Andrey, > Is this [1] a full scope of the tickets that MUST be resolved before the engine could be merged? Yes, we must resolve at least these tickets before merging. If you see any other release blockers fill free to attach them to this ticket.
> I think we have to add instructions to the readme file on how to turn a new SQL engine on. Sure, I think it should be the part of documentation ticket. > Also, I don't like the module name "ignite-calcite", because Calcite is an independent project. Personally, I see no problems here (but it's discussable). We have a lot of modules where the name is an independent project: "ignite-kafka", "ignite-spring", "ignite-kubernetes", "ignite-log4j", "ignite-zookeeper", etc. > So, would you mind renaming the module to e.g. "ignite-sql-engine" or "ignite-sql"? Module "ignite-indexing" also contains SQL engine, so names like "ignite-sql-engine" or "ignite-sql" will be ambiguous. чт, 30 дек. 2021 г. в 13:54, Andrey Mashenkov <andrey.mashen...@gmail.com>: > Alex, > it would be great to release a new SQL engine in 2.13 as an > experimental feature. > > Is this [1] a full scope of the tickets that MUST be resolved before the > engine could be merged? > I think we have to add instructions to the readme file on how to turn a new > SQL engine on. > > Also, I don't like the module name "ignite-calcite", because Calcite is an > independent project. > and Ignite just uses it. > So, would you mind renaming the module to e.g. "ignite-sql-engine" or > "ignite-sql"? > > [1] https://issues.apache.org/jira/browse/IGNITE-15436 > > On Thu, Dec 30, 2021 at 11:10 AM Zhenya Stanilovsky > <arzamas...@mail.ru.invalid> wrote: > > > > > Alex, great ! > > If someone wants to touch codebase somehow plz use this branch [1] > > Test passed can be found here [2] [3] > > > > [1] https://github.com/apache/ignite/tree/sql-calcite/modules/calcite > > [2] > > > https://github.com/apache/ignite/tree/sql-calcite/modules/calcite/src/test/java/org/apache/ignite/internal/processors/query/calcite > > [3] > > > https://github.com/apache/ignite/tree/sql-calcite/modules/calcite/src/test/sql > > > > > > > >> > > >>>Hello, Igniters! > > >>> > > >>>As you may already know there is the new Ignite SQL engine based on > > Apache > > >>>Calcite currently under development. > > >>> > > >>>Reasons to move from H2-based engine and motivation for creating the > new > > >>>one in details described in IEP-37 [1]. > > >>> > > >>>You can find all related to the new engine source code changes in the > > >>>"sql-calcite" branch [2]. > > >>> > > >>>Calcite-based SQL engine is not production-ready yet and has a lot of > > known > > >>>issues. In the future, the new engine should be fully independent of > > >>>"ignite-indexing" and H2, but now it relies on schema management and > > >>>indexes implemented in the "ignite-indexing" module and can't work > > without > > >>>the old engine. Despite all of the above mentioned, in the current > > state, > > >>>it has its own parsing, planning and execution flow and is almost as > > >>>functional as the H2-based SQL engine. > > >>> > > >>>Some users are already interested in the Calcite-based engine and > asking > > >>>about the development status and release dates. Calcite-based SQL > engine > > >>>will be the only SQL engine in Ignite 3.0. Perhaps even in 2.x we can > > get > > >>>rid of the H2-based engine at some time in the future. There is some > > syntax > > >>>difference between Calcite and H2 (Calcite is closer to SQL standards > > than > > >>>H2) and a totally new execution flow. After the release of this > feature, > > >>>users can try their queries and determine if any adaptation for them > is > > >>>required. With the new planning and execution flow, perhaps, some > > queries > > >>>will be executed more effectively, users can redirect such queries to > > the > > >>>new engine. > > >>> > > >>>I think we can provide an opportunity to users to try the new engine > and > > >>>release it as an experimental feature with the next Apache Ignite > > version > > >>>(2.13). > > >>> > > >>>What do you think? > > >> > > >> > > >> > > >> > > > > -- > Best regards, > Andrey V. Mashenkov >
