Re: COPY question

On 21 October 2010 23:28, Raymond O'Donnell <rod(at)iol(dot)ie> wrote:

> On 21/10/2010 22:16, Szymon Guz wrote:
>
>> Hi,
>> I still get the same error while using COPY FROM 'file'. I have to pass
>> the full directory for this to work, example:
>>
>> COPY first (a,b,c) FROM '1st_file.csv' with csv header delimiter ',';
>> ERROR: could not open file "1st_file.csv" for reading: No such file or
>> directory
>>
> >
>
>> The same when I use: '.1st_file.csv' or './1st_file.csv'.
>>
>> No errors when I give the full path.
>>
>
> Yes, that's documented behaviour - you need to give the full path.
>
>
I've noticed that already. Btw, I could use the file name, but it won't be
searched in my current directory.
"The path will be interpreted relative to the working directory of the
server process (normally the cluster's data directory), not the client's
working directory."

>
>
> Giving the full path is stupid, how can I use just the file name?
>>
>
> Why is it stupid?
>

Because I can't just move my scripts/queries to some other directory, where
I have files with the same names, but different content (just for tests).

Thanks for the answer.

regards
Szymon