Re: usage of ODBC

In article <2mf5fi$gp2_at_crl.crl.com> rjmac_at_crl.com (Robert J. McCallister) writes:
>From: rjmac_at_crl.com (Robert J. McCallister)
>Subject: Re: usage of ODBC
>Date: 19 Mar 1994 07:24:34 -0800
 

>Quinton Raj Yogeswaran Gregory (quintong_at_ifi.uio.no) wrote:
 

>:*Does anyone have any ide about to connect oracle which
>:*resides in unix maskin from pc by using ODBC.
>:*
>:*any comments will be appreciated.
 

>:*Thanks in advance.
 

>:* regards
>:* Quinton Gregory.

>I will assume you have installed your ODBC software properly and are
>having trouble connecting to your host. Go into your Control Panel and
>open the ODBC package. Select the Data Source(Driver) you set up to
>use for your connection to your host. Select the setup button and in
>the Network Address box enter the following.
 

> {net_prefix}:{hostname}:1024
 

> Select the help button in the ODBC Setup window for an
>explanation of what net_prefix and hostname mean. The 1024 is not
>documented in the help, this is the buffer size and the default is 4096,
>which if you allow the default to occur you will get a complaint from
>ODBC about the buffer size being mismatched or to large, I cannot
>remember the exact error message.
 

> We first set our host name to the name we have given the host but
>Oracle recommended we use the IP address. The network address we set was:
 

> T:{IP address}/1525:oracle:1024
 

>The 1525 after the IP address was added by the recommendation of Oracle,
>I am sorry I cannot remember where that is set in the configuration of
>SQL Net or on the host, but I rember the tech support from Oracle
>saying it is standard.

The 1525 after the IP address is a workaround to an ORACLE bug! Starting with V1.1.7.8 of SQL*NET for Windows, SQL*NET can no longer correctly read the services file. Unless the port entry for orasrv is the *FIRST* entry in the file, SQL*NET fails to find the port. I found this out late last year and, to the best of my knowledge, ORACLE has yet to fix this bug. As a result, we refuse to use this version of the product and insist that they supply us with V1.1.7.7b.

My suspicion is that you 1024 error message is also a bug that started in V1.1.7.8 since we do not always do this with V1.1.7.7b and do not generate an error message. THIS IS NOT TO SAY that using the 1024 is not the better way to do things. We have been struggling with a SQL*NET related network problem that is starting to look like using the 1024 minimizes if not eliminates it.

>Try this and if you have any more questions please feel free to
ask.

>Robert J. McCallister
Received on Tue Mar 22 1994 - 09:54:44 CET