[jsword-devel] Jsword development

[DM Smith]

I am back from Easter vacation and now have a bit more time to respond.

DM Smith wrote:
> Kurt Andrews wrote:
>> First the good news:
>>
>> The current build is working for Mac OS X 10.4.x
>>
>> Now the bad news
>>
>> WebStart is still broken (for me at least)

This is odd. We had another report of it being broken, but then I 
downgraded from requiring 1.5 to 1.4.2 and the result was that WebStart 
works for another that has 10.3.

Could you open WebStart and remove the application and then try it 
fresh? (I think your results might be cached.) Both WebStart buttons, 
too. Try the nightly one first.


>>
>> personally, i think the font for small verse numbers and these links 
>> could be just a bit larger.   turning off small verse numbers all 
>> together is even worse.
>
> Just change simple.xsl to your liking.

The other thing you could do is pick a bigger font. This is what I do.

>>
>> Are there any particular issues you'd like me to work on.  I have a 
>> complete java development environment installed with eclipse and 
>> subclipse, and I am at your service.
>
> Our road map is in Jira. See what you might like to do. Also, I 
> suggest poking around and finding and fixing bugs you find as a way of 
> becoming familiar with the code.
>
> More later, I have to leave to catch a plane for Easter vacation.

The biggies right now are: (Most of these are in Jira)
We don't support all SWORD modules. Such as GenBook and a particular 
dictionary format. GenBook is going to be the basis for supporting 
alternate versifications.

We don't support Personal Commentary. To do this we need to add write 
capability for SWORD modules.

Our search algorithm for determining the SWORD_PATH is different from 
SWORD. It should be the same as in SWORD's SWMgr.cpp, with the addition 
of looking at the current jsword locations first. We should also use the 
SWORD_PATH environment variable if set.

We should add the ability to download a module by specifying an URL to 
the raw zip or tar.gz file. (This should be trivial)

We should add the ability to have a local install site (e.g. CD-ROM, 
network, ...)

We should add the ability to run entirely from a USB key.

We should add the ability to enter an unlock code for modules that don't 
have one but need one. It would also be good to add validation code to 
determine whether the unlock code is correct. How to do this was 
discussed on sword-devel and can be found in those archives.

In the installer the size of the download is always wrong. This should 
be fixed.

I would like to see the installer become a separate executable, in 
addition to being a menu choice.

We should replace our home-grown plugin mechanism with an industry 
standard one, namely Eclipse's OSGI implementation.

We plan to replace the Swing implementation of BibleDesktop with 
Eclipse's SWT/JFace/RCP. The standalone installer would be a good place 
to start. There are a bunch of "issues" in Jira that make better sense 
after a migration (parallel texts, persistent preferences/state 
restoration, font rendering improvements, toolbar customization, hover 
definitions, ...) These would be things where Eclipse provides a 
standard, better or significantly different way. That is we don't want 
to waste our time on what we get for free by re-engineering, what would 
be a marginal improvement now or what would need to be significantly 
redone by re-engineering (reduplicated effort).

We plan to allow the direct use of an OSIS document (at least Bibles). 
The jsword-devel and sword-devel archives have extensive discussions on 
how to architect this. A good place to start would be to also allow for 
osis2mod emulation, which builds a SWORD module for a OSIS bible. The 
key characteristic of this should be that for any OSIS text, Ox, its 
conversion to a module, Mx, should be reverseable to an OSIS document, 
Oy, which may be different. Converting Oy to a module should result in 
Mx, not My. (Ox -> Mx -> Oy -> Mx) For any book, chapter, verse that is 
well-formed in the OSIS document, the concatenation of those verses from 
the module should also be well-formed.

With regard to search, we are very English centric, most specifically, 
we don't handle accents well. The user should be able to search with and 
without them against text that is accented or not and find the sought 
material.

At this time we don't index strong's numbers, morphology codes, 
cross-references, notes and any other non-canonical text (except by 
mistake!) It would be good to add these and allow them to be searched.

It would be good to add search hit highlighting.

I think it would be good to add a "Find related verses" capability. 
(e.g. we submit the text of a verse as a ranked fuzzy search, returning 
the most likely verses, using the threshold specified by the user)

Add ability to show/hide headings.