[ 
https://issues.apache.org/jira/browse/FLINK-21532?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17292688#comment-17292688
 ] 

Lsw_aka_laplace edited comment on FLINK-21532 at 3/1/21, 7:43 AM:
------------------------------------------------------------------

[~twalthr] Thx for ur reply.  Well, for the current issue, shall we just not 
normalize the key?  Actually it seem to work well for now in our own flink 
code. So I’m wondering it is reasonable or not to make this code change. Well, 
I'm sure this code change will cause some problem or not.


was (Author: neighborhood):
[~twalthr] Thx for ur reply.  Well, for the current issue, shall we just not 
normalize the key?  Actually  it works well for now in our own flink code. So 
I’m wondering it is reasonable or not to make this code change. Well, I'm sure 
this code change will cause some problem or not.

> Make CatalogTableImpl#toProperties and CatalogTableImpl#fromProperties case 
> sensitive
> -------------------------------------------------------------------------------------
>
>                 Key: FLINK-21532
>                 URL: https://issues.apache.org/jira/browse/FLINK-21532
>             Project: Flink
>          Issue Type: Improvement
>          Components: Table SQL / API
>            Reporter: Lsw_aka_laplace
>            Priority: Major
>
> For now, both legacy Table/SQL API and implementation and current Table/SQL 
> API and implementation are case sensitive when it comes to the key of table 
> properties. Though it is highly recommended  to be full of low case, key in 
> upper case still works well.
> But the following case upon current code seems a little misleading..
> Given a Ddl sql:
> """
>  create table a (f string) with ('K1' = 'xxx')
> """ 
> The property of corresponding `CatalogTableImpl` is  Map(K1->xxx).
> After calling `CatalogTableImpl#toProperties` and then 
> `CatalogTableImpl#fromProperties`, there comes a `CatalogTableImpl` with 
> properity:Map(k1->xxx). The letter in upper case has been converted into low 
> case.
>  
> After reading code, the reason seems that the two method mentioned above 
> normalize key by default and can not be configured.
>  
> As far as I'm concerned, There is an easy way that just  ` 
> DescriptorProperties descriptorProperties = new DescriptorProperties(false)`, 
> when means not to normalize key to low case.
>  
> While, for the fear that this should break some underlying rules and cause 
> some unexpected incompatibility. This part shall be under well discussed.
> What's  more, a better alternate way is welcomed~
>  
>  
>  



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to