As the documentation on heroku site didn’t work, here is the DATABASE_URL value I use on my env.

export DATABASE_URL=jdbc:postgresql://localhost/$(whoami)

And to manually start a local PostgreSQL database (if not automatically started)

pg_ctl -D /usr/local/var/postgres -l /usr/local/var/postgres/server.log start

Benoît initiated a github organisation with "A curated list of stuff about Awesome Living Documentation" and a maven plugin. And I joined him, and Cyrille too. So we’re developing a maven plugin implementing Cyrille’s principles for living documentation. Cyrille gave is repo dot-diagram, which is a dependency for living-documentation to the organisation.

I also recommend link:Cyrille’s book[https://leanpub.com/livingdocumentation]. You can also see link:his talk at Devoxx 2016 on YouTube[https://www.youtube.com/watch?v=Tw-wcps7WqU].

But my wife can’t play on her Wiko Highway 4G, the game "was not" able to locate the phone by GPS. I didn’t found why until I asked myself and try to "hack" the game on my phone with "devloper mode" and "fictive positions". The result is clear : the developers have well done their work and detect the activation of the "fictive positions" in the developer mode. So you can’t play and found pokemons.

As the result was the same on the wiko, I suspected the "fictive positions" were enabled by default on that phone. An it’s the case :

I activated the developer mode (go in the parameters > about the phone > click 4 times on the "build number"), then go back and enter in the "developer mode" new menu. The "fictives positions" were activated by default. So I unchecked the mark and retry the game.

And tada, the game works well now :D

I’ve it with the help of this conversion tool, but modified it a little to add :

The tool is not perfect but as it has to be used only one time, it’s enough to migrate the blog with less manually operations after the migration :

The last thing to do is to log in your hubpress admin page and publish your posts one by one, and the migration is done.

I was searching how to get a segment of an url (including some / characters) into a single parameter. The response is on stack overflow : use a regex.

The goal for me is to reroute a subset of RESTFUL call to the server really implementing the services.

http://www.journaldulapin.com/2012/11/15/recevoir-des-fichiers-en-bluetooth-sur-un-mac/

By default, you can’t receive files via BT on MacOs. You have to activate it in System Preferences/share.

And it’s what I made in my Codebaord Project to test Play! Framework. In my model, I have a class Domain that is only a database wrapper (with only annotations), and a controller Domains that has code that should be in the model (see old controller source with this wrong choice).

So, I started replace domain code in the domain classes.

Now the controller class is lighter :

and the model class is no more "empty" :

@Entity

public class Domain extends Model \{

public String name;

public long position;

public boolean isPublic;

public boolean isDefault;

public String toString() \{

return name;

}

public Domain delete() \{

if (! isDefault) \{

super.delete();

} else \{

throw new UnsupportedOperationException("dont_delete_default");

}

List<Domain> domains = Domain.find("position > ?", position).fetch();

for (Iterator iterator = domains.iterator(); iterator.hasNext(); ) \{

Domain domain = (Domain) iterator.next();

domain.position = domain.position - 1;

domain.save();

}

return this;

}

public Domain setDefault() \{

Domain defaultDomain = Domain.find("isDefault = ?", true).first();

if (defaultDomain != null) \{

defaultDomain.isDefault = false;

defaultDomain.save();

}

isDefault = true;

save();

return this;

}

public void move(String to) \{

if (to.equals("highest")) \{

List<Domain> domains = Domain.find("position < ?", position).fetch();

for (Iterator iterator = domains.iterator(); iterator.hasNext();) \{

Domain domain = (Domain) iterator.next();

domain.position = domain.position +1;

domain.save();

}

position = 1;

} else if (to.equals("higher")) \{

Domain domain_to_swap = Domain.find("position = ?", position - 1).first();

domain_to_swap.position = position;

position = position - 1;

domain_to_swap.save();

} else if (to.equals("lower")) \{

Domain domain_to_swap = Domain.find("position = ?", position + 1).first();

domain_to_swap.position = position;

position = position + 1;

domain_to_swap.save();

} else if (to.equals("lowest")) \{

List<Domain> domains = Domain.find("position > ?", position).fetch();

for (Iterator iterator = domains.iterator(); iterator.hasNext();) \{

Domain domain = (Domain) iterator.next();

domain.position = domain.position - 1;

domain.save();

}

position = Domain.count();

}

save();

}

}

[/code]

\1. it’s not possible to delete defaults datas

The delete button is not present if the object is the default one.

\2. it is not possible to uncheck a default value

Editing an object, the default attribute is read-only (shown with an image in my code) if it’s a default one; else a standard default check box is shown.

#\{if domain.isDefault }

<img alt="Toggle_check" src="@\{'/public/images/toggle_check.png'}" />

#\{/if}

#{else}

<input type="checkbox" id="domain_isDefault" name="domain.isDefault" value="true" $\{domain?.isDefault ? 'checked':''} />

<input type="hidden" name="domain.isDefault" value="false" />

<span class="errorExplanation">#\{error 'admin' /}</span></p>

#\{/else}

[/code]

To complete these tips on template, a third tip

\3. it is not possible to have twice object being checked as default

So if we’re setting an object to default, we must uncheck the old one. This control is done in the controller (and not in templates).

if (domain.isDefault) \{

Domain defaultDomain = Domain.find("isDefault = ?", true).first();

if (defaultDomain != null) \{

defaultDomain.isDefault = false;

defaultDomain.save();

}

}

domain.save();

[/code]

[to be continued …​]

The trick is simple :

I had to actions to do it :

To enable the two behaviours, the use of the render method, that needed to identify a template, is replace by the renderArgs.put that is independant of the template.

So here is the example updated to enable rendering the objects on view of different datas of on different templates of the same datas

In the example, due to ACL, if the user have no rights to view an issue, an "Access denied" page is shown in place of the standard view to edit the issue.

As we saw, the implementation of the Logic tier is easy with Play!. In this implementation, you can remove CORS on the tier.

The Presentation tier will also be relatively easy; the Logic tier will be considered as the model of the Presentation tier : the diagram below illustrated the architecture.

Play! on a 3-tier architecture

Instead of making calls to a classic Play! model (such as <code>Account.findAll()</code>), the  controller will  call the REST API of the logic tier, and render a view as a response :

The view code will be a classic html template code. There’s just a trick, as strings will be surrounded by quotes ("). The code will be <code>$\{object.property.getAsString()}</code> instead of <code>$\{object.property.getAsString()}</code>.

And that’s all. Really simple to make a 3-tier application with Play!

Related articles

So the action about ruby was easy :

For tomcat 7, I had a previous manual installation on ubuntu 11. As tomcat 7 is embedded with precise pangolin, I just had to copy the file missing fomr my previous manual installation :

Here a sample model example that could occurs in a bank application :

A client is specified with this model, that includes the password of the client.

This client has accounts in the bank, so a simple model would be :

One of the first service of the application when a client connects to the application is to show the list of his accounts :

The route is :

And the code of the controller is :

This code will expose the following json datas :

Oh, the password is visible.

So what we will do is to extend Play! with a new Controller class that will use add a Strategy to gson :

Now, we will wrtie this strategy. The goal is to exclude fiels annoted (we will use the tag @NoExpose) in the json output.

You can notice that with exclude fields, but that we can also exclude Classes. This can also be very interesting to exclude a model (so all datas of a particular table).

The main code is done. All we have to do now is just two basic actions.

-annotate the fields we want to exclude :

The documentation says :

For example, the 'play' command writes a lock file in dir.

WHY ?

So how do you share the installation between users ? How do you ensure nobody broke the installation ? I don’t like that choice. Here are the commands to still install to /opt (or other irectory of you choice).

What I’d like from the play team :

Play 2.0 is out but 1.x branch is not dead. Really ?

Issue #1304 is, in my opinion, a critical defect (absolutely impossible to use version 1.2.4 with PotgreSQL). It was opened on December 7th, 2011 and resolved on December 12th, 2011 : but no new release as been published since.

What kind of support of the framework can an enterprise expect ?

Play! is a great framework, but must have a clearer and better support policy for enterprise that use it in critical businesses.

and this additional line is present in the save of the controller :

I finally found the solution in this discussion of the play-framework group. As said by

Guillaume Bort, an hidden field must be added after the checkbox field to have the binding done when the checkbox is unset : when unset, the value is null and so no binding is done. So the hidden field is used to binf the false value.

And you can now remove the setting of the boolean field in your controller.

The goal of this pattern is to avoid code duplication using the @Before/@After annotations even for objects to be rendered. The trick is that these methods can also contains calls to the render method !

To avoid code duplication, you must use the @After annotation to a private method which will have the common code of your controllers and call render on these objects. And the "main" controller method of each view will render the specific objects of the view.

The example below render the same view on edition and creation of an object ( of design pattern #1), nut could be used on different views.

As you can see in this example, this pattern is particularly useful for combos that can appear in many views of an application.

Typically you’ll pass an issue to the form when editing an existing one, and no one for the form to create new ones. Taking codeboard as an example, this would result in the following code in the controller and the template :

This is not the better way to handle the form :

The good pattern is to create a new instance of the object in the model class :

The example below show the Issue model with management of default values that can be modified in others forms :

Now the difference between the method of creation and edition in the controller are just different by the initialisation of the issue object passed to the renderer :

Then the code for the drop down list in groovy is just 1 line of code now !

Trying to call WS services (with JSON format) from the front controllers is transparent.

But trying to make dynamic pages with JQuery querying from the front page the the back controller, the results were blocked due tothe Same Origin Policy. So I had to specify the Cross-Origin Resource Sharing on the back controller.

Here is the code to add on the controller of the backapplication : this will add the header to allow the caller to access the service :

You can sepecify a list of domains to accept as origin, or set it to * to allow all domains to call your service.

And the JQuery call :

You can generalize the CORS setting with the @Before annotation, so all your services in a class will be accessible :

Then your services coding in the controller is wery simple :

There’s now a ant task to use sonar on your project build by ant. You’ll have the same results, except that by default, you will not have the results of your tests (passed/failed) and the coverage of your tests. To add them in your sonar pass, you have to :

Here’s the sonar task that will include the two reports generated by junit and cobertura  tasks :

I’ve put the jar of the sonar plugin in ant library directory, so the classpath to sonar is left blank. And if you’ve install the sonar plugin in eclipse, you’ll have to remember the groupId and artefactId to link the project with the sonar results. And the advantage is that you’ll have the link betwwen the violations and the source line directly in eclipse.

More interesting info is the replacement of the use of the deprecated class People.

But this don’t mention the correction for the "Force close" when launching the app on Android 2.3.3.

The trick is the attribute outputpropety of the exec task. You can then use ${host.name} variable to use the hostname in your ant scripts.

So a better option is to use the UUID of the drive and set the auto mount option. To gert the UUID, launch gparted again and right click on the partition. In the dialog box you’ll have a item will the UUID of the partition. Just copy the value. Edit again /etc/fstab and set the line of mount point the the code below this time :

Related articles

In fact, it’s a bit hidden. You have to go to Menu → Download, and you’ll see previous installed apps, included Aspahalt 5.

Edit : by the way, I found it now on the market and there’s a free version !

So, if as me your favorite clock is TypoClock, here it is (as a reminder for me) : the thread on xda

Related articles

I use a NAS as my main drive to maintain my datas shared against my computers (and to not have to make copies when changing it). But my Buffallo crashed a disk for the second time (the same one,  I think it’s corrupted and I’ll try to have it replaced by the support).

But before sending it back to consumer’s service, I want to have my datas back on a rescue disk. So I wanted to use the safe disk to copy files on a third drive. The problem is that the disks use XFS filesystem. No way to copy files from Winows. Fortunately, I have a dual boot with Ubuntu, that natively can read XFS ;) . so below are the steps to achieve the copy process :

Get the packages to support XFS :

Now you have to identify the partition to mount : use gparted

You can use the drop down list on the top right to select your hard drive, then you’ll see the list of partition; just identify your partition with XFS filesystem (/dev/sdd6 in my case).

So next step is  to edit /etc/fstab to specify the mount characteristic (filesystem type : xfs and associated mount point : /media/whatYouWant ) ; just add the line above

No create the mount point for your drive and mount it :

It’s finished, you can access your drive and make your backups.

Searching on the web, it’s often said to use :

But in fact, I didn’t found a keyboard option.

Just launch :

On these virtual installations of Ubuntu I couldn’t have the Unity interface : the reason is simple, Unity need graphic acceleration, so do not forget to check the graphic acceleration in the configuration of your virtual machine.

The full answer is :

Then you can run a terminal and call su. Now you can run the last two steps in the shell

And Voilà! No reboot.  You’ll have the rotation on Google, the compass will work again and you’ll can play your favorites games (eg car race games). And personally, my clock I launch when I go to bed (it’s the way I noticed there was no more auto rotation).

Just go to "Control panel" then "Programs" and "Turn windows features on or off". You’ll just have to go down to "Internet Information Services" (or IIS) and select FTP Server AND IIS management console. See the screenshot below (in french) :

Then you have to administrate your IIS : in control panel, click System and Security, and then click Administrative Tools and launch IIS Manager. There just add a FTP site (right click on the Sites folder).

And you’re gone with your ftp site : no download, run as a service; the hard part is to know it’s included in Windows 7 and where to find it.

Details : Go Launcher with 3 more icons (for a total of 15) made by me for the dock (GTalk, TweetDeck & Facebook) - Wallpaper from LP.

- Screen 1 : TypoClockText - Battery Widget (may be later replaced by BattStatt) - MinimalistText

- Screen 3 : Smooth Calendar - Games icons (from iPhone theme) resized by me - Go LauncherImage folders - Dektop Visualizer (contact link) - 4 icons for apps by me

By default Telnet is not available in Windows 7. Just type the two following commands and the command will be "back" in your new Windows version :

pkgmgr /iu:TelnetClient pkgmgr /iu:TelnetServer

[/sourcecode]

Though you can see the http://developer.android.com/sdk/android-2.2-highlights.html, there’s a number of interesting new features I noticed. Some are from the Samsung user interface "TouchWiz".

The first one is the stamp on sms, e-mail with the numbber of unread items.

The management of the seven home screens is also reviewed : with the pinch gesture, you can have a view of the 7 small screen you can arrange with drag and drop and delete dragging to the  dust bin.

On my first home screen, I’ve put :

There’s some new standard applications too :

Yes, Press Reader is new and not "online" for the moment.

The management of applications has changed too :

and the market comes with his new features :

A great enhancement is the new feature of text selection, copy/cut/paste :

Once you have already made a copy or cut, there’s also  a paste icon available.

The newq notification bar have new toggle buttons :

The goggle enhanced search tool :

It’s nothing but a little change in the colors :

And some more colors again :

Posted from WordPress for Android

Then put it back in a nice new box (mine can host odd and sata) and plug the new plug via USB : you WD passport is back with all tour datas.

Posted from WordPress for Android

Posted from WordPress for Android

First we will add the Event class. If you use Eclipse and the Play! plugin, you can use the new play! menu :

image::screenshot-23.jpg[image,title="Play! Eclipse plugin"]]

The class will be very simple for this how-to :

Then we add the link between Task and Event; add this property to Task :

Now we will add a link in the current view to show the detail view, add a column with a html link :

Now we have to had the controller action task in Application :

Last action is to write the task.html template. It will be very similar to the index template, except we will use the template tag on the events property of the rendered task object :

The important part is the save action; here we pass the id of the master object as it will be used in the save action in the controller. So here is the save action in Application :

In the controller,

we get back a reference to the master object (via the id passed from the view),

we save the event entered in the form

we add it to the events list property of the master

we then save the master object

and we call again the action to show the detail of the task.

So here is our master-detail views how to finished

.

there a sass module in play. Just type the following line to get the module. It comes with fancy-buttons embedded.

Don’t forget to run eclipsify (or the command of your IDE) again and refresh Eclipse (your IDE). I’ll make this how-to on my datepicker project (from previous post).

Add the module to your application in conf/application.conf :

Not you just have two files to modify.

Create the sass stylesheet (main.sass) in /public/stylesheets. Import fancy-buttons and "style" your submit button :

Really simple, isn’t it ?

Add the call to the stylesheet in main.html :

And, voilà :

I’ll let your change the look : see more on author site.

Working on a personal project, I was looking for a java framework to develop a web interface. I didn’t want to use a "full stack" java framework and searched for a RAD  approach and found some information on wicket and trapestry and finally found Play!.

I first take a look at the live coding demo and was impressed : exactly (or almost) what I was looking for. Java and simple page templates (with groovy). Watch the video and take a look at the samples in the package and you’ll have a good idea of the power of the framework.

I’m just trying it since a week and can already create some application : I’ve already one done (just having to add some entities but the core is already done).

Most of my work was learning JQuery (that I didn’t know before) to make a dynamic web application : edit in place, ajax, …​ I’ve for example mix an editinplace on a date with a datepicker. So Play! made me discover JQuery I already heard but never get a try in the forest a javascript frameworks (prototype, mootools, script. aculo.us, …​ didn’t know where to begin).

There already some great modules as CRUD and Secure, to make the admin part of your application for example, as in the blog sample.

Perhaps more about Play! and JQuery soon ?

Following comments on my previous post (http://javathought.github.io/2009/06/09/reusable-ant-build-file/), I finally arrived on this excellent presentation from Douglas Bullard at JavaOne 2009. Read it, it will really improve you scripts.

The basis is to use the import.

PS :

Sorry for the delay of this post. I’ve moved my home, so I was packing all my stuff this summer.

I’ll post again soon, the next subject will be about ant

About Green ball plugin, not only, there’s no feature enhancement, but it’s an accessibility regression : color blind person don’t easily differentiate red and green. To have an idea of how blid color persons "see" this two colors, just take a look at this post : Great UI: Clarity and Color on the Presentation Layer, and learn good use of colors for your UIs.

So, just keep Hudson with blue ball and really useful plugins.

Setting your file ecnoding charset : by default, Eclipse uses cp1252, wihch is not quite ideal if you have to deal with internationalisation and character manipulation in your datas with differents charset.

Just go to the window- preferences menu and select General - Workspace in the tree. Then change the "Text file encoding" parameter :

Then select Other and choose UTF-8.

Don’t forget to set the encoding parameter to your javac javadoc target in your ant tasks :

That’s all.

If you have lists in your xml, you’ll probably handle them with Jtable or JList in your UI. Modifications of objects (from the edition form if you don’t edit directly in your list) will be seen in your table, but creation or deletion of an element in the list will not be seen. To see creation or deletion events in your UI lists, your object list have to be observable, i.e. implements ObservableList.

JAXB generate the following code for xml lists :

The way to create an an observable list is below; beans binding library contains a factory :

I would appreciate a way to modify the creation method in JAXB, but didn’t find it. So I created a class which implements the ObservableList interface (with an almost Decorator Pattern) and specify to JAXB to use this class to handle lists :

So, first step, create your own class for lists, that implements all method of ObservableList :

Then specify xjc to use a custom bindings settings. Use -b file.xjb, the file containing the following instructions :

This way, JAXB will call

The plugin from no0ne handle this UI requirement (with the library CollectionBeanProperty).

So I used another plugin, a mod of the original from jaxb-commons found on http://moscaville.com/. This one replace the vetoable events by fireChange events. It is less complete than the one from http://www.no0ne.org/, but interesting for learning more on JAXB.

First, this plugin has a two bugs :

Then, this plugin does not handle the lists (what the plugin of noOne does). This will be the subject of the next post.

However, I’ve been using Java and XML on the prject and experienced more on JAXB. So, I’ll tell more soon.

The way I use is to apply the Decorator pattern.

The reason is that  we only want to add row highlighting behaviour to existing cell renderer, and not to recode all the behaviour of cell renderers. This is particularly right for booleans values, which are shown by check boxes in JTable. With this pattern, you won’t have to rewrite the behaviour of the cell renderer.

So here is the code of our new cell renderer. It inherits of DefaultTableRenderer. It will apply a background to a cell depending of :

Then all we have to do to use this cell renderer is to get existing cell renderers from the table, apply our decorator on them and set this new renderers to the JTable.

Look at the second one : we keep the behaviour of the cells showing booleans as we apply our new renderer upon the existing one : the check boxes are kept to show booleans values in the JTable.

You can also find many useful poster or "Refcardz" on dzone.com. The Design Patterns one is also made by Jason. ou’ll find others on Java, Netbeans, eclipse or others.

Take onaother look at devirtuoso.com to find 9 more ones quick card with one on subversion (from  http://www.addedbytes.com/)

To have a list of Hudson plugin, just visit the hudson wiki.

Fortunaltely, there’s a quick and easy way to add a project in Hudson with source control under Dimensions. Just add a step to the build with shell script or Windows batch command and use the Dimensions command line dmcli :

You should be able to use the option -file also (which i didn’t try) : see this post on Serena forum for more information.

To get more help on dmcli option, just call dmcli without command and use help to get more information on internal commands. The cons of this is that you won’t have changes informations in the hudson change log :(.

On Windows, if you use Hudson as a service, the dmcli command won’t work directly : I suppose you must update the path before calling dmcli (or enter the full path to the command, but didn’t check it.

My personal choice is Swing, first af all because I use it for a while. For example, I worked on a simple application in 2003 to created thumbnails images for original (a web photo gallery) written by jimmac.

Developing this application, I developed some Swing widgets (a splashscreen, status bar, a toolbar with outlook behaviour) you can find on my old java page and download here.

So, I prefer Swing because :

The last reason will definitly confirm my choice of Swing. And really take a look at Romain’s blog to his impressive Swing GUI notes (and even his photos).

I’m back from some vacations.

The topic I want to discuss this time is the use of ant build files. I don’t like the build files generated by netbeans. Why ? Working in a professional environment, you have to master all your building chain : know what happens in the continuous integration process, know exactly what you deliver to your production team (in a professional environment developers package the release and deliver it to production team; only production authorized team deploy the package on production systems), …​.

So I don’t like the maze of ant files generated by Netbeans. Here is the reusable ant file I used. you can rapidly use it back in a new project with fex modification (mainly the project name).

This file contains targets for :

You can had dependances from other projects in the prebuild target, for example add this :

Hopes this can help you. And don’t hesitate to report possible errors.

But you can go further with JAXB. You can heavily customize JAXB with annotations. Using the same phonebook example, you can introduce annotations in your schema to make JAXB meet your needs.

A phonebook is made of contacts, so developing the mockup presented in part 1, we will add the contacts in our schema :

What JAXB will generate with this schema. A new class will be generated : Contact. It’s what we wanted, but take a look at Phonebook class. There will be a new field and an associated get method :

Not quite what we expected has the get method returns a list, so we expected the get method to be getContacts which is clearer : the method does not return one contact.

To correct this we’ll use annotations, to have a Contact class and a getContacts method :

You can use those annotations for more structural customisations :

I looked at hibernate, which now make XML bindings, but, from what I’ve seen, the binding is managed by the relational mapping. As I do not have a database in my application, I had to search elsewhere.

So I took a look at JAXB. It’s the tool I needed : you specify your xml schema, run xjc and you have your object mapping done. As example, I’ll use a schema I’m writing at the moment (phonebook.xsd). Now, all you have to write, once the classes are generated is 4 lines of code (plus catching the exceptions:

Let’s comment the source :

But you have three things to know to better use it : first if you want to directly unmarshal to your root class and to write an object representation to an xml file, the master object class has to be tagged @xmlRootElement, if you want to write simplier code.

Take the following schema, that does not generate the @XmlRootElement annotation :

If you use that schema, you’ll have an error (java.lang.ClassCastException: javax.xml.bind.JAXBElement cannot be cast to package.name.Phonebook) when unmarshalling and an other (unable to marshal type "package.name.Phonebook" as an element because it is missing an @XmlRootElement annotation) when marshalling. You’ll have to write the following java code to unmarshall and marshall without errors with the schema above :

Instead, a better way to declare your schema (and have simplier java code) is :

The two others tips to knows will be the subjects of the next posts.

As Hudson, you just uncompress the archive wher you want (for a standalone deployment), run and it works. Then you create your project, choose your build configuration (there’s default configuration for ant, maven 1 and 2, and console).

Just a think to now : you have to specify your SCM url in the maven format (even if you use ant) : for subversion, it will look like <code> scm:svn:https://host:8443/svn/Repository/trunk </code>, otherwise you’ll have a message like Provider message: No such provider: 's'.

First impressions testing it, compared to Hudson : at first sight, I prefer the dashboard of Hudson, which have a better view of the past builds with a meteo indicator computed on past 5 builds.This give you a view of the stability of the builds.

image::hudson-1.png[Hudson’s view of builds,title="hudson-1"]]

Continuum is organized ny projects groups, So you have a first view on the group with counters on successful project, with a counter (a project counts for one).

image::continuum-1.png[The project groups view of continuum,title="continuum-1"]]

When you click on a group you then have a view on the projects of the group :

image::continuum-2.png[the view of the projects in the group,title="continuum-2"]]

There’s a release button on each group in the main view, and on each project in a group view. You’ll find the documentation on the release functionality on Continuum website. I should try this soon. For the moment, I’ve not been able to get the prepare options to continue the process.

I find the navigation in Hudson more intutuive, but it’s quick to findhow to navigate in Hudson. The organisation of project in groups should also be useful in a large organisation, matching groups of project with the team’s organisation.

I’ll also try integration with Jira later.

so it’s really possible and you really have a good C IDE. I’ve used Eclipse C environment for a while and it’s really good C environment. I’ve tried Netbeans since I’m developing parts of my Java projects in Netbeans, but I stopped for the moment : I could not be able to use the debugger in Netbeans, quite a hige issue for an IDE.

So, until I try it again and resolve that issue in Netbeans, Eclipse is for me the "Java IDE" for C development.

So to achieve this pattern, the has to follow the process, with the cost overhead. I think the distibuted model of mercurial as a smaller overhead in managing feature changes. This process is much simplier with a SCM that implements the process : developers summit their changes attached to a change request, all in the same trunk. Before a release the SCM manager just pick the validated chage requests. In such SCM, branchs are used to maintain releases, not the features.

Thus, Oracle will acquire new activities to concurrence their competitors. Especially, the sun server activity as Oracle didn’t ship hardware before. A good news to Sun’s customers. They will also acquire concurrent activity of their actual solutions, as MySql.

Personally, I’m particularly interested in the future of Netbeans, of course. And also, the raise of VirtualBox. Will they improve the solution to compete with VMWare ?

As I’ll write a desktop application, I’ll Netbeans for the IDE.

Next, I need a SCM (which is a Key Process Area of CMM Level 2) or rather a version control system. But to choose it, before I got my request/issue tracker. It’s Jira. My short list was svn (subversion) and mercurial (hg), the later is a distributed version control system (VCS), which differ from traditional centralised ones. I elected svn, though mercurial principlpe is interesting, but svn intregration in Jira is simplier (the plugin is mature).

svn is not the more mature VCS I used (I wouldn’t tell it’s a SCM), but linked with Jira, we get a missing feature. Correctly used, the changeset are linked to requests, but I found some lack of flexibility (how to decide which request will integrate a release, if the last changeset cumulates all requests ?), though I need to learn more about svn. I think to achieve flexibility, you have to work with branches; but this is harder than in others SCM.

I decided to use a continuous integration (CI) tool also. The list I tried was CruiseControland Hudson. I selected Hudson, because it’s was simpler to install and configure. The configuration is intuitive, not the one for CC. And it simply integrates with Jira too. So in Jira, you track the changeset that fixes your request and the successful build that integrates your request. I began to look at apache continuum, but hudson was already configured and it seems it has not the simplicity of hudson. Perhaps I’ll try it more later.

So here is my "developer suite" :

I’ll have a look at natbeans' code coverage plugin next time and compare it to JTest (which does not only do code coverage analysys).

First, what is HCR ? The precision, because you may think it does not work, but there’s some limitation in what you can change in your code to fiw it while debugging; it is not possible to do the following :

To summarize, you can change "internal" code of methods.

Now, how does it works on these two IDEs :

In Eclipse, you just save your code, then your code is recompiled (assuming automatic build is set - see menu Project → Build automatically). Then the modified code is available in the current debugging session, the instruction pointer is back at the beginning of your method.

The difference in Netbeans is that you have to explicit specify that you want your modified code to be "pushed" in the debugging sessions through the menu Debug → Apply code changes. By the way, if you search more information on the net, search for Hot Code Replace for Eclipse and Apply Code Changes for Netbeans.

So, is VE up to Matisse features ? The response is quickly given : No !

VE does not give you the rapid placement of components you have in Netbeans. In Eclipse, you have to place exactly the component where you want it to be to have a professional form.

In Netbeans, magnetism is active and place component in a professional way with the GroupLayout. Spacing is automatic, so field text drop perfectly behind labels. Components can be "attached" to borders, so that they grow with the containing panel. A component can have the width of a group of others components. Indentation of components is given, and so on …​

You’ll not have this rapid form design in VE.

But, what makes Netbeans rules in GUI design against VE is the implementation of JSR-295 Beans Binding ! You simply drag a bean class on the form to have it available for binding. Right-click on a component, then bind and you’ll have it binded to the property you want. You can start learn beansbinding with this introduction.

Opposite, Eclipse gives you nothing. You’ll really spare time developing your GUI in netbeans. In 1 or 2 day you can design an effective form that works, not only a mock-up but a full funtionnal GUI ready to deliver.

edit : And the other advantage of Netbeans is the ease of internationalizing your application. No code to write, just write your strings in any language. The IDE can automatically set up the code to load localized labels.

Next time, I’ll speak about the debugging.

Both are good at editing source code (fortunately). There’s no major difference on basic tasks to develop an application. I’ll give you the major differences I’ve found using them to develop J2SE applications. You’ll have all what you need to develop, make unit tests (with JUnit), package (create jar files), automate your builds (with ant). And you’ll find additional plugins to help you. So on which feature will you make your choice.

First, it seems Netbeans has better integration of some java technologies. E. g., I’ve develop some applications manipulating xml data files. I choose JAXB to handle the xml-object binding. Netbeans natively supports JAXB. You create your schemas and Netbeans is ready to generate your classes mapping your xml datas. Under Eclipse you’ll have to install JAXB and the associated plugin. Quite a minor disagreement.

Further, Netbeans is from far better when profiling Java applications. You launch your application by the profiling menu : a screen ask you parameters to specify what you want to profile, you click run and …​ this is all. Your application is being profiled as the result is there quickly. Under Eclipse, you’ll have to install an agent outside of eclipse, laucnh the agent (outside of eclipse again), and laucnh the profiling. And the bad news is : it’s very slow. The overhead of profiling under eclipse is just unbearable. I’ve never been able to wait the end of the profiling.

And what about developing desktop applications ? What about Swing GUI builder and beansbing. I’ve been waiting too long for a GUI builder in Eclipse. VisualEditor (VE) seems to have been quite inactive for a while (see VE project page) , but is back active (you’ll have to search deeper to find new versions on the wiki page of VE). But I didn’t tried it (when you disappoint and loose a client, it’s difficult to get it back). Perhaps I’ll try it again and give it a chance. There’s also a new alternative  on eclipse : VisualSwing4Eclipse. I ddn’t tried it. But facing this situation, Netbeans comes with embedded Swing GUI Builder (formerly Project Matisse). It comes with great wizards; you’ll find very goods tutorials on netbeans.org. It comes with BeansBinding integration, creates well design Swing application implementing JSR-296 Swing Application Framework to provide responsive desktop applications.  You rapidly build professional GUI for your applications (I personally build the desktop client of an application with MQ-Series middleware in one day).

So, is Netbeans the winner ?

On the other hand, in a professionnal environment, you work with other tools to achieve my tasks and goals. So the integrations with them is an important part in the choice of the ideal (not perfect) IDE. In my case, I work with SCM (software configuration management), code analysis tool, and issue tracking software, which of them have plugins for eclipse but not for Netbeans. SCM is the most important part in the integration aspect. This fact is perhaps due to the relationship with WSAD/RAD.

Will this change if, finally, IBM buys SUN ?

So, I’ll continue to develop on Eclipse and keep Netbeans for GUI and profiling.

I’m Pascal and this blog is a new one where I’ll post all my notes about Java. I could also write old stuff about Java that I wrote elsewhere to keep only one place for my writings about Java.

This will also be a place to stock my sniplets.

Hope I’ll see you soon back here.