Disqus - Latest Comments for mochoa

Re: Oracle Web Services Manager book

mochoa — Tue, 12 Apr 2011 08:18:48 -0000

Fixed.

Re: Revisiting JDBC Connection Pooling and Spring Data Source implementation

mochoa — Mon, 28 Mar 2011 06:38:38 -0000

Fixed. Best regards, Marcelo.

Re: Dealing with JDK1.5 libraries on Oracle 10g

mochoa — Mon, 03 Jan 2011 06:02:21 -0000

Hi:
Any library or code compiled with JDK 1.4 will run without any problem on 11g.
Mostly of libraries available today are compiled with JDK 1.4 and runs perfect.
Marcelo.

Re: New Oracle Lucene Domain Index release based on Lucene 3.0.2

mochoa — Mon, 01 Nov 2010 19:42:55 -0000

Hanifnejad:
If you are using latest LDI (3.0.2) simply connect as SYSDBA and grant this role:
SQL> grant LUCENEUSER to scott;
Previous versions do not have this role, simply use:
SQL> begin
dbms_java.grant_permission('SCOTT','SYS:java.util.logging.LoggingPermission', 'control', '' );
commit;
end;
/

and thats all. Best regards, Marcelo.

Re: New Oracle Lucene Domain Index release based on Lucene 3.0.2

mochoa — Tue, 21 Sep 2010 17:53:06 -0000

Hi:
If you change to thin driver the parameter db.str should be:
db.str=orcl4.oracle.local:1521:orcl4
which means using JDBC thin syntaxt, but I am not sure if this connection string will works with SQL*Plus syntax.
Best regards, Marcelo

Re: New Oracle Lucene Domain Index release based on Lucene 3.0.2

mochoa — Mon, 20 Sep 2010 12:28:06 -0000

Hi:
This seem to be a problem of the value of you LD_LIBRARY_PATH configured in your environment.
loadjava.xml Ant scripts is using JDBC OCI connections, so the LD_LIBRARY_PATH should be defined to the directories where the 64bits .so files are installed.
Please check this configuration, another alterntative is to modificate loadjava.xml Ant task to use -thin flag and change the connection string value, but It will change the behaviour of other task that uses SQLPlus, for example.
Best regards, Marcelo.

Re: Auto complete functionality with latest Lucene Domain Index

mochoa — Sat, 05 Jun 2010 08:05:27 -0000

Jose:
En mi perfil del programa ACE aparezco como en la parte de "Database Management & Performance" pero en realidad sí, tengo mas un perfil de desarrollo del lado del servidor que en la BD propiamente dicha.
Mi CV on-line puede clarificar un poco mas la idea: http://mochoa.sites.exa.uni...
Slds. Marcelo.

Re: New release of Lucene Domain Index based on Lucene 2.9.2

mochoa — Wed, 31 Mar 2010 07:45:39 -0000

Hi:
I don't have a collection of XMLType table with above number of documents to test, I have made test with WikiPedia Spanish/English dump stored as XMLType in object-relation way and it works properly in mi noteook (http://marceloochoa.blogspo....
Sure Lucene scale well for this number of documents.
For the implementation of INPATH or WITHIN Oracle Text operator it is necessary to override or extend the class DefaultUserDataStore which is a default implementation and for the XMLType it only performs text extraction discarding node information.
With WikiPedia due only have three important nodes the trick is to use directly XMLTYpe XPath extraction, for example:
create index pages_lidx_all on pages p (value(p))
indextype is Lucene.LuceneIndex
parameters('BatchCount:115;MergeFactor:5000;ParallelDegree:2;PopulateIndex:true;DefaultColumn:text;SyncMode:OnLine;LogLevel:INFO;PerFieldAnalyzer:title(org.apache.lucene.analysis.EnglishWikipediaAnalyzer),comment(org.apache.lucene.analysis.EnglishWikipediaAnalyzer),text(org.apache.lucene.analysis.EnglishWikipediaAnalyzer);ExtraCols:extractValue(object_value,''/page/title'') "title",extractValue(object_value,''/page/revision/comment'') "comment",extract(object_value,''/page/revision/text/text()'').getClobVal() "text",extractValue(object_value,''/page/revision/timestamp'') "revisionDate";FormatCols:revisionDate(day);IncludeMasterColumn:false;LobStorageParameters:DISABLE STORAGE IN ROW CHUNK 8192 CACHE FILESYSTEM_LIKE_LOGGING COMPRESS');

Note that I am extracting text, comment, title and timestamp XML elements in the index.
For your number of XML Elements included in the XMLType object a typical implementation is to convert nodes to Lucene Field names like:
< A_Node >
< B_Node >text1
< C_Node >text2

transformed into:
A_Node-B_Node: text1
A_Node-C_Node: text2
with this Lucene Field names you can use lcontains('A_Node-B_Node:qry_text1 OR A_Node-C_Node:qry_text2') for example.
Best regards, Marcelo.

PD: If you need further assistance to implement a custom DataStore Class just drop me an email to marcelo.ochoa <at> gmail

Re: New release of Lucene Domain Index based on Lucene 2.9.2

mochoa — Tue, 30 Mar 2010 16:22:53 -0000

Hi:
I can't figure out which is your question.
You mean if its possible to index 500M XML documents stored in XMLDB?
The answer is it depends on the size of the XML, the structure, your hardware, etc. etc.
Best regards, Marcelo.

Re: New release of Lucene Domain Index based on Lucene 2.9.1

mochoa — Tue, 09 Mar 2010 16:54:08 -0000

Roberto:
ant jit-lucene-classes
Do Java-to-assembler compilation in advance.
Oracle 11g JIT technology evaluates every method call to update most-usage classes, then a background process do Java-to-assembler translation once it detects a most used method.
If you don't want that the database detects these most used methods, this target compiled all LDI classes in advance to get faster response time on LDI indexes and operators.
Best regards, Marcelo.

Re: New release of Lucene Domain Index based on Lucene 2.9.1

mochoa — Tue, 09 Mar 2010 15:07:37 -0000

Hi Roberto:
A this point the first step (Lucene schema creation) was done. Please check if there where errors.
Or simple try to login with SQLPlus as:
sqlplus LUCENE/LUCENE@<my_sid>
To see if the schema for LDI was properly created.
Best regards, Marcelo.

Re: New release of Lucene Domain Index based on Lucene 2.9.1

mochoa — Tue, 09 Mar 2010 14:33:32 -0000

Hi:
Did you put your build.properties file at $HOME directory?
The correct output from Ant script is:
arguments: '-nodefiner' '-v' '-s' '-g' 'public' '-u' 'LUCENE/***@test' '../../build/contrib/ojvm/lucene-ojvm-2.9.2.jar'
note that db.usr should be LUCENE and not '-u' '${db.usr}/***${db.pwd}@${db.str}'
If build.properties file is not at $HOME directory (Unix) or c:\Documents and Setting\user on windows it will not work.
Best regards, Marcelo.

Re: Experiences on Java programming within the databases, tips & tricks, tools, open source libraries and more

mochoa — Mon, 14 Dec 2009 18:46:19 -0000

Kuassi:
There wasn't material :(
This is the unconference way ;)
We took a look at the code and I have done comments them on the fly :)
Marcelo.

Re: Oracle Web Services Manager book

mochoa — Wed, 21 Oct 2009 07:35:28 -0000

Fixed...
Best regards, Marcelo.

Re: Unleash Oracle ODCI API - OOW09 Voting Session

mochoa — Thu, 18 Jun 2009 12:54:41 -0000

Right, but you can contribute spreading the post ;)

Re: New binary release of Lucene Oracle Data Cartridge using Lucene 2.4.0

mochoa — Sun, 08 Feb 2009 16:20:18 -0000

marcelo.ochoa@gmail.com
Regards, Marcelo.

Re: New binary release of Lucene Oracle Data Cartridge using Lucene 2.4.0

mochoa — Sun, 08 Feb 2009 07:55:10 -0000

Hi David:
Latest release was tested using a VMWare virtual machine with:
- Mandriva 2009 (basically a clone of RH5) kernel 2.6.24.4
- Oracle 10g 10.2.0.3.0
- Diskspace
/dev/sda1 5,6G 5,0G 300M 95% /
/dev/sdb1 2,0G 1,3G 679M 65% /u01/app/oracle/oradata
/dev/sdc1 1004M 484M 470M 51% /home
- Meminfo
MemTotal: 1555672 kB
SwapTotal: 417648 kB
- GNU C Library stable release version 2.7
If you want I can send you a complete Oracle RDA output file to compare the configurations, just drop me an email to my personal account.
Best regards, Marcelo.

Re: New binary release of Lucene Oracle Data Cartridge using Lucene 2.4.0

mochoa — Sun, 25 Jan 2009 15:47:20 -0000

Hi David:
I just uploaded two new binary distribution of 2.4.0 version.
https://sourceforge.net/pro...
With these distributions you don't need to build lucene sources, all the required libraries are compiled at lib directory.
On the other hand NCOMP consume a lot of disk space and memory to perform the compilation.
I tested this binary distribution with a Linux VM using 1.5Gb ram and more than 400Mb at /tmp and 600Mb at $ORACLE_HOME free space.
Best regards, Marcelo.

PD: I added alpha to the binary dist because I don't have phighlight pipeline function completed yet, but the other code is the production code.
Also I am starting a short vacation and I will not have internet connection for the next two weeks :(

Re: New binary release of Lucene Oracle Data Cartridge using Lucene 2.4.0

mochoa — Sun, 25 Jan 2009 13:07:45 -0000

Hi David:
This is strange, I made lucene-ojvm-2.4-10g-bin compiled into an Oracle 10g using the JDK placed into ORACLE_HOME/jdk directory.
Are you sure that you are using this download:
http://downloads.sourceforg...
Best regards, Marcelo.

PD: org_apache_lucene_util_Installer is not used by Lucene Domain Index, but if you have more classes with dependencies to jdk1.5 may be is bad compilation.

Re: Using APEX into Oracle WebLogic Server 10.3

mochoa — Thu, 15 Jan 2009 06:08:59 -0000

Sure, see this post:
http://ora-00001.blogspot.c...

Re: New binary release of Lucene Oracle Data Cartridge using Lucene 2.4.0

mochoa — Thu, 15 Jan 2009 06:06:22 -0000

Fred:
Lucene Domain Index 2.4.0 was releases the same day as Lucene 2.4.0 binary release :)
Usually I download latest CVS code a month before the new planned release, test and update changes affected by the new API and release it
Also I follow user and dev Lucene lists to see the changes and register new functionality to integrate with the code.
Now, I am working in a maintenance release of 2.4.0 which adds Highlighting functionality using an ancillary operator lhighlight() and removes the usage of some planned deprecated API such as Hits.
Performance is pretty good, I have done a test over Wikipedia Dump parsing/highlighting 79Mb of XMLType text and returning 756K of code snipped in 60 second with my notebook.
Due the ancillary operator have a limitation of working in one column at time, I am working in a pipeline table function which accepts ref_cursor parameter with some other key and returns a new ref-cursor with the column highlighted.
Stay tuned.
Best regards, Marcelo.

Re: Using APEX into Oracle WebLogic Server 10.3

mochoa — Mon, 01 Dec 2008 05:09:51 -0000

Hi Sebastian:
I think that Apex will work with your 10.2.0.1 DB version, but installing latest patchset is always good.
First check at Apex readme if there is some bug related with 10.2.0.1 DB and the OWA PLSQL packages included on it.
DBPrism mod_plsql emulation is aware of the Embedded PL/SQL gateway functionality and it has been working since Oracle 7x version.
It only uses owa* packages and the packages of Apex in that case.
Best reagards, Marcelo.

Re: XMLDB and Lucene Domain Index

mochoa — Sat, 11 Oct 2008 17:10:33 -0000

Hi Sanjeev:
There is a new DBPrism CMS release ready for downloading:
http://www.dbprism.com.ar/l...
This new release is full tested on Linux but due is installer works with Apache Ant sure there is no problem to execute on Windows.
First be sure that you have workwing a Java SE envirement and Apache Ant.
Latest release is for Oracle 11g, but there I can provide you a simple patch to run with 10g standard edition one+
Best regards, Marcelo.