[sword-devel] Early Church Fathers

Jerry Hastings sword-devel@crosswire.org
Sun, 22 Apr 2001 12:02:23 -0700

After giving the sentence idea a bit more thought.

In the Bible format now used the verse is the smallest division. And verses 
are the stuff of which all other divisions are made from. There are things 
smaller than a verse, like notes and Strong's numbers, But they are treated 
as part of a verse. There are also special verses, such as the zero verses 
and blank verses. And all this is derived from a one verse per line, or 
line by line, concept. Larger divisions point to smaller divisions and 
ultimately to a verse which points to an offset in the data file.

Can we define a basic building block like the verse/line of the bible 
format for free form? And is it needed?

I think we could come up with rules to decide when to start a new block, 
such as when starting a new sentence. Perhaps the start of a table or a 
cell would also start a new block. But just as a verse could have 
numberings and notes, a basic block could have multiple items on it.

Then what is the next level up for a division? At what level do you get all 
the formatting, font, background and other display info to put the smallest 
blocks in proper display context?

But is all this needed? Html help files work ok. If we could just display 
works in that format in a panel, it would be a good start. They have TOCs 
and indexes. Build ThML into that and it would be nice. I like being able 
to search for text and go right to the sentence, but if pages are small or 
highlighting is good, going to a page works.


At 04:03 PM 4/22/2001 +1000, Paul Gear wrote:

>There is a need in Sword for text that are free format, i.e. without any
>imposed divisions.  I know all texts have divisions, but by providing a text
>type that has no divisions, arbitrary divisions could be added with external
>indexes (or indices, if you prefer), which provide the type of key that they
>want and link that to an offset into the text.
>Jerry's suggestion about sentences might work in most cases, but what do 
>you do
>about tables and figures?  By making the text just a stream, you avoid the
>problem of having to classify the type of every text.  (Of course, you create
>other problems, but that's not relevant to my suggestion right now.  ;-)