Re: PG8.3->10 migration data differences - Mailing list pgsql-general

From Adrian Klaver
Subject Re: PG8.3->10 migration data differences
Date
Msg-id d0f34873-c63c-6484-38bf-1fc707c19aaf@aklaver.com
Whole thread Raw
In response to Re: PG8.3->10 migration data differences  (Csaba Ragasits <ragasits.csaba@gmail.com>)
List pgsql-general
On 9/11/18 1:41 AM, Csaba Ragasits wrote:
> Hello,
> 
> We would like to migrate from 8.3 to 10 version. We've hundreds 
> databases with different structures. That reason we're working on an 
> automatic data comparing process.
> 
> I've found the following storage settings:
> - pg83: Date/time type storage:               64-bit integers
> - pg10: Date/time type storage:               64-bit integers
> 
> When I running the following select from psql (Select '09/10/18 
> 07:10:25.110'::timestamp;)
> The results are same:
> -pg83: 2018-09-10 07:10:25.11
> -pg10: 2018-09-10 07:10:25.11
> 
> When I select it from a table (Select v.entry_timestamp from t_vis v):
> The results are different:
> - pg83: 2015-08-28 21:25:07.70
> - pg10: 2015-08-28 21:25:07.7

Actually the result is the same:

select '2015-08-28 21:25:07.70'::timestamp = '2015-08-28 
21:25:07.7'::timestamp;

  ?column?
----------
  t

The formatting is different.

> 
> The field type:
> entry_timestamp TIMESTAMP WITHOUT TIME ZONE NOT NULL,
> 
> Our difference every time the last 0 values. The time values are same, 
> but our field based data comparing mechanism every time mark it as error.

As Ron stated it should not.

> 
> thx,
> Csaba
> 

-- 
Adrian Klaver
adrian.klaver@aklaver.com


pgsql-general by date:

Previous
From: Arup Rakshit
Date:
Subject: Why order by column not using index with distinct keyword in selectclause?
Next
From: Tom Lane
Date:
Subject: Re: Why order by column not using index with distinct keyword in select clause?