Thread: Odbc driver

Byron,

Postgres 6.3.2, Odbc driver 6.40.0004


I get this

'Access violation at address xxxxxxx in module 'PSQLODBC.DLL'.
Read of address yyyyyyyy'

calling SQLGetData( hstmt, FieldNo, SQL_BINARY, SQLPOINTER( @i), 0, @iv);

note the 5th parameter -> 0

over a TEXT field with sizes of 4000 or more bytes.

if I try it with small fields (3000 or so bytes) there is no problem.

ok, I tracked it down and it seems to work until iv=4099,
when it must return 4100, boom !

I wonder why a text saved by Notepad is 4183 bytes and this
same text within Postgres is reported as 4099 chars long, but
maybe this is another history.

Sergio

Sergio wrote:
>
> Byron,
>
> Postgres 6.3.2, Odbc driver 6.40.0004
>

<snipped real problem, since I know nothing about it!>

(although, I do know that the communications protocol changed 6.3->6.4.
Might an upgrade help?)

>
> I wonder why a text saved by Notepad is 4183 bytes and this
> same text within Postgres is reported as 4099 chars long, but
> maybe this is another history.

Hmm, this text wouldn't have 84 lines, would it? That'd be an average of
~49 characters per line, so it's reasonable. DOS/Win uses CR/LF line
ending pairs, while the unix backend only stores LF (or is that CR?
Doesn't matter, 2 bytes vs. one byte)

Just think of it as the earliest example of Microsoft code bloat ;-)

Ross
--
Ross J. Reedstrom, Ph.D., <reedstrm@rice.edu>
NSBRI Research Scientist/Programmer
Computer and Information Technology Institute
Rice University, 6100 S. Main St.,  Houston, TX 77005

Sergio <ser@perio.unlp.edu.ar> el día Tue, 30 Mar 1999 12:40:10 -0300,
escribió:

>
>Byron,
>
>Postgres 6.3.2, Odbc driver 6.40.0004
>
>
>I get this
>
>'Access violation at address xxxxxxx in module 'PSQLODBC.DLL'.
>Read of address yyyyyyyy'
>
>calling SQLGetData( hstmt, FieldNo, SQL_BINARY, SQLPOINTER( @i), 0, @iv);
>
>note the 5th parameter -> 0
>
>over a TEXT field with sizes of 4000 or more bytes.
>
>if I try it with small fields (3000 or so bytes) there is no problem.
>
>ok, I tracked it down and it seems to work until iv=4099,
>when it must return 4100, boom !

I _think_ I found the problem:

in convert.c in the function copy_and_convert_field, you declare

char tempBuf[TEXT_FIELD_SIZE+5];

wich is fixed at _compile_ time, no ?, so the TEXT_FIELD_SIZE wich
is an option in the Odbc driver cfg, is ignored.

I'm rigth ? (because changing Max LongVarChar in the cfg doesn't
seems to affect nothing)

Sergio