Re: [xsl] xml -> htmlhelp and character 8220

[Allin Cottrell <cottrell@xxxxxxx>]

On Fri, 12 Nov 2004, Jirka Kosek wrote:

> Output encoding for HTML Help couldn't be implicit as HTML Help is 
> very buggy piece of software. It doesn't support UTF-8 or 
> character entity references so all characters must be written raw 
> in some single-byte encoding.

Granted.  After some more experimentation I see what the relevant 
difference is between docbook-xsl-1.59.1 (with which I did not 
perceive a problem) and dockbook-xsl-1.67.0 (which triggered my 
reports of a problem).  The earlier stylesheets silently 
discarded characters, such as left and right double quotes, which 
could not be represented in the output encoding for the htmlhelp toc 
file.  With the new stylesheets, you get an error message regarding 
such characters.  The new behavior seems right.

As I mentioned in a previous post, I am working around this by 
outputting utf-8 then using gnu recode to turn the toc into windows 
cp1252 (this circumambulation being required since saxon 6.5.3 does 
not support cp1252 output).

In the process I have discovered a bug in gnu recode version 3.6 (or 
libiconv): recode produces corrupt output from the u8..cp1252 
conversion unless supplied with the -x: flag, to prevent it from 
using libiconv.

Allin Cottrell