[sword-devel] Sword support of indents and line breaks

Peter von Kaehne refdoc at gmx.net
Sat Apr 13 04:35:21 MST 2013


Can I try and summarise what I think is going on here?

Basis of problem:

1) CrossWire has a whitespace/title/poetry problem which has been
discussed at nauseam for many years. But it has neither been thoroughly
resolved nor even isolated. Various parties have blamed each other -
module makers, osis2mod maintainer, filter maintainer, frontend
developers (and maybe others, including bystanders) DOI - module maker
and happy to blame anyone, including myself as long as it finally
improves.

2) IBT had a specific need and had someone in their midst willing to
cater to that need, using our CrossWire's efforts but was miffed by
problem (1)

3) IBT - outside of the Wycliffe/UBS "tradition" of using Paratext - has
developed it own use of USFM with specific rules about markup, sometimes
breaking semantic rules.

The problem grows:

4) IBT's lead programmer decides to solve IBT's immediate problem by
applying his own patches to sword and maintain his own version of
libsword and OSIS with little or no communication with Sword people and
no attempt to isolate and solve the underlying problem (1) either.

The problem becomes apparent:

5) People in IBT's area start to use smartphones and Mac's and
subsequently IBT's offer of xulsword is not sufficient anymore - making
(1) obvious to everyone and letting (4) fail.

The problem becomes a bunfight:

6) Instead of looking at (1), (2) and (3) respectively and finding both
the remaining whitespace bugs + fix IBT's "dodgy" USFM encoding we are
fighting for two positions which are actually both not particularly
good.

I note John's recent email which says that translators do not understand
and/or care about semantic markup, just how it looks like and that it is
understandable. I think this is fair enough - only the "understandable"
varies from platform to platform. What is understandable on paper will
not work on gobible and will create a mess in other places too. So, as
computer people it is our job to find the patterns and encode the
patters correctly, semantically, including on occasion feeding back to
the translaters - "yes, I know what you want to achieve, but _this_ is
the way you should encode it, we will fix the presentation afterwards."

The <milestone/> element is indeed intended to encode translater intent
for the odd and bizarre. But IBT's desire is not odd and bizarre. Only
its solutions are. A lot of times when the <milestone> is used there are
perfectly adequate structural OSIS elements available - q, p, div,
poetry markup and a few more. A few milestones will subsequently maybe
still remain. But not even 10% by my reckoning. 

On CrossWire's side though we need 

a) the ability to add per language or per module the means (maybe CSS?_
to ensure that certain language conventions which are necessary for
readability do not fall foul of our (mis)understanding how certain
structures should look like. If a language requires that e.g. a <q/>
element is not just encoded with a ", but will include a linebreak, an
indent, a longdash and a few more things at the beginning and some
others at the end, then this should be acceptable and reproducible.
Routinely and easily.

b) a final push to get the USFM/OSIS/module witespace/interverse
content/intraverse extra content/title etc mess sorted. 

Peter
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part
URL: <http://www.crosswire.org/pipermail/sword-devel/attachments/20130413/5c2a5544/attachment.sig>


More information about the sword-devel mailing list