We are in the process of migrating all our applications.
We are changing :

- the server OS from Novell NetWare 3.11 and 3.12 to Microsoft NT4,
- the client environment from Microsoft Windows 3.11 to Microsoft NT4,
- Sybase SQLServer from v 4.2.2 and to Sybase Adaptive Server
- Sybase Open Client from DBLIB 4.2.5 to Open Client 11.1.1
- application from Powerbuilder 4 to Powerbuilder V5
- from DB_LIB to CT_LIB (when possible).

We are based in France and there are lots of accentuated characters in
the French language. Almost all the applications we are used are Windows

Today's situation

I have had some problems with charsets and at the moment although all
servers are started with cp850 as the default charset the database
usually contains iso_1 data.

I make the following assumption, please correct me if I am wrong; if I
edit a file using EDIT (dos program) and all looks OK then this is cp850
and if I edit it using NOTEPAD (windows programs) and all looks OK then
it is iso_1.

In today's situation, on a system 10 SQLServer :

- if I BCP out with –J cp850, I get a file that looks like iso_1,
- if I BCP out with –J iso_1, I get a file that does not look like iso_1

nor CP_850

This is something that I would like to change.

This is what I want to do

I have 70 servers to install and I would prefer to get things right from
the start.
Since most of our applications are Windows Applications and since the
charset for Windows is iso_1, I was thinking of starting the server with
iso_1 as the default charset to avoid any unnecessary conversions.

This is how (I think) the applications react.

With a CT_LIB 32 bit application (10.0.4 or 11.1.1), after looking up
the info in LOCALES.DAT , Open Client determines that the charset should
be iso_1. No translation occurs.

With a CT_LIB 16 bit application (10.0.4), after looking up the info in
LOCALES.DAT , Open Client determines also that the charset should be
iso_1. No translation occurs.

I have been using Embarcadero's products DBArtisan and RapidSQL to
administer the servers and to develop SQL code. The 16-bit version of
the programs is a DB-LIB application and the new 32-bit version is also
a DB-LIB application.

With the 16 bit version (10.0.4), open client decides that the charset
used by application is iso_1 (unless I modify the default values in the
W3DBLIB.MSG file), no translation occurs and data is displayed

With the 32-bit version (10.0.4 or 11.1.1), open client decides that the
charset used by the application is cp850, the server therefore converts
the data from iso_1 to cp850 and when displayed by Windows Application
does not look too good.

This is the problem.

The way things are in a 32-bit environment, if I input some data in
using RapidSQL or DBArtisan (DB-LIB) it does not look too good if I
access with a CT_LIB application and vice versa.

What I would like to be able to modify is the default value used by the
DB-LIB 32-bit applications.

Has anyone else encountered the problem and is there a known remedy ?


Jean Pierre CALLANT


email :