How can i do that?

[rpc]

Hi all,
We have a problem with DOA.
We have Oracle 9i database with many tables. Some tables have varchar(4000) columns & use there many characters like "ř

 

[Marco Kalter]

What exactly do you mean by "With PL/SQL this isn't a problem"?

 

[rpc]

When I run PL/SQL developer and execute same SQL containg varchar2 variable, the PL/SQL developer do SQL correctly.
Same SQL executing with using D.O.A. it is incorrect.

DOA return error from Oracle, that the string in variable isn't terminated by zero character.

 

[rpc]

Error result returned by DOA from Oracle:

ORA-01480: trailing null missing from STR bind value

TOracleSession.BytesPerCharacter = bcAutoDetect,
code is UTF-8 & string length is more than 1333 characters.

 

[Marco Kalter]

As you probably know, PL/SQL Developer uses Direct Oracle Access as well. What version of Direct Oracle Access are you using?

 

[rpc]

Yes, I know.

I using DOA v 3.4.7 for Delphi.

 

[rpc]

BTW: This problem is only when we use bind variables.

 

[Marco Kalter]

There is no 3.4.7. Can you check again?

 

[rpc]

Sorry,

DOA v 3.4.6.4 (for Delphi 7)

 

[Marco Kalter]

Okay. Can you check if TOracleSession.BytesPerCharacter is set to bcAutoDetect? Can you also let me know if PL/SQL Developer and your Direct Oracle Access application are using the same Oracle Home?

 

[rpc]

Yes, TOracleSession.BytesPerCharacter is bcAutoDetect.

PL/SQL using same Oracle Home as DOA - It's same PC, same user profile, same database...

Function MaxVarchar return 1333 characters, because CharacterSize (autodetected by GetBytesPerCharacter) is 3 and UseOCI80 is true.

But 1333 isn't correct max len (

 

[Marco Kalter]

I tried to help (as my previous 4 replies might suggest), but overlooked your last post.

 

[rpc]

Dear Marco Kalter,
I am from Czech Republic, where we using same characters like "ě

 

[mglass]

I am using DOA 4.0.5, and still encountering this problem. It appears that DOA isn't correctly truncating the string as it should be (when using its worst-case estimates on string length).

When can we see this fixed? In particular, I would want to see more accurate estimates based on variable-length encoding, which is very common. Otherwise, one should be able to turn this feature off and let Oracle handle it.