On 01/05/2017 11:46 AM, Adrian Klaver wrote:
> On 01/05/2017 08:31 AM, Rob Sargent wrote:
>>
>>
>> On 01/05/2017 05:44 AM, vod vos wrote:
>>> I finally figured it out as follows:
>>>
>>> 1. modified the corresponding data type of the columns to the csv file
>>>
>>> 2. if null values existed, defined the data type to varchar. The null
>>> values cause problem too.
>>>
>>> so 1100 culumns work well now.
>>>
>>> This problem wasted me three days. I have lots of csv data to COPY.
>>>
>>>
>> Yes, you cost yourself a lot of time by not showing the original table
>> definition into which you were trying insert data.
>
> Given that the table had 1100 columns I am not sure I wanted to see it:)
>
> Still the OP did give it to us in description:
>
> https://www.postgresql.org/message-id/15969913dd3.ea2ff58529997.7460368287916683127%40zoho.com
>
> "I create a table with 1100 columns with data type of varchar, and
> hope the COPY command will auto transfer the csv data that contains
> some character and date, most of which are numeric."
>
> In retrospect I should have pressed for was a more complete
> description of the data. I underestimated this description:
>
> "And some the values in the csv file contain nulls, do this null
> values matter? "
>
>
My apologies for missing that. Was sure there would be room for some
normalization but so be it: OP's happy, I'm happy