[sword-devel] I implore you...
dfhmch at googlemail.com
Mon Jun 10 06:32:02 MST 2013
At the time we prepared the TurNTB module for release, we were told that
SWORD did not support mid-verse titles.
Accordingly, the module was built after implementing the negotiated
workarounds in the USFM file for Acts.
My difficulty is this:
Starting with received USFM files in which there are section headings tagged
(or similar headings with other kinds of tag),
there is at present no proven successful method to create a working module
in which these headings are (a) all present and (b) all correctly placed.
usfm2osis.py places section titles BEFORE the eID milestone of the previous
osis2mod expects to find section titles somewhere between the eID of the
and the sID of the verse following the title.
If it doesn't find them there, the OSIS transformations described elsewhere
(the adding of the attribute subType="x-preverse") simply doesn't happen.
In the output of usfm2osis (.py or .pl) there are simply no gaps between the
eID milestone of verse N and the sID milestone of verse N+1.
Hence I believe that what Chris calls "OSIS best practice" is at odds with
the SWORD utility.
I may be wrong; others in CrossWire are far more experienced programmers
than I am.
I'm not proficient in C++ in any way, and would be counted as a beginner in
Yet after my email exchange with DM Smith two weeks ago (before his
I still think we have have evidence that seems to have eluded most of the
That this is hindering folk such as myself and Peter making accurate high
quality modules would be a huge understatement.
The other side to this is that section titles at the start of a chapter are
usually missing after the two stage process involving usfm2osis and
For this too I have useful further evidence.
Inserting a verse 0 tag *\v 0* in every chapter of a set of USFM files makes
an improvement of sorts (albeit not for every front-end app.)
After this empirical kludge, at least Xiphos displayed the section titles at
the start of each chapter. That other front-ends don't can be explained by
the more fundamental issue.
usfm2osis places titles where mod2osis cannot use them aright.
I hope this helps. It's something that's been bugging me since January 2012,
when the TurNTB module was released with misplaced titles.
Peter persuaded me then that "good was good enough" and that it was better
to release the imperfect module than not to release it at all.
Yet, IMHO, making and releasing "less than perfect" modules, when we have
full access to the translators' source files does not show CrossWire in a
I was going to send a private email to Chris about usfm2osis.py - yet now
that the current thread has asked for testable instances for mid-verse
titles, I think I must bring this wider topic to the fore.
It would be better that I were completely wrong, than that I might be right
and the issues continue to exist and remain inadequately understood or
This really should be in a new thread, as I don't wish to hinder the release
of SWORD 1.7
If that's preferred, then let's start a new thread. I can easily repost the
text of this email.
Yours in Christ's service,
View this message in context: http://sword-dev.350566.n4.nabble.com/I-implore-you-tp4652437p4652467.html
Sent from the SWORD Dev mailing list archive at Nabble.com.
More information about the sword-devel