Re: DBD::Oracle, 10g Lob Refetch problem
[email protected] (John Scoles)
| Newsgroups | perl.dbi.users |
|---|---|
| Message-ID | <[email protected]> |
On 20/12/2010 3:17 PM, Furst, Carl wrote: > Hello, > > I just built DBD::Oracle 1.26 on Solaris SPARC 2.10 using perl 5.8.5 32 bit > against client 10.0.2.4 > > We've been having trouble since day one. The biggest problem is that we are > having a problem writing LOB fields. We get the following error: > > DBD::Oracle::st execute failed: ORA-00903: invalid table name (DBD > ERROR: OCIStmtExecute/LOB refetch) > To start we will need to see the output from the perl Makefile.PL and the make, and make test to see if there is something wrong with the way it is being built > We think it's the LOB refetch that's causing the issue. > > the encoding of the database and the NLS_LANG parameter are both UTF8 > nls_lang specifically is AMERICAN_AMERICA.UTF8 > > If anyone has any advice about this, it would be a big help. > Next we will need to see an example of your code as there are at least 4 different ways to write lob fields with DBD::Oracle > My questions are the following: > 1) do we need an actual Oracle server to build the DBD - if so what libs > would we need to link against? No you do not need a server you only need the client. Instant client should work fine > 2) Has anyone else experienced this; building again lib32 client libs. Solaris SPARC tends to have more problems that other OS when attempting to compile DBD::Oracle in 32bit but since you have it working you should be past this point > 3) What role does oraperl have in all this? If oraperl fails to compile, is > that a blocker for DBD? > No it is a separate and deprecated hunk of code for perl 4. It may be a symptom of a deeper problem so we need to see the Makefile.pl and make output to know for sure. Cheers John Scoles > Thanks in advance, > > Carl Furst > CMS Developer > MLB Advanced Media > > > > > > > > ********************************************************** > > MLB.com: Where Baseball is Always On