ignite-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ivan Pavlukhin <vololo...@gmail.com>
Subject Re: Question: Table scheme 'default'
Date Sat, 01 Feb 2020 07:20:55 GMT
> I think this case should fail with error like "only cache created tables
> can be removed with drop table. ", not with "scheme not found."
> SQL-engine is supposed to accept and understand values it returns itself.

Agree. Feel free to contribute a fix. My expectation here is that it
will not be a trivial one.

пт, 31 янв. 2020 г. в 18:22, Vladimir Steshin <vladsz83@gmail.com>:
>
> //Got a connectionConnection conn = ...;
>
> //execute() - is just a helper function. Creates prepared statement, pass
> params...
> //Get all the tables
> List<List<?>> lst = execute(conn, "select SCHEMA_NAME, TABLE_NAME from
> SYS.TABLES");
>
> for( List<?> row : lst ){
>         String schemaName = (String)row.get(0);
>         String tableName = (String)row.get(1);
>
>         //Shows: "schema: default, table: PERSON"
>         System.out.println("schema: " +  schemName + ", table: " +
> tableName));
>
>         //Fails with with: java.sql.SQLException: Failed to parse query.
> Схема "DEFAULT" не найдена
>         execute( conn, "drop table "+schemaName + "."+tableName+"'" );
> }
>
> I think this case should fail with error like "only cache created tables
> can be removed with drop table. ", not with "scheme not found."
> SQL-engine is supposed to accept and understand values it returns itself.
>
>
> пт, 31 янв. 2020 г. в 11:29, Ivan Pavlukhin <vololo100@gmail.com>:
>
> > Vladimir,
> >
> > I will try to shed a light on this.
> > 1. Statically configured tables in caches (QueryEntity) by default
> > resides in a schema having the same name as a cache it belongs to
> > ("default" is a cache and a schema name for a table Person in your
> > example).
> > 2. As you might know SQL literals are case-insensitive if they used
> > without quotes. And Ignite SQL uses a quite naive approach for a
> > case-insensitive comparison. It simply converts all case-insensitive
> > names to upper case and compares after that.
> > 3. And we observe consequences of such naive comparison. A schema name
> > (equal to a cache name) is treated case-sensitive and saved as is
> > ("default" in lower case). Query "SELECT ... FROM default.Person"
> > passes schema name as case-insensitive. Ignite converts a schema from
> > the query to upper case "DEFAULT" and searches for a schema with that
> > name (using ordinary case-sensitive string comparison). But as there
> > is only "default" in lower case it fails to find a schema.
> >
> > So, it is a bug. But for me personally it looks like the bug causes a
> > not major usability problem.
> >
> > ср, 29 янв. 2020 г. в 12:27, Nikolay Izhikov <nizhikov@apache.org>:
> > >
> > > Vladimir.
> > >
> > > Please, provide a simple, self-contained reproducer to the issue you
> > described.
> > >
> > > > 29 янв. 2020 г., в 11:16, Vladimir Steshin <vladsz83@gmail.com>
> > написал(а):
> > > >
> > > > Hi ppl. Wanted to ask is issue a bug. Minor propably.
> > > >
> > > > Recently noticed that one simple query returns a value which can't be
> > found
> > > > 'as is' in following query. I tried to find and drop all the tables I
> > > > created with "create table":
> > > >
> > > > select schema_name, table_name, cache_name from sys.tables
> > > >
> > > > And there are tables which were not created with 'create table...'. For
> > > > example, table 'default.Person' from the test/tutorial I was
> > researching.
> > > > Ok. I tried to delete them too. Now I know this is prohibited. But why
> > I
> > > > got error: scheme 'default' not found? I just received this schema name
> > > > from previous query.
> > > >
> > > > To avoid this I had to capitalize schema name taking it into "":
> > > > "default".Person. Then I got normal error like 'only cache created
> > tables
> > > > can be removed with drop table.'.
> > > >
> > > > One select returns value which can't be found in second select
> > > > (delete/drop). Isn't it a bug from SQL's point of view?
> > >
> >
> >
> > --
> > Best regards,
> > Ivan Pavlukhin
> >



-- 
Best regards,
Ivan Pavlukhin

Mime
View raw message