Re: Weird listbox issue (Gupta(SqlWindows), Oracle)

On Dec 2, 10:08 am, "MVk - Monika V. kycka" <m.vaitk..._at_gmail.com> wrote:
> On 2 Gruo, 18:32, Shakespeare <what..._at_xs4all.nl> wrote:
>
>
>
> > Actually, I think it IS the best thing. Better than cluttering your code
> > with '""||() and so on... Views are for free, you don't have to pay for
> > them...
>
> > And instead of testing all kind of modifications to your code, I suggest
> > you try to find the difference in the databases. It can't be an Oracle
> > bug, for it does work in one of the databases.
>
> > I would perform these queries in SQL Plus and check for the differences
> > in the result set (null values, lengths, number of records returned)
>
> > Shakespeare
>
> It's not the matter of price. For the fast solution, we created some
> views, but they are simply not the best sollution. Our software is
> huge and creating dozens of views just to represent simple cumulated
> select data (e.g. for a filter listbox) is not practical and time
> consuming, especialy for places with already existing code. Writing
> queries in Pl/Sql Developer (or SQL Plus - doesn't matter which one)
> is no help here because I can't recreate this situation - data is
> selected well in both databases.
>
> Comparing db parameters is the only step I see for further checking.
> But taking in mind that there are more than 250 db parameters (of
> course, not each is of the matter or may differ), but lets say I'll
> find 30 parameters that differ and looks potentialy important. How to
> know which one(s) is(are) faulting data capture? I hoped someone would
> have any idea or give a direction what to look for/base on - charsets,
> nls settings, cursor sharing, oracle version difference in dbs, server
> setting differences (if different servers), etc? Besides, as I have
> mentioned, problem occurres on the real database with real data where
> we have limited access and have to make requests to the client for
> exact information we need. Well, the burocracy... Ok, I'll try getting
> all the parameters from both dbs tomorow, see what differs and write
> news here. Hope, then there would be some ideas of what could be
> wrong.
>
> Thank's for cooperating. :)
>
> MVk - Monika V. kycka
>
> ps. Ah.. And I almost forgot! Data is the same in both databases as
> there was dump created from real db and imported to test db.
> ps2. I'm just a programmer and db or server administration is not my
> field. That's why I would be greatfull for any help. :]

Don't know a thing about Gupta, but I can guess that if you are using substrb on NAME and have different results with data that was exported and imported, you may have a characterset issue. So you would need to post:

The character set of the prod db.
The character set of the test db.
The NLS values of the database.
The NLS values of any other tier.
The NLS values of any session involved, including the export/import. The exact (to all findable decimal places - what does sqlplus say when you start it?) of all parts of this, including database, operating system, and apps. Could be a Gupta bug or missing synonym or something like that, too.

I'm guessing you will find some difference there, could be as fundamental as the db character sets. exp/imp (oh yeah, there are different tools with similar names, be sure and specify which you used) will oh-so-helpfully translate character values for you, as can NLS settings dynamically. If you have middle tier startups for test and prod, look for differences in the environment.

Aside from that, I doubt there would be init.ora differences to make this kind of effect.

jg

--
@home.com is bogus.
"Cheech & Chong celebrity roast brought to you by Windows" - yeah, Up
In Smoke for sure.