Transation Isolation Levels

Discussing with a colegue about defining transactions in our application we finished by talking about transaction isolation levels and finally I got into problems explaining the problems faced in concurrent transactions (I wasn't able to tell the difference between non-repeatable and phantom reads [sigh/]). I decided to write down for further references. Transaction isolation levels are defined according to the state of data during the following 3 scenarios (they compair the data accessed from 2 different transactions): * dirty reads occur when a transaction reads data written by concurrent uncommited transaction; * non-repeatable reads occur when a transaction re-reads data it has previously read and finds that data has been modified by another transaction (that committed since the initial read); * phantom read occur when a transaction re-executes a query returning a set of rows that satisfy a search condition and finds that the set of rows satisfying the condition has changed due to another recently-committed transaction; Note: even if it seems that non-repeatable reads and phantom reads are the same thing, the difference is made from the fact that in the 1st case the same rows are returned but their data is modified, while in the 2nd case another set of rows is returned. Defined isolation levels:

DB Isolation Level	Dirty Read	Non-Repeatable Reads	Phantom read
Read uncommitted	Y	Y	Y
Read committed	N	Y	Y
Repeatable read	N	N	Y
Serializable	N	N	N

Java Level	Dirty Read	Non-Repeatable Reads	Phantom read
TRANSACTION_NONE	Y	Y	Y
TRANSACTION_READ_UNCOMMITTED	Y	Y	Y
TRANSACTION_READ_COMMITED	N	Y	Y
TRANSACTION_REPEATABLE_READ	N	N	Y
TRANSACTION_SERIALIZABLE	N	N	N

Still I would like to see more details on the separation made on non-repeatable reads and phantom reads.

Best tip in town

This is the best tip in town by far and I got a quick answer too. Raible Designs ~ Eclipse tip o' the day Good job! 10x a lot

Are you gonna test you java GUI?

If yes pay a visit here: TRIJUG Wiki: TestingGUIs. A non exhaustive list, but you may find what you are looking for.

Becoming an Architect

Becoming an Architect Luke Hohmann: Most people think of architecture as technical, and the architect as a technical person. And absolutely, the architect must be technical. But there's also a social aspect of the architect role that I think is not well communicated or understood. The architect is the person who will say, 'This is the way we do things.' [...] Conversely, you can tell when a system hasn't had an architect, because it lacks consistency in things like error handling, naming, and so on. In a system I'm working on right now, for example, I noticed that some of the database tables have a unique identifier column named id, other tables have a unique identifier column named id, and other tables do something completely different. So table foo would have a fooid column, and table bar would have an id column. I finally went to the development team and said, "Let me guess. You've had more than one DBA [Database Administrator] in your time here." And they said, "Oh yeah, we've had five of them." Every DBA had their own way of naming columns. They didn't have an architect to ensure a technical consistency in column naming. There wasn't a notion of conceptual integrity from a technical perspective. [...] Humans are failure machines. We're not success machines. We're failure machines. We fail all the time. And it's only through processing the feedback of our failure that we learn how to correct for them and do better. That is why it is important to stick with the choices you make and understand how well they worked.

LGPL your project on...

Assuming you are intending on using Java and a license like LGPL then here are some choices.

• SourceForge
• Savannah (GNU)
• java.net
• Codehaus - if you are interesting.
• Tigris - if you fit their mission.
• Apache Incubator

This list was created by some other blogger to who I must thank. Unfortunately I have lost the original link so I excuse myself in front of him [sorry/].

freshmeat harvest - part 1

This is the first part of a long harvest time on freshmeat announced projects: * Linux

• X File Explorer
• Filer: small file-manager
• Simple Samba Commander
• Linux Feed Reader
• Xi-Batch: job scheduling and workload management system
• ConsultComm Project Timekeeper: lightweight, cross-platform program that allows anyone managing multiple projects, clients, or tasks to effectively keep track of exactly how much time has been spent on each project
• deplate: converting documents written in an unobtrusive, wiki-like markup to LaTeX, HTML, or "HTML slides"

* Java

• UUID: UUID implementation in Java
• Maxine: simplistic JMX microkernel
• OpenWFE: workflow engine
• applet2app: converts applet to standalone
• UMLGraph: Declarative drawing of UML diagrams
• Blitz-Javaspaces: JavaSpaces technology

* Misc

• Jabberwocky: IDE for LISP
• DidiWiki: small and simple WikiWikiWeb implementation written in C
• csvToSql
• Vex - A Visual Editor for XML
• CVSGraph

Tomcat - JRockit

Found this on this weblog: Random thoughts:

Turns out that it wasn't Tomcat 5 that provided the 15-20ms/page performance. Instead, it was the JRockit 8 JVM that did the trick. I got the same performance with Tomcat 4, but switching to Sun JVM /w server settings increased the rendering time to about 40ms/page. Oh well.

I definitely must try this.

Manageability quoting Rod Johnson

Manageability - Rod Johnson on J2EE Without EJB: An interesting listing of EJB offerings and their utility:

• Declarative Transactions: Good but not always appropiate.
• Remoting: Best choice today for remoting.
• Clustering:Entity and SFSB have weak implementations.SLSB clustering isn't rocket science. Clustering more important on the web and database tier.
• Thread Management:Locking entire bean instances is naive. Multi-threaded services without read/write instance variables make better sense (i.e. Servlets).
• Instance Pooling:Doesn't make sense with today's more modern garbage collectors.
• Resource Management:Not truly tied to the EJB container.
• Security:Security requirements tend to be more complex than what EJB role-based security provides.
• Business Object Management Use of JNDI unnecessarily complex.

Understanding Proxy ...

Try to think, try to do...... Nice blog on Proxy. If you feel Proxy is something hidden just take the code and understand it. Use: byte[] proxyClassFile = ProxyGenerator.generateProxyClass(proxyName, interfaces);

NoCo first version

A short update on the previous post. I have already finished the first version of the CoNo plugin extension. I have name it NoCo :-). I intend to send it to the initial authors - in case they are still interested in it (they are not active for the last 2 years, but who knows). Otherwise I will try to find a solution to post it somewhere. I will come back with the features description of the current version.

Holiday plans

I've been in holiday for one week and a lot of plans have risen into my mind. I will present here shortly:

• Eclipse CoNo plugin extension: to save notes files to a specified (per project) directory - this allows not to save the notes into cvs or to have to add lots of .cvsignore files
• Eclipse Protocol plugin extension with javadoc tags: this will allow the same functionality as the original plugin, but will not force you to reformat the sources in order to take advantage of it
• Aspectwerkz code digging
• Aspectwerkz ajdoc implementation
• Spring framework fullblown demo

I also intend to start and finish reading Hibernate in Action as I consider it one of the most interesting frameworks (and not just interesting, but very very usefull). I think I have lots of plans for only one week left for my holiday.

tough refactoring

Long time no see. It was a hard month. I hope that the days to come to be more normal. Oke, back to business. I got on my head an applet client with about 150 classes. Told that it worked. Although at the beginning I didn't agree, now I'm very sure, and also, I'm not very happy about it. Eveything seems to me to be very messy (code duplication, a lot of unusefull delegation and lots of classes unused). I started by attacking the communication layer, which I thought was very important to my application. I've tried a refactoring and little by little the code is 70% rewritten (this is not very good, but now the communication is pretty neat). Other interesting part is some panels that are generated code and displayed according to user interactions. The generated code contained also lots of code duplication and unused methods and fields. I've fixed this too, so the code in a generated panel has been reduced by about 40%. I have only added 4 generic classes that add the default behavior. I am left with aproximatively 60% of this client, which is represented by specific GUI controls developed by extending the Swing. Huston, we've got problems. Number of methods/class: 11. Not so bad. But looking deeply, the most used controls have between 15 and 47 methods, and this is definitely not good. But the thing that bothers me the most is the duplicated code. I hate this code. You find yourself on the last days and find out a little bug. You fix it, but you must replicate that code in other x classe. Then you find out you just introduced another one, and your work must be redone. This way you can loose the delivery time. Keeping this in mind, I want to start eliminating the duplicated code. First idea: create a few generic controls with the attached default behavior and then let the others extend. But this solution is no good now, as all the developed controls extend a Swing control. I must find out something else. Secondly, I want to reduce interfaces offered by these controls. 47 public methods represents to much. For the moment I guess not only 15 of these are really used. Then why to publish such an interface... just for the future sake??? I am really trapped with this problem. As always, I don't have enough time to try to create some working skeleton and then only append the needed features. I must deal with the whole existing stuff. What I propose to myself is: - try to extract the common code and then find a solution for this code to be called from each already developed control; - step through all the controls and change the published interface and so finding out what is the really needed interface. After this, reduce the quantity of code. Maybe this step will be the first one, so the extraction step to be easier to be done. This is what I'm thinking this evening (in fact late night) after studying the code for 2 days. Tomorrow is a new day, and I think more ideas will be in my head. Till next time, ta-daaa.

AOP intro

A few introductory links on AOP.

• What is AOP? An introduction by Renaud Pawlak and the JAC development team.
• Aspect-Oriented Programming and JBoss by Bill Burke and Adrian Brock -- Is aspect-oriented programming (AOP) the OOP of the oughts? It promises to apply common behavior across different object models. Bill Burke and Adrian Brock explore AOP by example in a JBoss application.
• Aspect-Oriented Refactoring Series - Part 1, by Ramnivas Laddad, author of AspectJ in Action
• Aspect-Oriented Refactoring Series - Part 2, by Ramnivas Laddad

I got two essential books on my library:

• Mastering AspectJ
• AspectJ in Action

debut [old post of mine :-)]

New times may require sometimes new actions. I am changing again my employer. Why? I think I did this for the future. I think I did this hoping to be different. 'Being different' suggested me to give a try also to blogging. What will be here? I'm not very sure yet, but I think it will be something with the smell of personal-professional thoughts. First of all, I work in the software industry, java software industry. Every day trying to be keep uptodate with the tons of new information. Every day finding/discovering something new. Every day trying to deal with time planning, trying to get the most from this day. How can you/must you organize your time to be able to successfully fulfill all the requirements and needs. This is what I think software industry is all about. Let's imagine the following scenario on a daily basis: * get your tasks done * read the news (your prof-life depends on them). Try some blogs. * check some specs/docs to keep the brain cells uptodate * scan the mails coming from the main mail lists you are in * answer some forum questions to verify that your brain cells are still working * go home, enjoy your family * and last but not least ... relax. A new day is coming. :)