Re: [DISCUSS] Sql-client lack support for new features

2020-07-06 Thread DONG, Weike
Thanks Jinsong for the suggestions. +1 for #1, as users are constantly puzzled by the inconsistencies between the Table / SQL module and SQL Client. Best, Weike On Mon, Jul 6, 2020 at 11:55 AM Jingsong Li wrote: > +1 for expanding the scope of #2, a SQL test framework(Should not be too > heav

Re: [DISCUSS] Sql-client lack support for new features

2020-07-05 Thread Jingsong Li
+1 for expanding the scope of #2, a SQL test framework(Should not be too heavy but clear) and guidelines for developers. Best, Jingsong On Thu, Jul 2, 2020 at 12:30 PM Kurt Young wrote: > Thanks Jingsong for bringing this up discussion and sorry for the late > reply. > > I'm in general +1 for #

Re: [DISCUSS] Sql-client lack support for new features

2020-07-01 Thread Kurt Young
Thanks Jingsong for bringing this up discussion and sorry for the late reply. I'm in general +1 for #1, and want to expand the scope of #2. First of all, I think the approach Jingsong proposed in #2 can help with covering more e2e use cases of SQL, which also draws a clean line between how to des

Re: [DISCUSS] Sql-client lack support for new features

2020-06-18 Thread Jark Wu
+1 for #1 I think this is what we are currently doing, that forward SQL statements to TableEnv#executeSql, e.g. FLINK-17113, FLINK-18059. But IMO the SQL CLI specific statements (EXIT, QUIT) should still stay only in SQL CLI. Another idea is that, the reviewer/committer should check tests are both

Re: [DISCUSS] Sql-client lack support for new features

2020-06-18 Thread Rui Li
Thanks Jingsong for bringing up the discussion. Perhaps we can do both #1 and #2? I like the idea that SQL Client should just forward SQL statements to TableEnvironment. IIUC, with TableEnvironment::executeSql in place, most statements can be forwarded. Only a very limited set of statements need to

Re: [DISCUSS] Sql-client lack support for new features

2020-06-18 Thread Benchao Li
Thanks Jingsong for bringing up this discussion! +1 for #1 I think it would always be a good practice to have a unify implementation for two different user-facing API of the same functionality. The divergence of two different implementations always bothers users. In the past, users complains a lo

[DISCUSS] Sql-client lack support for new features

2020-06-18 Thread Jingsong Li
Hi all, I'd like to start a discussion for the new features lacking support of Sql-client. I've seen the new DDL syntax that SQL client lacks support for many times. For every new DDL syntax, we need to add support in sql-client. Add a corresponding SqlCommand in sql-client, otherwise this DDL is