NK/FLIP-303%3A+Support+REPLACE+TABLE+AS+SELECT+statement
Best regards,
Yuxia
- 原始邮件 -
发件人: "Timo Walther"
收件人: "dev"
发送时间: 星期三, 2024年 6 月 12日 下午 10:19:14
主题: Re: [DISCUSS] FLIP-463: Schema Definition in CREATE TABLE AS
Statement
I just noticed the CREATE TABLE
Looking forward for this feature.
>> > > >
>> > > > Regards,
>> > > > Jeyhun
>> > > >
>> > > > On Thu, Jun 13, 2024 at 4:48 PM Sergio Pena
>> > > > > >
>> > > > wrote:
>> > > >
>&g
>
> > > > wrote:
> > > >
> > > > > Sure Yuxia, I just added the support for RTAS statements too.
> > > > >
> > > > > - Sergio
> > > > >
> > > > > On Wed, Jun 12, 2024 at 8:22 PM yuxia >
> &
>
> > > > > Hi, Sergio.
> > > > > Thanks for driving the FLIP. Given we also has REPLACE TABLE AS
> > > > > Statement[1] and it's almost same with CREATE TABLE AS Statement,
> > > > > would you mind also support
> Statement
> > > > to CREATE TABLE AS Statement
> > > >
> > > >
> > > > [1]
> > > >
> > >
> >
> https://cwiki.apache.org/confluence/display/FLINK/FLIP-303%3A+Support+REPLACE+TABLE+AS+SELECT+statement
>
; >
> > > [1]
> > >
> >
> https://cwiki.apache.org/confluence/display/FLINK/FLIP-303%3A+Support+REPLACE+TABLE+AS+SELECT+statement
> > >
> > > Best regards,
> > > Yuxia
> > >
> > > - 原始邮件 -
> > > 发件人: "
> Yuxia
> >
> > ----- 原始邮件 -----
> > 发件人: "Timo Walther"
> > 收件人: "dev"
> > 发送时间: 星期三, 2024年 6 月 12日 下午 10:19:14
> > 主题: Re: [DISCUSS] FLIP-463: Schema Definition in CREATE TABLE AS
> Statement
> >
> > > I just noticed t
LECT+statement
>
> Best regards,
> Yuxia
>
> - 原始邮件 -
> 发件人: "Timo Walther"
> 收件人: "dev"
> 发送时间: 星期三, 2024年 6 月 12日 下午 10:19:14
> 主题: Re: [DISCUSS] FLIP-463: Schema Definition in CREATE TABLE AS Statement
>
> > I just noticed the CREATE T
atement
to CREATE TABLE AS Statement
[1]
https://cwiki.apache.org/confluence/display/FLINK/FLIP-303%3A+Support+REPLACE+TABLE+AS+SELECT+statement
Best regards,
Yuxia
- 原始邮件 -
发件人: "Timo Walther"
收件人: "dev"
发送时间: 星期三, 2024年 6 月 12日 下午 10:19:14
主题: Re: [DISCUSS] FLIP-46
> I just noticed the CREATE TABLE LIKE statement allows the definition
> of new columns in the CREATE part. The difference
> with this CTAS proposal is that TABLE LIKE appends the new columns at
> the end of the schema instead of adding them
> at the beginning like this proposal and Mysql do.
Thi
Hey Yanquan,
I just added the WITH clause to the CTAS syntax in the flip.
> If we have a source table with primary keys and partition keys defined,
> what is the default behavior if PARTITIONED and DISTRIBUTED not specified
> in the CTAS statement, It should not be inherited by default？
You're ri
Hi Sergio, thanks for driving it, +1 for this.
I have some comments:
1. If we have a source table with primary keys and partition keys defined,
what is the default behavior if PARTITIONED and DISTRIBUTED not specified
in the CTAS statement, It should not be inherited by default？
2. I suggest provi
I just noticed the CREATE TABLE LIKE statement allows the definition of new
columns in the CREATE part. The difference
with this CTAS proposal is that TABLE LIKE appends the new columns at the
end of the schema instead of adding them
at the beginning like this proposal and Mysql do.
> create table
Thanks Timo for answering Jeyhun questions.
To add info more about your questions Jeyhun. This proposal is not handling
NULL/NOT_NULL types. I noticed that
the current CTAS impl. (as Timo said) adds this constraint as part of the
resulting schema. And when defining
a primary key in the CREATE part
Hi Sergio,
thanks for proposing this FLIP for finalizing the CTAS statement.
Adopting the logic from MySQL for deriving and potentially overwriting
parts of the schema should be easy to understand for everyone. So +1 for
the FLIP in general.
> How do you handle CTAS statements with SELECT cl
Hi Sergio,
Thanks for driving this FLIP. +1 for it.
I have a few questions:
- How do you handle CTAS statements with SELECT clauses that have (implicit
or explicit) NULLABLE or NOT NULLABLE columns? Could you also mention the
casting rules in the FLIP for this case?
- In some cases, redefining th
Hi David,
The CTAS feature is already part of Flink (proposed in FLIP-218 [1]). The
new FLIP-463 is just to extend the CTAS syntax to allow for adding new
columns to
the created table that are not part of the generated schema. I think
FLIP-218 [1] was discussed in the mail list somewhere, but I co
Hi Sergio,
Sounds good . I am relatively new to this area and had some basic questions:
I notice in [1] it talks of materialized views. And CREATE view can already
take the AS keyword. It would be useful to me to understand when we would use
each of these.
- I assume the table will be kept up t
18 matches
Mail list logo
