Re: Fwd: Problem to get UTF8-CSV-File

On 22/12/11 20:40, Bill Ward wrote:
> Agreed. Any Oracle initialization environment variables need to be set
> before your Perl script starts. Another approach is to set them and then
> re-exec() from within the script ... though be careful to avoid infinitely
> re-execing the same script!

This is not correct. With the script as written, no Oracle related code
is executed until connect time, so enviroment variables which are interpreted
by Oracle can be set any time before the first connect.

On the other hand, variables interpreted by the run-time loader, such as
LD_LIBRARY_PATH, must be set before perl itself is loaded. So, re-execing
is necessary for a change to LD_LIBRARY_PATH, but not, but not for a
change of NLS_LANG.

>> hello ralf
>> hello jens
>>
>> Jens Rehsack schrieb:
>> [...]
>>
>>   We have a problem. We want to get an UTF8-CSV-File from
>>> UTF8-Oracle-Database. It seems to be that the CSV-File is always charset
>>> "latin-1". What can we do to get UTF-8-File?
>>>   Here is our Testscript.
>>>
>>> #------ our Testscript--------------------**----#
>>> #!/usr/local/bin/perl -w -C64 use strict;
>>>
>> [...]
>>
>>> $ENV{NLS_LANG}="GERMAN_**GERMANY.AL32UTF8";
>>>
>> [...]
>>
>> I seem to remember that it's to late for the oracle-client libraries to
>> set NLS_LANG here inside the script - I guess at this time they are loaded
>> already and have finished initializing their charset, thus ignoring any
>> change in NLS_LANG.
>>
>> try to set NLS_LANG before starting the script. either by "export"ing it
>> or by doing
>>
>> NLS_LANG="GERMAN_GERMANY.**AL32UTF8" ./your-test-script
>>
>> greetings
>> hermann
>>
>
>
>


-- 
Charles Jardine - Computing Service, University of Cambridge
[email protected]    Tel: +44 1223 334506, Fax: +44 1223 334679