[jsword-devel] feather request: book downloading and index generating api

P. R. B. dysbiote at yahoo.com
Fri Dec 8 22:44:15 MST 2006

Another thing to consider is to put the JSword and Common code into a basic OSGi bundle / RCP plugin jar and have BibleDesktop run an OSGi framework implementation (something like Apache Felix). The Crosswire jars would stay independent of RCP, yet RCP developers could treat the jars as standard plugins. Some abstraction classes would still need to be made to the JSword/Common code for RCP'rs, to reduce the conflicts that Zhaojun and I ran into. 

I don't know enough about OSGi to say what other pros and cons there are.

As for making BibleDesktop a standard GUI RCP: Yes, you'd be dependent on SWT. 

Any other thoughts? It seems like we've got some good ideas floating around.

----- Original Message ----
From: David <lzj369 at gmail.com>
To: J-Sword Developers Mailing List <jsword-devel at crosswire.org>
Sent: Friday, December 8, 2006 10:50:14 PM
Subject: Re: [jsword-devel] feather request: book downloading and index generating api

The main advantage for RCP that it hides the low level details from business logic. A big plus is cross platform is piece of cake. We can click several times to make it run on almost every platform like linux, windows,mac ,ppc, you name it. It is native code call from java layer, thus very fast.

The biggest challenge is the size.

 The default size is like 9mb. 

Another advantage is update: RCP has built-in update site support. It even supports anatomical update! even scheduled update! 

Plus, it is de facto standard for JAVA IDE now. and IBM is behind it.

as far as BD gui, it is hard to tell because it depends. My experince tells me that it is very easy to mimic the current gui with two views. Installation view can be done , I have it done already, it can be shared. The other part is main gui: it should be a Form base with a broswer GUI or two. 

The preference can be done by using RCP built-in preference api. It is very simple. I can share my code by providing a simple tutorial.

BTW, all code are open sourced.  

I have different opinion about the classloader. Yes, jsword's class loader creates an issue for RCP. However, I do not think removing it is a good idea. We can add addition classes or addition methods to jword (Jsword, and common) to solve this issue. Maybe is is time to create an standard eclipse plugin. I can offer help on this one.

Next week I will release all the code to a place so everyone can share and modify.  The GUI like a clone of WEBSOWRD on crosswire.com

In Christ,


On 12/8/06, DM Smith <dmsmith555 at yahoo.com> wrote:
Zhaojun and Phillip,

I think there is enough interest that it would be good to meld all this into one effort.

Phillip, I'm sorry you have burned out. Perhaps if we had collaborated we could have shared the load. 

Please help me understand the advantages and disadvantages of going to RCP for JSword and Common. And what impact it would have on the BibleDesktop GUI. How portable is RCP? Will it run on all platforms that Java runs? Or is it limited like SWT?

I've done a bunch of reading, but have been focusing on adding behavior to what we have. So I have not really experimented too much.

In Him

On Dec 8, 2006, at 9:17 PM, P. R. B. wrote:

Hi gang,

I ran into the same problem with RCP. I ultimately turned the core JSword libraries (common and jsword -- the parts I was after) into an RCP plug-in that could accommodate the RCP Job class and use plug-ins to extend the installer, filter, and driver (RCP needs to manage class loaders itself, so the class utilities and property files were removed). The key to changing the job code, for me, was to replace JSword-Job calls  with IProgressMonitor calls, and making all calls run in the caller's thread. The IProgressMonitor interface serves essentially the same purpose and the JSword Job class, so the change was fairly straight-forward.

I have an example of the changes that I made at the bottom of this e-mail.

 Off-topic: I hit project burn-out with this endeavor a few weeks back and the fate of the code is undecided (I take project burn-out pretty hard). The JSword plug-in is functional and the RCP application is 100% working. If anyone wants to play with it or is interested in helping me resurrect the thing back to life, please let me know.



Example changes to convert JSword Job code to use IProgressMonitor  (for non-RCP'rs, IProgressMonitor and IStatus are part of the Eclipse/RCP core API):

Installer interface:

    IStatus install(IProgressMonitor monitor, Book book);

HttpSwordInstaller class:
    public IStatus install(IProgressMonitor monitor, Book book){
        // Is the book already installed? Then nothing to do.

        if (Books.installed().getBook(book.getName()) != null)
            monitor.beginTask(Msg.INSTALLING.toString(book.getName()), 1);

            return Status.OK_STATUS;

        final SwordBookMetaData sbmd = (SwordBookMetaData) book.getBookMetaData();
            // the task has as many steps to perform as there are parts to the download

            monitor.beginTask(Msg.INSTALLING.toString(book.getName()), (getSize(book) / 4096) + 1);
            URL temp = NetUtil.getTemporaryURL("swd", ZIP_SUFFIX); //$NON-NLS-1$

            // download the book. Each chunk downloaded contributes to the monitor's work.
            download(monitor, directory + '/' + PACKAGE_DIR, sbmd.getInitials() + ZIP_SUFFIX, temp);

            // Once the unzipping is started, we need to continue
            File dldir = SwordBookPath.getDownloadDir();
            if (!monitor.isCanceled())
                IOUtil.unpackZip(NetUtil.getAsFile(temp), dldir);
                SwordBookDriver.registerNewBook(sbmd, dldir);
        catch (Exception ex)

            Reporter.informUser(this, ex);
            return new Status(IStatus.ERROR, JswordActivator.PLUGIN_ID, 0, Msg.UNKNOWN_ERROR.toString(), ex);

        return Status.OK_STATUS;

----- Original Message ----
From: David <
lzj369 at gmail.com>
To: J-Sword Developers Mailing List <
jsword-devel at crosswire.org>
Sent: Friday, December 8, 2006 10:26:01 AM
Subject: Re: [jsword-devel] feather request: book downloading and index generating api

Hi, DM,
If you just want to migrate to SWT, I personally think it does not worth it. SWT just another set of API, like SWING.  However, if you want to migrate to NETBEAN RCP or Eclipse RCP, I will applause your effort. (The application size will be increased by 6-9MB).  For Eclipse RCP, the work is just create veiws and call jword api. 

The work I am doing is based on Eclipse RCP. Jsword is a base part of it. The plan is to build a p2p system that can enable blog , forum sharing, of course file sharing(  JXTA.)  next year.  It will be a set of plugins. 

Right now, I am rewriting the system to look like exactly what is on web sword interface. After all, people are familar with web interface.  I , however did a websword plugin with tomcat embedded(yes, a server on user's desktop).  It is cool. However, due to other considerations, I put it on hold. 

Since you are interest, I will put the source code to Google code next week.  I tried sourceforge, still does not get approved.  An alternative is I put the code temporily into jwsord SVN, so you can look at it. Again, next week. :) 

On 12/8/06, DM Smith <
dmsmith555 at yahoo.com> wrote: We hope to migrate the UI to SWT. So I am very interested in your RCP

work. Is there a place where I can see what you have done. If possible,
I'd like to minimize your pain of re-integrating the JSword as it changes. 

David wrote:
> Eclipse RCP platform has its own threading api. Use can choose the job

> run on backgroud or foreground. I maybe need to double check why the
> process failed. (I believe it is because of threading) 
> For book installation, I added two mirror methods to do downloading

> and copying.  It works fine. The only issue is the api and package
> keep changing and I have to compare every time there is a new release 
> and merge the code.
> For indexing, I have not made it to work correctly.  I will study the

> code next week. The basic idea is kick off a background job and call
> jsword api to generate index. No thread for indexing itself is needed. 
> For Web sword, the book installation is on server side. Never mind.

> It will not be used for end users. I put it on hold because I have
> several othe plugins need to be done soon.
 > Zhaojun
> On 12/8/06, *DM Smith* <
dmsmith555 at yahoo.com
> <mailto:dmsmith555 at yahoo.com>> wrote:
>     David wrote:

>     > Hi, DM and fellow developers,
>     >
>     > I am developing web sword(90% done, pure j2ee implementation, will
>     > help web hosting) and sword on eclipse RCP.  Issues were raised up 

>     > when I try to integrate install book and generate indices. The code
>     > now has job api build in and also has the reporter to
>     communicate with
>     > user UI.
>     > 

>     > In order to accelerate the acceptance of jword library,  the api for
>     > downloading books and api for index processing need to seperated
>     from
>     > Job api.
>     I'm not sure I understand the problem. 

>     The index api (org.crosswire.jsword.index) is independent of the
>     Job and
>     Reporter apis. The Lucene implementation is not.
>     Likewise for the install api ( org.crosswire.jsword.index
 ) and the
>     sword
>     implementation.
>     Both the Job api and the Reporter api are listener based. If there
>     is no
>     listener for Job events or Reporter events, then those are not heard. 

>     Any listener of your choosing can be provided or not provided. It
>     is up
>     to you.
>     The purpose of the Job and Reporter apis is to provide asynchronous
>     communication of a potentially background task thread. For 

>     example, you
>     will notice in BibleDesktop that you can download and/or index
>     more than
>     one Book at a time. Each download and index is on its own thread
>     and it

 >     communicates back to BibleDesktop asynchronously of its progress
>     or any
>     problems that are encountered.
>     In a web environment asynchronous communication of long-lived

 >     threads on
>     the server may prove to be a challenge, but it should be possible.
>     _______________________________________________
>     jsword-devel mailing list
jsword-devel at crosswire.org <mailto:
jsword-devel at crosswire.org>
>      http://www.crosswire.org/mailman/listinfo/jsword-devel

> ------------------------------------------------------------------------
> _______________________________________________
 > jsword-devel mailing list
jsword-devel at crosswire.org
> http://www.crosswire.org/mailman/listinfo/jsword-devel 


jsword-devel mailing list

jsword-devel at crosswire.org

jsword-devel mailing list

jsword-devel at crosswire.org

 Check out 
the all-new Yahoo! Mail beta - Fire up a more powerful email and get things done faster._______________________________________________
jsword-devel mailing list

jsword-devel at crosswire.org


jsword-devel mailing list
jsword-devel at crosswire.org


jsword-devel mailing list
jsword-devel at crosswire.org

Cheap talk?
Check out Yahoo! Messenger's low PC-to-Phone call rates.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.crosswire.org/pipermail/jsword-devel/attachments/20061208/5c59d6d1/attachment-0001.html 

More information about the jsword-devel mailing list