Re: BUG #14038: substring cuts unicode char in half, allowing to save broken utf8 into table

I see, thanks Tom!

On 3/21/16, 5:04 PM, "Tom Lane" <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:

>Reece Pegues <RPegues(at)tripwire(dot)com> writes:
>> Looks like the database is created with ENCODING = 'SQL_ASCII'
>
>Basically what that does is defeats all encoding checks inside the
>backend; it'll store whatever bytes you give it. So yeah, substring()
>is expected to deal in bytes not characters in this encoding.
>
>> So I assume it was thus saving the data that way, and then if the client
>> encoding is utf8 it tried to encode to that and failed?
>
>If client declares its encoding, the backend will verify correct encoding
>before transmitting data; but if the database encoding is SQL_ASCII then
>no actual conversion happens, only a validity check at transmit/receive.
>
> regards, tom lane