Re: Apparent DBD::Oracle 1.16 unicode regression
From: Lincoln A. Baxter (lab_at_lincolnbaxter.com)
Date: 12/02/04
- Next message: Greg Sabino Mullane: "Re: DBIx::DBH - Perl extension for simplifying database connections"
- Previous message: Lincoln A. Baxter: "Re: Apparent DBD::Oracle 1.16 unicode regression"
- Maybe in reply to: Lincoln A. Baxter: "Re: Apparent DBD::Oracle 1.16 unicode regression"
- Next in thread: Tim Bunce: "Re: Apparent DBD::Oracle 1.16 unicode regression"
- Reply: Tim Bunce: "Re: Apparent DBD::Oracle 1.16 unicode regression"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
To: Jesse Vincent <jesse@bestpractical.com> Date: Wed, 01 Dec 2004 21:00:23 -0500
On Wed, 2004-12-01 at 19:01 -0500, Jesse Vincent wrote:
>
>
> On Wed, Dec 01, 2004 at 06:59:18PM -0500, Lincoln A. Baxter wrote:
> > On Wed, 2004-12-01 at 12:18 -0500, Jesse Vincent wrote:
> > > The bug I'm describing isn't something
> > > I've managed to duplicate in-house.
Have you tried?
> Everyone who's hitting it is
> > > running perl 5.8.3 or newer (Some are running 5.8.5). They're all
> > > running against Oracle 9.2.0.x.
> >
> > So, you are allowing them to make a direct oracle connection? Assuming
> > you are, then, can you ask one of those complaining to at least provide
> > you with the output requested?
> >
> > I suspect an oracle client version mismatch with the server.
>
> Is there a change in 1.16 that would cause this to become an issue,
> since backing down to 1.15 makes the issue vanish. And it appears that
> _everyone_ I have on Oracle hits the issue when going up to 1.16.
>
Well, you can always diff the files... it is open source after all.
The obvious answer is yes, HUGE changes. It's documented in the changes
file. Oracle Unicode support was non-trivial to add. Things were
significantly reworked, and there were 7 (or was it 8?) release
candidates before 1.16 was officially released. The LOB code was more
problematic than the non-LOB code. And it may not have been as throughly
tested as some of the other code, although the long test is pretty
detailed. And it is working fine at my employer, where we have it
running in production. It works with both the 9.2 client to an 8.1.7
database, and from a 9.2 client to a 9.2 database.
Is is possible that some combinations of conditions that don't work were
missed in all the testing that was done? You bet. We found a bunch of
combinations that we need to except. mostly having to do with OLDER
oracle client software or OLDER versions of oracle on the database side.
The later is apparently not the problem for you, but I would think that
you could ask at least one of your customers for the information I asked
you for. BTW: Have them use YOUR database (in ORACLE_USERID) while
installing 1.16.
Since you say this breaks _every_ one of your customers, however, I find
myself wondering if you have something mis-configured, or at least
pretty unique in your database or environment. I assume your customers
told you DBD-Oracle-1.16 installed cleanly?
Lincoln
- Next message: Greg Sabino Mullane: "Re: DBIx::DBH - Perl extension for simplifying database connections"
- Previous message: Lincoln A. Baxter: "Re: Apparent DBD::Oracle 1.16 unicode regression"
- Maybe in reply to: Lincoln A. Baxter: "Re: Apparent DBD::Oracle 1.16 unicode regression"
- Next in thread: Tim Bunce: "Re: Apparent DBD::Oracle 1.16 unicode regression"
- Reply: Tim Bunce: "Re: Apparent DBD::Oracle 1.16 unicode regression"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
