I was wondering why my build was always broken and corrupt and I found out that in my SVN repository the eclipse_1206.so file was missing…and indeed it was NOT committed.
I first thought I made a mistake, so I removed and recommitted everything. And again, the eclipse_1206.so wasn’t there.

So I started googling and I found this which directed me to this. Apparently *.so files are igored by SVN by default even though there is no svn:ignore property set.

Solution:
In order to solve this I did the following:

1. Open the file ~/.subversion/config
2. Remove the comment from the line which starts with # global-ignores
3. The line looks like this now: global-ignores = *.o *.lo *.la #*# .*.rej *.rej .*~ *~ .#* .DS_Store

As you can see the .so file extension is not defined anymore and suddenly .so files also show up in my SVN Commit Dialog (I am using svnX on MaxOSX).

This also explains a bit my most recent issue I was having with a broken Eclipse build.

I am not too sure if this is a good default by SVN to ignore .so files as those are shared objects in Unix Environments which are important for Multi Platform builds in my case. Anyway I found this out and solved the problem.

My problem until today was, that this kind of transaction handling was not working in my DWRServlet, because this is a servlet not under my control. My first approach was to create a new servlet which extens DWRServlet so that I could modify the behaviour , but that didn’t work completely, because DWR does not pass Exceptions to my servlet. DWR handles all exceptions, marshals them so that the Exception can be passed back to the browser to be displayed in your calling Javascript code.  That means I cannot catch the exception in my custom servlet  because it is already caught and handled. Damn!

So I debugged and digged through a complete DWR call and found the @GlobalFilter annotation. That seems to be exactly what I need. This annotation can be applied to a class and this class also needs to implement the AjaxFilter interface. DWR then applies all filters to each method call made through the DWRServlet. This class needs to be registered the same way as you register your other DWR annotated classes (e.g. the classes annotated with the @RemoteProxy or @RemoteMethod annotation)

So I came up with the following class which borrowed some code of the existing org.directwebremoting.hibernate.H3SessionAjaxFilter which does almost exactly what I need, except the rollback() of my transaction in case of an error.

@GlobalFilter
public class DWRFilter implements AjaxFilter {

	private static final Logger LOGGER = LoggerFactory.getLogger(DWRFilter.class.getName());

	public Object doFilter(Object obj, Method method, Object[] params,
			AjaxFilterChain chain) throws Exception {

        SessionFactory sessionFactory = Activator.getSessionFactory();

        Transaction transaction = null;
        if (sessionFactory != null)
        {
            Session session = sessionFactory.getCurrentSession();
            transaction = session.beginTransaction();
        }
        else
        {
            LOGGER.error("SessionFactory not initialized for this web application.");
        }

        Object reply = null;
		try {
			reply = chain.doFilter(obj, method, params);

			if (transaction != null)
			{
			    transaction.commit();
			}
		} catch (Exception e) {
			LOGGER.error("Error while committing transaction in DWRFilter. Transaction will be rolled back now...",e);
                        if(transaction != null){
			    transaction.rollback();
                        }

			throw new InvocationTargetException(e);

		}

        return reply;
	}

}

This basically does exactly what I want and is basically something like the Open Session In View pattern for DWR and Hibernate. For each DWR call it is the following flow:

• begin transaction
• do work
• commit transaction
• or rollback transaction in case of an exception

This way I have a central place for my DWR-Transaction handling.

This is my first solution to this problem and maybe it doesn’t catch every scenario, e.g. especially when you are using special features of DWR like batch calls etc. I haven’t used those features and for my current features using simple calls it works very well.

public static void main(String[] args) {

OutputStream out = new ByteArrayOutputStream();

System.setOut(new PrintStream(out));

System.out.println(“Test”);

String sysout_content = out.toString();

System.out.println();

}

After having spent almost an entire day to help our new developer to get setup his development environment containing the source code, eclipse, svn, appserver, database, user accounts I made a new resolution for every system I will build in the future:

The system about to be built:

• has to have a development environment which can be setup within less than an hour from a single location / disc, automatically
• to do this it has to use sensible defaults –> customization can happen later on a case by case basis
• the developer has to start working with it right away
• build and deployment on local environment has to happen automatically. required appservers etc. have to be part of the stuff installed ealier

I started developing with Ruby On Rails about 3 month ago and launched my first project a few days ago, I realized that the ROR already have taken those things into account. Everything is bundled inside the project folder and after checking it out from the SVN you are ready to go by just typing “script/server” in the command line. Same for setup “rails myProject”

I just started with Rails out of curiosity but I already love the way I can develop my ideas and web-projects. Besides that I love to develop in JAVA and currently in the field of OSGI on the server-side and there we basically need the ease of use and the spirit of Rails.

Ok, I switched a bit between topics, so I try to conclude my new years resolution:

“I want to build systems which solve complex tasks, and are easy to setup, fun to develop and fun to use.”

Komponenten:

• Macbook 13″ 2.16 GHz Intel Core 2 Duo (gekauft Oktober 2007),2 GB RAM
• 24″ TFT Samsung SyncMaster 245B
• Schreibtischlampe
• DENON Verstaerker als Stereoanlage

Durchschnittskosten pro 24h: 0,56 EUR (normale Freizeitarbeit, kein Freelancing)
Durchschnittskosten pro Jahr:  ca. 205 EUR

Minimal / Maximal:

Mit Minimal meine ich, dass der die Helligkeitseinstellungen beim Display bzw. die Lautstaerke der Stereoanlage minimal war. Maximal heisst beim Display volle Helligkeit und Lautstaerke etwas ueber Zimmerlautstaerke.

Fazit

Interessant was die Helligkeit allein ausmacht. Auch wenn alle Geraete ausgeschaltet sind, werden noch ca. 5 Watt gezogen. Das gilt auch fuer den Sleep-Modus des Macbooks, wo immer noch 9 Watt messbar waren, auch wenn alle anderen Geraet ausgeschaltet waren.

Aus diesem Grund schalte ich ab jetzt immer die komplette Steckdosenleiste ab, wenn ich das Haus verlasse.

I was running into exceptions like this:

Caused by: org.hibernate.exception.SQLGrammarException: could not load an entity: [app.model.entities.Jobs#8]
at org.hibernate.exception.SQLStateConverter.convert(SQLStateConverter.java:90)
at org.hibernate.exception.JDBCExceptionHelper.convert(JDBCExceptionHelper.java:66)
at org.hibernate.loader.Loader.loadEntity(Loader.java:1895)
at org.hibernate.loader.entity.AbstractEntityLoader.load(AbstractEntityLoader.java:71)
at org.hibernate.loader.entity.AbstractEntityLoader.load(AbstractEntityLoader.java:65)
at org.hibernate.persister.entity.AbstractEntityPersister.load(AbstractEntityPersister.java:3072)
at org.hibernate.event.def.DefaultLoadEventListener.loadFromDatasource(DefaultLoadEventListener.java:434)
at org.hibernate.event.def.DefaultLoadEventListener.doLoad(DefaultLoadEventListener.java:415)
at org.hibernate.event.def.DefaultLoadEventListener.load(DefaultLoadEventListener.java:165)
at org.hibernate.event.def.DefaultLoadEventListener.proxyOrLoad(DefaultLoadEventListener.java:207)
at org.hibernate.event.def.DefaultLoadEventListener.onLoad(DefaultLoadEventListener.java:126)
at org.hibernate.impl.SessionImpl.fireLoad(SessionImpl.java:905)
at org.hibernate.impl.SessionImpl.load(SessionImpl.java:822)
at org.hibernate.impl.SessionImpl.load(SessionImpl.java:815)
at org.springframework.orm.hibernate3.HibernateTemplate$3.doInHibernate(HibernateTemplate.java:569)
at org.springframework.orm.hibernate3.HibernateTemplate.doExecute(HibernateTemplate.java:419)
… 30 more
Caused by: com.mysql.jdbc.exceptions.MySQLSyntaxErrorException: Table ‘database.Files’ doesn’t exist

Although all my table names are lowercase it seems that Hibernate tries to create case-sensitive SQL queries which relate to the name of my POJOs (e.g. MyPojo) .

While this was working on my OSX it didn’t on Fedora Linux on EC2.

Solution:

In my /etc/my.cnf I had to add the following configuration property:

lower_case_table_names=1

Thanks to the following articles which helped me:

http://www.parkroad.co.za/mysql-case-sensitivity-table-names-between-windows-and-linux-systems

http://dev.mysql.com/doc/refman/5.0/en/identifier-case-sensitivity.html

https://forum.hibernate.org/viewtopic.php?p=2394017

It was complaining that the mysql gem was not up to date or something and I had to do a “gem install mysql”. But that resulted in errors as decribed in the following blog posts:
http://wonko.com/post/how-to-install-the-mysqlruby-gem-on-mac-os-x-leopard#comment-5544

http://boonedocks.net/mike/archives/175-MAMP-and-the-Ruby-MySQL-Gem.html

I had issues as I was using XAMPP and “gem install mysql” was complaining about missing mysql.h I had to upgrade my XAMPP and additionally download the XAMPP developer tools.

Go to http://sourceforge.net/projects/xampp/files/ and download:
xampp-macosx-1.7.2a.dmg
xampp-macosx-1.7.2a-dev.dmg

Of course first I had to remove/backup my existing XAMPP folder under /Applications/xampp Then I installed the new version and after that the xampp-macosx-1.7.2a-dev.dmg developer pack. this xampp-macosx-1.7.2a-dev.dmg contains the mysql.h header files and so on. After that dev tools installation the XAMPP folder has a new folder called “include” which has all the mysql header files like mysql.h Then the following command line worked for me:

sudo ARCHFLAGS=”-arch i386″ gem install mysql — –with-mysql-dir=/Applications/XAMPP/xamppfiles/

Note I was pointing to the xamppfiles base-dir as the mysql-dir, because then all the paths are relative as if this would be the mysql root dir.

Update: But as I am writing this, I was running into new trouble. It seems I need to update to the 64bit MySQL Version because of Snow Leopard, because now I am running into new problems:

Couldn’t create database for {”username”=>”root”, “adapter”=>”mysql”, “database”=>”mydatabase”, “password”=>nil, “socket”=>”/tmp/mysql.sock”}, charset: utf8, collation: utf8_unicode_ci (if you set the charset manually, make sure you have a matching collation)

According to http://www.techskater.com/ruby-on-rails/problems-with-mysql-gem-and-rake-on-snow-leopard/ this seems to be related to Snow Leopard and MySql 32bit vs. 64bit. On Snow Leopard the Gem Mysql needs to be build against a 64bit mysql or so.

I am currently reading http://weblog.rubyonrails.org/2009/8/30/upgrading-to-snow-leopard to resolve the issues and update MySQL. Let’s see…

Update 2: So I have finally installed MySQL 64Bit as suggested on the page above and I can start it. A “rake db:create” has also worked now as I could verify via MySQL Query browser. Let’s see what is the next thing which is not working…

Update 3: It really looks I am done and I am currently working through the Getting Started Guide.

Conclusion: 3 hours wasted to get it running on my OSX Snow Leopard. I did the same yesterday on my Windows machine at work and I didn’t have any such problems. It seems to be only related to Snow Leopard.  You might just read this page http://weblog.rubyonrails.org/2009/8/30/upgrading-to-snow-leopard and ignore everything above, but I leave it in this blog post for further reference. The difference is that you might not continue using the MySQL from XAMPP but a standalone mySQL. But you can still use phpMyAdmin from XAMPP or MySQL Query Browser to manage your database.

Something what always enjoyed in the Java world was the availability of easy debugging. In PHP in the past I have used ugly “echo …” debugging statements and never really digged any deeper. Now the time has come and I read that it is possible too but maybe with some more setup efforts.

Environment

• Mac OSX 10.6.1 (Snow Leopard)
• XAMMP for Mac OS
• Latest Eclipse PDT Full – I downloaded the 32bit Cocoa
• XDebug

The following articles helped me to get setup:

• Setting up Xdebug on Mac OS X (or Win32 / Linux)
• http://www.eclipse.org/pdt/articles/debugger/os-php-eclipse-pdt-debug-pdf.pdf
• http://www.oos-shop.de/xdebug-installation/
• http://www.eclipse.org/pdt/documents/XDebugGuide.pdf

Even though the articles are pretty good I had some minor problems, especially figuring out what exactly I have to add to my php.ini.

That’s why I give you what I have added to the very bottom of my php.ini which lives in the following path: /Applications/xampp/etc/php.ini

[xdebug]
zend_extension=/Applications/xampp/xamppfiles/lib/php/php5/extensions/no-debug-non-zts-20060613/xdebug.so
xdebug.auto_trace=1
xdebug.remote_enable=On
xdebug.remote_host=”localhost”
xdebug.remote_port=9000
xdebug.remote_handler=”dbgp”

The xdebug.so was mentioned in first tutorial above and I got it from here (Komodo IDE: http://aspn.activestate.com/ASPN/Downloads/Komodo/RemoteDebugging)

Here are some important infos and screenshots which might be helpful in understanding my setup:

• php executable: /usr/bin/php
• php.ini: /Applications/xampp/etc/php.ini

One of the most important things was the Path-Mapping, which maps your Eclipse Project structure to the URLs of your web-application.
URL of my app in the browser: http://localhost/zendtest/ZendFramework-1.9.3PL1-minimal/quickstart
Local-Path: /Applications/xampp/xamppfiles/htdocs/zendtest/ZendFramework-1.9.3PL1-minimal/quickstart

This mechanism is important, because it tells the debugger when to stop. If those settings are not correct (and they weren’t correct in my case) then the debugger will not recognize to which file the current URL in the browser belongs to and basically will not stop or will not stop at where you have placed your breakpoint.

Last but not least some screenshots which show the my setup and configuration:

Here is a page which helped me a lot with the migration. Pay attention to the “dummy entries” which are mentioned there:
https://wiki.nodomain.cc/projekte/blog-migration

One outstanding thing is to migrate the Serendipity permalinks 100% to wordpress.

In Serendipity I had the following structure:

http://www.keksrolle.de/index.php?/archives/139-Freemarker-Templates-Layout-Decorators-with-Shared-Variables-in-Data-Model.html

The problem is the “index.php?” –> the question mark spoils the links. I got it working with just /index.php/archives/… but my pages are indexed with the index.php?/archives. Shouldn’t be a huge problem as people just get to the homepage.

1. Purpose and Goal

Make the recipients understand purpose and goal. You are writing the email for a reason and not for the sake of writing a mail. Make that as short as possible.
Just as you do for your meetings: Give each email an AGENDA. People should know what your email is about before they read it.

2. Structure is key!

Structure your email into sections:
- Introduction & problem description: Why are you writing this mail and what do you expect.
- Symptoms: Say what you did, what you see and believe is wrong and what you would have expected to see instead. Never say: “Software XYZ is not working!“, because nobody knows what and how you expect it to work.
- Solution proposal: If you believe you have an idea what could be the reason for the problem and how it can be solved, then say that. But outline that this is only a guess. If you are wrong, then no problem.
- Conclusion: This is most important. After you have written and explained a lot, people are usually more confused than before. Summarize your mail in a few short notes.

3. Be polite

Never think your problem is the most important problem. Better always expect that everybody couldn’t care less about your stupid problem. If you think like that, you spent more thoughts into writing the mail and making a point. Do this in polite words. Remember you are dealing with human beings just like you who all have their own little problems, maybe a bad day and at the end of the day have to do their jobs – just like you. So just behave like you would expect it from everybody else too. If your colleagues are all assholes, then stand out of the crowd and do NOT be one!

4. Use screenshots, screenshots and screenshots!

In the IT and Software Development industry, problems are usually about software which happens on your screen. Instead of writing 10 long sentences about a problem you see on e.g. a Web page, just take a screenshot, add some red arrows and some circles to it and send it. If one screenshot isn’t enough, send another one.
Don’t even think of the larger size of your email. That doesn’t matter. Hard disk space is for free these days and productivity and information is more important!. Time is money.
I use SnagIt! for Windows or Skitch on my Mac for taking screenshots and adding annotations and stuff. Some of my mails consist only of screenshots because that might explain the problem without any words and the recipient sees exactly what I see. Make sure you use short keys to make screenshots. I just press the “Print” button which brings up SnagIt. It is possible to write a mail with a screenshot within 5 seconds.

Example:
It makes clear that I want to point out that specific part of the text.

5. Do NOT use abbreviations

Instead of don’t, write do NOT.
Instead of can’t write cannot
Always write the word not in capital letters and maybe bold, if you want to highlight that this is important. When reading emails too fast that small word can be missed and maybe it turns the sense of your mail upside down.

6. Favor notes or bullet points over full sentences

If you are listing a couple of related thoughts, then just put it into bullet points or a numbered list instead of writing it into full sentences. That has the following advantages:
- people can refer to each bullet point / numbered item which is very exact. You immediately know what they are referring to.
- easier to see what belongs together. In long sentences it can get hard to distinguish between the different things
- they are shorter and people can read it faster. If something needs more explanation, people will ask you.

7. Summarize your mail

Always add a conclusion or summary to your mail. After you have written a lot of text, people are usually more confused because of information overload.
Thus summarize your mail at the end in 3-4 bullet points. It helps to get back to the main purpose and goal of your mail.