During christmas break, I checked my (very) old personal Java applications. One of the first comments was from 2001. A long time ago. I had no UI framework, no testing framework and no automatic build process. All was hand-made I was amazed that I could read and edit the source code without problems.

One of my biggest and first applications, that I wrote for my personal use, was "MP3Tool". It is a tool to manage (very) large MP3 inventories, it has cool automations, supports offline edit/merge via XML, supports database loading and has command line support. It has support for ID3v1 and ID3v2. I have implemented my own library to read and write ID3 TAG infos (not complete, but works still great).

The tool is a "developer" tool that is not very user-friendly, but it is 11 years old and has tons of features. It was developed from a developer for a developer If someone needs a Java tool for mass editing of MP3 Files - here it is. It is not Open Source because it was not relevant in 2001. It is Free for everyone and if someone needs the source code it should not be a problem to upload it somewhere.

The tool has a German and English language file. I don't know if the auto detection works but you can change the language file in the settings, via Toolbar. I have not tried it on MacOS, but used it very often with Linux. If you search documentation - I have never used one!

I use the tool from time to time and still love it because it just works. After 11 years, I think it is ok to share it with you.

To be sure - The tool is not a SIB Visions Tool - It is a retro style swing application developed long time ago

File overview

File optimization

TAG conversion

We made some tests with TableView and missed productivity features like Load-on-demand and database support. So we tried to use our DataBooks with a TableView. We need a TableView that is re-usable and shows data from remote storages like databases or csv files. It should support multi-column sorting and of course, load records on demand. A nice feature would be support for edit data without boilerplate code.

Our first simple use-case: Show records from a database table

The result:

TableView with IDataBook

The implementation is not finished and is more a PoC, but it works great! The source code is available here.
It is also possible to edit data, but the different datatypes are not yet supported!

Simply use the table view to show dynamic content. Allow editing without additional source code. Don't re-invent the wheel again!

The source code of the test application is available here.

JavaFx 2.x has some nice UI components. It has new APIs and it is completely different to Swing. Of course, you can mix JavaFx components with Swing components, but the LaF is not the same! The new styling mechanism of JavaFx are really great.

But for me, its not a big step forward in the evolution. It is another UI library, of course a visually appealing. I miss an application framework or components with built-in support for modern applications. From my point of view, there is not enough focus on business applications. A business application is not a simple table or chart with some nice looking bars. A business app should handle millions of records, should allow keyboard navigation in all variants, should be fast, should save costs and time.

What should be different?

A standard JavaFx 2 application looks very nice because the standard style is modern. Also styling options are better than ever.

For me, the whole API is too much focused on Lists, POJOs, Generics and Object Bindings. Of course, it is cool for small apps, but you waste time with boilerplate code. If you have a database application with 5 simple tables that contains master data, and you need edit screens, you duplicate your source code 5 times because you have no support for dynamic usage. If you work without Generics, it is not really funny because JavaFx uses it everywhere.

Load-on-demand of records is possible if you do it manually, but it is not included in the controls. "Touch-scrolling" would be great. Compared to other UI libraries (swing, swt, qt), you can't change the default operation of UI controls because you don't have access to internal listeners, scrollbars, Behaviour, ... Some parts are customizable but these are more the exception.

The whole event handling is sometimes very complex, because you don't know what event type you have to use, where the type is declared and which event is the correct one. And inner classes are not the best solution for clarity. Because of Generics you have a lot of configuration work and you are not sure if everything works as expected. Without powerful IDEs you have no chance. Sometimes you have events but nothing happened....
That parts are not the real problem because it requires a certain amount of time to understand new classes. But must things always be complex rather than simple?

But it is great to write source code instead of crazy scripts.

What else?

The version number 2 is too high, because it is still not mature. JavaFx shows what is possible (scene graph, effects, web and video content) but it is not simple to use. But maybe it is perfect to combine only some parts with other mature toolkits like Swing or Swt?

It is not the library that bothers me, but it brings little relief for a developer. Why not just use Pivot? Yes... the web view is powerful

What should be done?

Focus on Business applications. Integrate an application framework that saves development time. Controls should not be so restrictive!

We use Eclipse Galileo for all our projects. It has some problems with CPU load if you use Subclipse with large Java projects. But if you disable SVN decorators, it is ok (not perfect, but still better than the rest)...

We don't use every new Eclipse version for development because it needs a lot of time to check and change our projects. Don't believe that everything works out-of-the-box.

In the last days, we tried to update our Android plugin and... the new version does not support our "old" Eclipse version. We need at least Eclipse Helios... sh*t. We opted for the update, hoping that it does not take too much time.

Installation of Indigo and all our plugins was straight forward. But the "new" JDT have some side effects. It was not possible to compile JVx because KeyValueList and IdentityKeyValueList have a "Name clash". The problem details are described here. Not an Eclipse problem, a generics problem. But even that is not great.

After we set the compliance level for JVx from 1.5 to 1.6, it was possible to compile the project. Now we have Name clash warnings instead of errors. Not great but we can live with it. After we "solved" the first problem we had the next with our Web projects, because the Project Facets use Java 1.5 compliance level. But we learned that it is enough to set the level to 1.6 . To set the level from 1.5. to 1.6, we had to change the JRE for our Server environment(s) from 1.5 to 1.6. It is not a problem to use 1.6 but JVx is 1.5 compatible and it is not great to test without 1.5, but it is what it is.

Most of our projects are now migrated and again it was not funny to search "no" problems. We look forward to the next waste of time.

If you need details about our migration steps, write a comment

The new version of our first application - 1.2 - includes JVx 1.0.

We replaced the old JVx version with the new one and used the APIs.

The current version of WebUI - 0.10 - supports JVx 1.0.

Our showcase application already uses WebUI 0.10.

Our Showcase application is up-to-date! We have updated all libraries to the latest versions.
The source files are available here.

Test the new REST interface:

List of Contacts
List of Educations

Use admin as username and password.

JVx 1.0 is out and we have some Source Code statistics for you!

Our last statistics were determined for JVx 0.8. It is cool to compare 0.8 with 1.0. We have again worked very efficiently because 1.0 contains approximately 420 Features and Changes compared to 0.8. The Source Code grew by 18.767 lines, since 0.8. That means, we need 44 Code lines for one feature

Here are the numbers:

Hm... we thought that we reach 100.000 LoC with JVx 1.0, but we "only" have 97.454. Compared to the features of JVx, we have still too few lines of code. Maybe we should use a code generator or Object Mapper or simply integrate a bunch of new features?

Some additional numbers:

The test result is not perfect but we are happy with it. We should nevertheless pay more attention to the test coverage... and we still need a solution for automatic UI tests.

We proudly present JVx 1.0!

What a year... The first release of VisionX is out, and we finished our work on JVx 1.0.
This release is awesome. It contains a total of 230 changes to 0.9.

We implemented really cool productivity features, some fancy UI features and we fixed some bugs. But the complete list is really long

Our original Roadmap for 1.0 had fewer Features, but thanks to the community, JVx 1.0 now contains a lot of wonderful things.

In the following weeks we plan to write more documentation especially about the new features and update our showcase and demo apps. It is not possible to write about everything in this posting - and it is also Christmas. One of the next postings will be some statistics about the JVx source code - LoC, quality, ...

We think it is also time for more WebUI... maybe another implementation... let's see. And of course, mobile devices are still very interesting. We started first attempts with Android some months ago. We made some POCs with Pivot and JavaFx 2.0 and who knows what the next year brings.

And we have still a lot of new ideas to speed-up the development time. Without JVx 1.0 you develop as fast as possible - but with JVx 1.0 you develop with the speed of light.

Now, a very short look at JVx 1.0

• Database
  • Support for PostgreSql and MySql enum datatypes (automatic link cell editors)
  • High performance metadata cache
  • Block fetching
  • Scriptable db import scripts
• Server
  
  • REST support for all objects and actions
  • DetachedSession and DirectServerSession introduced
  • DirectObjectConnection introduced
  • Transparent config encryption
  • LCO based security checks
  • Reduced communication requests
  • Better support for custom storages
  • Support for custom session manager and object provider

• User Interface
  
  • Intuitive link cell editors
  • New cell formatting with image and insets
  • JNLP services integrated
  • Focus rect handling
  • En/Disable translation per component
  • New Locale handling

You find the complete list in the Changelog.

And last but not least, we look forward to your feedback!

MySql and PostgreSql have some useful custom datatypes. Enum is one of them. MySql has a Set datatype too.

With an Enum, you define a list of allowed values, but only one value is possible for the column (same behaviour as check constraints in other databases)

With a Set, you define a list of allowed values and multiple values are possible for the column (more like a n:m relation). But you should not use Set datatypes, because it makes your data model unreadable and everyone wants clean data models?

With JVx 1.0 it is possible to use Enum or Set datatypes. On UI side you get a link cell editor (dropdown) or choice cell editor (e.g. checkbox).

Look at the image on the right. You see 3 columns. The first is a simple text column, the second is a linked cell editor (dropdown) and the third is a checkbox. Nothing special? Not until you see the database table.

The table and types, in PostgreSQL, are defined as:

The server-side storage handles the enum/set detection automatically and sets the linked cell editor and allowed values into the MetaData. You don't have additional work on UI side to get above table!
And this is really cool, isn't it!

But we recommend a new creation style for your storages in your life-cycle objects, to get the new features:

Old style:

New style:

Why we recommend this?

With 1.0 we introduce db specific storages like MySqlDBStorage and PostgreSQLDBStorage. This storages support database specific features that are not standardized.
We avoid/reduce boilerplate code whenever possible and who wants to change source code if the database is changed?

With the new style, you get the best storage available for your database, automatically. Of course it is possible to create e.g. a MySqlDBStorage() manually, but keep in mind that you are fixed to MySql.

After you changed your life-cycle objects, you get linked cell editors for your enum columns. But in many cases you use enum datatypes as check constraints e.g. Yes/No columns. In that case, a linked cell editor is not very cool - you need a choice cell editor.

JVx defines some default choice cell editors, via

in the Application class. With this mechanism it is very easy to define choice cell editors for your enum types.

Set datatypes are very special and generally, we don't like unreadable data models. But JVx should support it as good as possible.

We handle Set datatypes like enum datatypes and we show a link or choice cell editor. The user can only select one value. A developer can set multiple values via API (as comma separated list).

Have you ever seen this comfort in other frameworks? Let us know