10.2B01

Posted by Admin on 03-May-2010 00:09

I've just seen service pack 1 for 10.2B on my ESD (added last Friday).

Anybody tried it already? Any comments, warnings, recomendations?

All Replies

Posted by jmls on 03-May-2010 01:01

We installed it this weekend. So far so good. Will let you know later

today, after live rollout

Didn't see the announcement on any email though ...

Julian

Posted by Admin on 03-May-2010 01:16

Didn't see the announcement on any email though ...

Me neither.

Posted by bsgruenba on 03-May-2010 10:17

I installed it this weekend on Windows and Linux. I have not seen any problems yet, but then no gains either.

I'm particularly interested in banging on one of the bug fixes related to Prodatasets and Java OpenClient. I plan on getting to that later this week.

Posted by Tim Kuehn on 04-May-2010 08:38

10.2B01 has a fix for a rather nasty bug...

ID: P164132
Title: "Different number of results returned with   USE-INDEX verses no index specification"
Created: 04/19/2010
Last Modified: 05/03/2010
Status: Unverified


Symptoms: 
·    Different number of results returned with USE-INDEX verses no   index specification
·    Different number of results returned with USE-INDEX verses   natural index selection
·  Query predicate   that selects two single component indices for query shows fewer records than   exist to meet the criteria.
·    Displaying the recids for the missing records shows that they are beyond the   32 bit boundaries.  (These are 64 bit recids).


Facts: 
·  Table   contains data for both 32 and 64 bit recids.
·    Changing code to make use of the USE-INDEX function instead of the natural   index selection locates all records for the given table.
·  proutil   dbname -C idxcheck has no errors.
·  proutil   dbname -C idxfix has no errors.
·  proutil   dbname -C idxbuild has no errors.
·  proutil   dbname -C idxbuild does not address the problem.
·  dbtool   dbname (option 3, 4 or 5) does not detect the problem.
·    Database was dumped and loaded via binary dump and load with
·  All   Supported Operating Systems
·    OpenEdge 10.2A 64-bit
·    OpenEdge 10.1C 64-bit
·    OpenEdge 10.1B 64-bit
·    OpenEdge 10.2A03 64-bit Service Pack


Changes: 
Upgraded application and schema from 10.1B03 to 10.2A03.


Cause: 
·  Bug# OE00197068
·  When   retrieving data from the database and relying on natural index selection,   based on predicate formation, a significant number of records that match the   predicate criteria are not returned from the database. When the query uses   multiple indexes and the result-set includes records with 32-bit and 64-bit   rowids, the records with 64-bit rowids are not returned to the client.


Fixes: 
Upgrade to 10.2B01.

A workaround for this issue is to modify the code to make use of the   USE-INDEX statement.

Posted by Admin on 12-May-2010 03:49

I'm particularly interested in banging on one of the bug fixes related
to Prodatasets and Java OpenClient. I plan on getting to that later this
week.

If this is the performance problem when retrieving datasets from a Java OpenClient, there's a noticeable gain. I didn't go really deep to see if there are side effects, I'm planning on doing that later.

Gilles

This thread is closed