[sword-devel] Fix for &

Karl Kleinpaste karl at charcoal.com
Thu Sep 28 07:23:03 MST 2006

So Sword really is positioned as a referee.  I can't say exactly why,
but that surprises me.  Probably it's twenty-plus years in network
environments, where the fundamental interoperability maxim credited to
Jon Postel -- "Be liberal in what you accept, and conservative in what
you send" -- is deeply ingrained in me.

In any event, given the perspective, a complete fix must necessarily
include both [a] resurrecting the commented-out substitutions for both
ThML and OSIS (as I mentioned, there are OSIS modules such as WEB
which fail to display properly due to these problems; also, HunUj, as
an OSIS problem example I bumped into during this process, uses &# and
so probably needs to be repaired elsehow) as well as [b] simply
removing the passThruUnknownEsc control.  That boolean was already
init'd false anyway, but by this design it shouldn't even exist.
What's been described is a design decision which simply forbids the
very idea.

And Chris, please take a look at using my script for converting Clarke
to a far more readable form.  Completely outside the question of
fixing the buggy mishandling of &entity;, my original motivation which
led to this deeper question was that Clarke's forced line breaks are
not visually correct and should be repaired.  The script is a fast and
brutal one-step conversion, though the result should probably also be
recompressed if the update is to replace the current Clarke; the
pipeline just ends with imp2vs.  An updated, finalized version could
be available from crosswire.org literally in less than an hour, for
someone with the right access privs.


PS- In surfing up the exact Postel quotation, I encountered this:

"In essence, implementations may relax the requirements on its inputs
 and can be stricter on the conditions of its outputs."

This puts the XML "hard stop" perspective in a rather backward light.
And I daresay that the Postel (which is to say, the Internet's)
perspective can claim to have stood the test of time quite a bit
longer than XML.

More information about the sword-devel mailing list