[sword-devel] 1 bug left for 1.5.8
chrislit at crosswire.org
Tue Apr 5 01:17:39 MST 2005
Okay, I've taken a look at the test.osis.xml file Christopher posted in
the bug tracker and found the following. This is based on a module
compiled with osis2mod (from SVN) and the same module inspected in
BibleCS 1.5.8pre3, as well as the module exported with mod2osis &
mod2imp (from SVN).
(Quoting from Matthew chapter 1 in the document test file:)
> This is a test file to show off the problems encountered with Sword
> OSIS2Mod and Sword for windows rendering. In the examples below I've
> used '[' and ']' in the place of '<' and '>'
> This file was written using XMLSpy and validated successfully against
Just to confirm, it also validates in oXygen.
> This file is split up into several chapters each showing a different
> behaviour. The contents of each chapter is described below:
> 0) Introduction test:
> chapter 0: introduction which contains multiple parapgraphs is not
> rendered properly (some RTF rubbish)
Works/renders fine in 1.5.8pre3. Maybe the problem has already been
> 1) Paragraph tests: additionally chapterTitle in chapters 2-4 is just
The chapterTitle attribute on <chapter> elements is preserved by
osis2mod, but ignored by the engine. I have no idea what this attribute
is for, but would guess it is something we forgot to remove when we
changed how titles are done. In any case, the correct way to encode all
titles that occur in the text would be with <title>.
> chapter 2: paragraph markers outside of verse tags (paragraphs enclose
> verse tags) - resulting a complete mess in Sword for Windows RTF
It's unattractive but not RTF gibberish, so I suspect this has already
> chapter 3: paragraph markers inside verses, with construct like this:
> [verse s][p]text[/p][p][verse e][verse s]text[/p][p][verse e]. This is
> the workaround for problems introduced in chapter 2.
> chapter 4: paragraph markers completelly embeded into verses - all [p]
> and [/p] markers are inside the verse tags. Also considered by me as a
> woraround to problems in chapter 2
Workarounds are bad.
> 2) notes/xrefs problems:
> chapter 5: xrefs crossing paragraph boundaries
This is also just the problem above that was fixed.
> 3) problems with div's and titles
> chapter 6: div and title ignored if placed before the chapter start
This is correct, but where should content between two chapters go?
Should it go in the introduction to the following chapter?
> chapter 7: two separate divs and titles inside a chapter - possibly
> missing newline after the div end tag
This is just a criticism of rendering style, so not really a bug. And
it's a criticism specific to BibleCS. Short answer: no, div should not
generate paragraph breaks.
Best practice OSIS requires a structure of:
<div type="book"> with other div types embedded in that, notably <div
type="section">. But the lowest level of the primary OSIS hierarchy is
<p>. <p> should and does generate paragraph breaks. It never allows
embedding of <div>s, so <p> should be the lowest level always. (This
ignores simple Bibles without any paragraph structure, just
> chapter 8 and 9: div spanning two chapters - almost ok
Same issue as chapter 7.
> chapter 10: div and title just after the chapter start tag - also
> almost correct
Same issue as chapter 7, also. But there is also a problem here with
osis2mod. It's making two copies of <title> elements that occur before
the first verse. (One in the chapter intro, one in the verse 1 preverse
title.) It is only the second of these that is being rendered. Instead,
perhaps we should render the intro content and eliminate any preverse
titles in verse 1. This is an osis2mod and/or frontend issue (not API).
> chapter 11: two embedded divs just after the chapter start results in
> first title missing
The preverse code is only grabbing the previous title, so only the last
title is rendered. Implementing the above would fix this.
> chapter 12: two embedded divs in the middle of the chapter - the
> titles are in the wrong order
Again it's the preverse code. It's only grabbing the previous title to
put IT before the verse number. All remaining titles will appear in
proper order. (For German-speakers, it's just like verb-second. Only the
very last verb in a clause gets promoted to the second position in the
> chapter 13: three embedded divs just after the chapter start - two
> sections missing and verses concatenated
This one is the encoder's fault. :) All of the verses have the same
osisID (Matt.13.1), so they are concatenated into that verse. Aside from
the behavior parallel to chapter 11, osis2mod is functioning correctly.
> chapter 14: three embedded divs in the middle of the chapter - wrong
> titles order
Same as 12.
> chapter 15: a note in the title - not rendered as a note!
The data is correct, the rendering (in BibleCS) is incorrect.
> 4) Various tests
> chapter 16: test of alternative passages
<rdg> isn't handled, but it also isn't very interesting. I'll make a
note to add a handler in the filters, but all it will do is italicize
<rdg> is not for alternative passages, just alterate readings. So you
wouldn't have "<note type="alternative"><catchWord>something</catchWord>
<rdg>First version of the text</rdg> <rdg>Second version of the
text</rdg></note>" and expect there to be a toggle between the two
readings. The primary reading occurs in the text (and possibly in the
The OSIS 2.1 draft manual gives the following example of intended usage:
<verse osisID="NRSV.Song.2.1">I am a rose<note
osisRef="NRSV.Song.2.1 at s[rose]">Heb <rdg>crocus</rdg></note> of Sharon,
a lily of the valleys.</verse>
Thanks for going to the trouble of writing the test file. I'll continue
to use it to test until we get the import & rendering improved. It
illustrated a lot of areas where both osis2mod and BibleCS could use
some work, but I don't think any of the items need to hold up the
release of 1.5.8 (the API). I would recommend we attend to the osis2mod
and BibleCS problems before releasing them. (I realize that much of the
BibleCS code responsible for its deficiencies is in the OSIS2RTF filter,
which is technically part of the API, but I don't think this is used by
anything other than BibleCS, so I wouldn't see a problem with releasing
1.5.8 with a buggy OSIS2RTF, so long as it gets fixed before the next
BibleCS itself is released.)
More information about the sword-devel