 'YYYY-MM-DD' format. The supported range is '1000-01-01' to '9999-12-31'.
The DATE type is used for values with a date part with time part. Dingo retrieves and displays DATE values in 'HH-mm-ss' format.
The DATETIME type is used for values that contain both date and time parts. Dingo retrieves and displays DATETIME values in 'YYYY-MM-DD hh:mm:ss' format. The supported range is '1000-01-01 00:00:00' to '9999-12-31 23:59:59'.
It returns less rows than the correct result set occasionally.
0.3.0-SNAPSHOT
No response
No response
[ ] CSV files with boolean type(default null) cannot be import
[ ] Import tools can be compute the import record count
v0.4.0
No response
No response
Do the following JDBC call:
DatabaseMetaData metaData = connection.getMetaData();
ResultSet resultSet = metaData.getTables(null, null, "%", null);Then NullPointerException is thrown.
0.3.0-SNAPSHOT
No response
No response
If we do a query like
select * from A join B on A.id = B.name where A.size >100
The query will fail.
0.3.0-SNAPSHOT
No response
No response
Driver server and driver client can be implemented on avatica's HttpServer and remote driver. Need to implement the server function in order to bridge remote request to local jdbc driver.
The store api now provide a count function to return the number of item in a table part. It can be used to accelerate count query if only the number of all items is concern.
In order to ensure the quality of code delivery, increase the CI/CD process using docker image
MySQL implements bool type as tinyint(1), so a == TRUE is different from a IS TRUE, the former returns false if a is not 1 because TRUE is 1, the latter returns true if a is a number and is not 0.
DingoDB do have bool type, but we need operators such as IS TRUE to mimic this behavior.
The purpose of a DingoDB Improvement Proposal (DIP) is to introduce any major change into DingoDB. This is required in order to balance the need to support new features, uses cases, while avoiding accidentally introducing half thought-out interfaces that cause needless problems when changed.
Any of the following should be considered a major change:
What are the "public interfaces" of the project? All of the following are public interfaces that people build around:
A DIP should contain the following sections:
Anyone can initiate a DIP, but preferably someone with the intention of implementing it.
Description of the problem to be solved.
Describe how the feature will be implemented, or the problem will be solved. If possible, include mocks, screenshots, or screencasts (even if from different tools).
Describe any new additions to the model, views or REST endpoints. Describe any changes to existing sub modules.
Describe any packages that are required. Are they actively maintained? What are their licenses?
Describe any database migrations that are necessary, or updates to stored URLs.
Describe alternative approaches that were considered and rejected.
Refactor the Dingo Server and implement the new server using raft KV.
Originally calcite jdbc driver is used to access DingoDB. Implement basic functionality in dingo-driver to replace it so we can simplify and customize more in 'sql -> db' process.
need to support CAST (sth as type) function. Actually, some CAST operation is added even not used in sql explicitly. For example, a AVG aggregation of INTEGER type requires the result be casted back to INTEGER.
Create a table
create table test(
id int,
data varchar array,
primary key(id)
)then insert some values
insert into test values (1, array['a', 'b', 'c'])the execution would fail.
0.4.0-SNAPSHOT
No response
No response
Currently only the primary fields of a table are used to construct the key of the kv store. If the store is a whole kv store, the keys of different table will be mixed up. So a fixed-length table id should be prepend to the key to identify data of the table.
NullpointerExeptionv0.4.0
No response
No response
to execute sql like "insert into test1 select * from test".
Projects and filters are allowed.
Generally a key-value store implementation don't need to know the row type of a table. Keep the encoding/decoding secret inside the exec engine can make the store implementation and resource management less coupled with the main execution layer of DingoDB.
The new api module can be called dingo-kvstore-api, the original 'dingo-store-localmodule can be migrated todingo-kvstore-rocks. Here explicit rocks` is used for there may be multiple storage implementation.
A columnar store api may be needed, so leave the dingo-stroe-api there for future work.
Null is a special value in sql. To predicate if a value is null or not, special function IS NULL and IS NOT NULL is needed.
As says in titile. Add AVG function in aggregation.
The result of expression evaluating are not correctly typed, make avro encoding fail for mismatched schema. So sometimes updating a table would crash.
0.3.0
No response
No response
The newest version of Apache Calcite is 1.30.0, which is matching the version of Avartica 1.20.0.
A new operator should be added to support the sorting of the query results.
According to the implementation of Calcite, 'limit' and 'offset' are supported in the same operator.
The table definition is stored in JSON format, so some fields are lost after reading.
0.4.0-SNAPSHOT
No response
No response
When the operators run, we can get some statistical datum, e.g. the timestamp of starting and stopping, the count of tuple processed. These datum can be pushed up with the finish token, and collected at the root of execution tree, for debug and profiling purpose.
When the size of the result set is larger than the fetch size, the same data would be returned again and again, because of the naive implementation of the fetch method in dingo-driver which only supports fetch data one-shot.
DingoDB目标是实现行、列的混合存储，当前Store-api层面有点混乱，造成行模式下本地rocksdb和分布式kv访问接口不清晰。
[ ] Support to create table from Annotation using Java Pojo
[ ] Support Date type of column
[ ] Support Time type of column
[ ] Support Timestamp type of column
[ ] Support boolean type of column
[ ] Support String type of column
A bug happened!
0.3.0-SNAPSHOT
No response
No response
DingoConfiguration does not support reading object configuration.
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
Django
The Web framework for perfectionists with deadlines.
Laravel
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft
Open source projects and samples from Microsoft.
Google
Google ❤️ Open Source for everyone.
Alibaba
Alibaba Open Source for everyone
D3
Data-Driven Documents codes.
Tencent
China tencent open source team.