On 12.11.2007 11:13:52 Vincent Hennebert wrote:
> Hi,
> 
> > Author: jeremias
> > Date: Mon Nov 12 01:40:16 2007
> > New Revision: 594067
> > 
> > URL: http://svn.apache.org/viewvc?rev=594067&view=rev
> > Log:
> > Avoid null values in generated Font classes so the encoding can be 
> > inspected.
> > Don't warn about missing hyphenation characters for fonts such as Symbol 
> > and ZapfDingbats which don't have the default hyphenation character.
> 
> Is this really necessary? 

I think, yes.

> If hyphenation must occur within a block with 
> the Symbol or ZapfDingbats fonts, doesn’t that mean that something is 
> wrong somewhere, and may lead to an output unexpected by the user? 

Not necessarily. Maybe someone likes to decorate his documents with lots
of characters from one of two fonts. :-)

> Moreover I can’t see how hyphenation can be performed with ZapfDingbats 
> (and even for Symbol, although that might be discussed), simply due to 
> a lack of hyphenation patterns.

Sure. That's why I actually chose this path. I didn't want to make
LineLM more complicated by handling the differences there. Ultimately,
the hyphenation character will probably never be used when one of those
two fonts is active.

> The encoding-agnostic version from the revision just before actually 
> looked fine to me.

There was one problem: You get lots of meaningless warnings about the
hyphenation char being substituted. If I just switch of the warning
completely, I'm back where I started. Having a "cannot map character"
when using a Symbol font loaded from a Type 1 font which is not very
helpful.

The way it is now I get properly warned with a meaningful message that
the hyphenation character I have chosen is not available for the current
font. And I don't have to manually redefine the hyphenation character in
FO each time I use Symbol or ZapfDingbats.

> WDYT?
> Vincent


Jeremias Maerki

Reply via email to