[jsword-devel] Bundling JSword with FireBible

Mon Nov 14 11:32:37 MST 2011

And Bible couldn't use smartcn because it was too large and slow for mobile
phones so we used mmseg4j.  From memory the only change required was
changing AnalyzerFactory.properties as below:
zh.Analyzer=org.crosswire.jsword.index.lucene.analysis.Mmseg4jChineseLuceneAnalyzer

and adding the attached file.

Regards
Martin

On 14 November 2011 08:36, Brian Fernandes <infernalproteus at gmail.com>wrote:

> DM,
>
> Thanks for the notes. Given my need to release sooner rather than later, I
> will stick to JSword 1.6 for now; as that is the version  that FireBible
> currently recommends, there are just too many index related issues to take
> care of to allow a quick release.
>
> I think I will *not* bundle the libraries into FireBible but make some
> sort of downloading provision where FireBible will automatically download
> and unzip the JSword distribution (probably through a set of dialogs with
> minimal user input). For future FireBible updates, I will update the
> version of JSword it points to.
>
> I've been out of the JSword space for too long, so if there are any
> specific areas or bugs in JSword that you could use help with, please let
> me know. How is the alternate versification work coming? If I can help
> there, please let me know how.
>
> In Him,
> Brian.
>
>
> On 11/13/2011 5:12 PM, DM Smith wrote:
>
>> On 11/10/2011 05:14 AM, Brian Fernandes wrote:
>>
>>> Currently when you install FireBible you are expected to point it to a
>>> JSword installation on your system, the FireBible extension is small and
>>> does not include JSword.
>>> When I submitted this for inclusion in the Firefox add-ons list, they
>>> told me that the whole installation process was just too complicated;
>>> besides the fact that you must have Java installed, you also need to
>>> download JSword, unzip, point to this location in FireBible. I can see
>>> their point, especially for non dev types. FireBible is currently listed
>>> (but hidden from most public) in their extension registry for this reason
>>> and I intend to correct this now, at least the JSword installation bit.
>>>
>>> I intend to include the JSword binaries in the extension, but continue
>>> to give users the ability to specify an external JSword installation if
>>> they wish. This takes the extension size up from 300KB to nearly 4MB. But
>>> that is with JSword 1.6. With the latest nightly builds, the sizes of JARs
>>> are significantly higher and I have a few questions:
>>>
>>> a) I assume a large number of fixes and enhancements (like faster/better
>>> searching) have been made since 1.6 and they are in the nightly builds.
>>> Does it make sense to include a nightly build in the extension or should I
>>> stick to 1.6 for now? Is there a more recent stable build that I could use?
>>>
>>
>> The nightly build is probably the best at this point, but test it first.
>>
>> I know there are some problems with the BibleDesktop resources.
>>
>>
>>> b) In the nightly binary, there are multiple versions of Lucene - 2.9.1,
>>> 3.0.2 and 3.0.3. Is this intentional or can I stick with Lucene 3.0.3?
>>>
>> For the nightly build, you should be using only the latest. The others
>> should not be there. That is a problem with our upgrade process.
>>
>>
>>>
>>> c) I know the newest version of Lucene, Lucene 3, will not work with
>>> indices created with Lucene 1. Some time ago DM mentioned that the code to
>>> detect that the index present was a Lucene 1 index and thus invalid was not
>>> complete. Has any work been done on that front or will users still have to
>>> manually delete their indices?
>>>
>>
>> That's correct. They will appear to work but won't. Also the nightly
>> build also assumes that the indexes are built with it.
>>
>> The index detection has not been completed. It will need to be before a
>> release is done.
>>
>> Also, Java 4, 5 and 7 differ in the Unicode standard they support. (5 and
>> 6 were the same.) This will require new indexes for some languages, but
>> without analysis it'd be best to rebuild all non-English language indexes.
>> (It hasn't affected ASCII and probably not affected Latin 1.)
>>
>>
>>> d) What is lucene-smartcn? That JAR is 3.5MB in size and almost equals
>>> the size of all the other JARs that make up JSword. From what I could find
>>> online, it adds better Chinese indexing but I would like to avoid including
>>> this as it has a very significant impact on the size of the extension.
>>> i) Will Lucene / JSword still work if this file is absent?
>>>
>> Yes. But you may need to change a resource file to point to the other
>> Chinese analyzer.
>>
>>
>>> ii) If yes, will indexing of Chinenese modules be completely broken or
>>> just not as "smart" if I remove the library from the distribution?
>>>
>> You'll have to try and see whether it is broken. :) Regarding "smart", it
>> includes a Chinese dictionary that helps it know what are words (a gross
>> oversimplification).
>>
>>
>> For searching to work it has to use the same analyzer with which the
>> index was created.
>>
>>
>>>
>>> Thanks,
>>> Brian.
>>>
>>> ______________________________**_________________
>>> jsword-devel mailing list
>>> jsword-devel at crosswire.org
>>> http://www.crosswire.org/**mailman/listinfo/jsword-devel<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<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<http://www.crosswire.org/mailman/listinfo/jsword-devel>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.crosswire.org/pipermail/jsword-devel/attachments/20111114/9aef0a7a/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Mmseg4jChineseLuceneAnalyzer.java
Type: application/octet-stream
Size: 1047 bytes
Desc: not available
URL: <http://www.crosswire.org/pipermail/jsword-devel/attachments/20111114/9aef0a7a/attachment.obj>