RE: DOCBOOK-APPS: Announce: early beta of DiffMk 2.0

["Rankine, Alastair J (Alastair)" <arankine at avaya dot com>]

> Several people on this list have asked about DiffMk, my tool for
> generating changebars automatically for DocBook (and XHTML and XML
> Spec) documents.
> 
> The brave at heart will find a first beta of my Java rewrite of this
> program at http://nwalsh.com/java/diffmk/

Hi Norm,

Looks like this is a great tool. I recently tried it out on some of my
DocBook XML documents, but did encounter some issues, enough to prevent
the production of a PDF from the output. (Though after some checking it
seems that the XSL FO stylesheets ignore the revisionflag attributes
anyway...)

The good news is that I did get an HTML version of my document using
Saxon 6.4.3 and the XSL stylesheets. Some comments though:


1. Firstly it seems that the output of the tool does not parse using
SGML parsers. Specifically, entity references such as ‘ get
converted into their unicode equivalent, and this seems to confuse
OpenSP (I'm using 1.3.4).


2. Perhaps more seriously the output documents no longer validate
according to the DTD. For example I had changed an appendix to a sect1,
and the result looked like this:

	<sect1 revisionflag="added"><appendix revisionflag="deleted">
		...
	</appendix></sect1>

   Of course appendix is not valid inside a sect1. I appreciate this is
a difficult problem to solve in general.

   I came across a paper recently on diffing of tree-based structures.
Maybe such an algorithm would be appropriate for this tool?


3. If the external entities have not changed, is it possible to include
the entity references in the diff output rather than the entire entity
itself? (Does that sentence make sense? :) This is related to point 1,
but is more a convenience thing for large external entities.


4. Why did it convert

	<address><email>arankine@avaya.com</email></email>

   into

	<address
format="linespecific"><email>arankine@avaya.com</email></address>

   ?


5. Using the FO stylesheet, I get the following error with PassiveTeX.
The "Runaway argument" message makes me suspicious...

	               </5:inline>
	               Grabbed content
	               End Grabbed content
	               <5:footnote-body
	                   0:font-size = "8pt"  >
	Runaway argument?
	\utfeight@chardef \FOfontsize \XML@attrib
0:font-size-adjust\relax \FOfontsizea
	djust \ETC.
	! TeX capacity exceeded, sorry [main memory size=263001].
	<argument> ...feight@chardef \FOlastlineendindent
	                                                  \XML@attrib
0:leader-align...
	l.1053 ...fo:block></fo:table-cell></fo:table-row>
	
<fo:table-row><fo:table-ce...
	
	!  ==> Fatal error occurred, the output PDF file not finished!


6. Is it possible to do _both_ chunking and changebars?


7. What about the production of a catalog of changes? I hesitate to
suggest this, but it would be nice to see this controlled in a separate
HTML frame, with links to next/prev change.


Thanks again for a great tool, I'm sure I'll get a lot of use out of it.

Cheers,

--
 Alastair Rankine        arankine@avaya.com
 Avaya Labs Services R&D    Westminster, CO