User-Agent: Thunderbird 2.0.0.23 (Macintosh/20090812)
MIME-Version: 1.0
References: <19189.17593.814557.704925@morse.mittelbach-online.de>           
            <4AF5A79B.9040003@residenset.net>           
            <19189.49743.74826.581704@morse.mittelbach-online.de>           
            <4AF6DC52.2030401@residenset.net>
            <19192.40419.459314.486313@morse.mittelbach-online.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 8bit
Message-ID:  <4AF94FFF.2090004@residenset.net>
Date:         Tue, 10 Nov 2009 12:35:27 +0100
Reply-To: Mailing list for the LaTeX3 project
              <LATEX-L@LISTSERV.UNI-HEIDELBERG.DE>
Sender: Mailing list for the LaTeX3 project <LATEX-L@LISTSERV.UNI-HEIDELBERG.DE>
From: =?ISO-8859-1?Q?Lars_Hellstr=F6m?= <Lars.Hellstrom@RESIDENSET.NET>
Subject: Re: an object type for heading commands
To: LATEX-L@LISTSERV.UNI-HEIDELBERG.DE
In-Reply-To:  <19192.40419.459314.486313@morse.mittelbach-online.de>
Precedence: list
Status: R

Frank Mittelbach skrev:
> Lars,
[snip]
>  > Depends on whether you look at it graphically or semantically. For 
>  > designing a general API, I think you really need to look at it 
>  > semantically.
> 
> yes agreed, but we can only go semantically up to a certain level, otherwise
> you need to specify 100+ arguments and still haven't covered all
> possibilities.
> 
> what i'm currently thinking might be reasonable is
> 
>  - subtitle (which is text that is semantically related to main title)
>  - motto/quote text (text which is semantically unrelated to main title)
> 
> and support both. 
> 
>  - text credits (which is author copyright or similar info but not
>                  semantically specified further)
> 
> if more control is needed then that type is not specialized enough to support
> it.

This suggests you're open to using a different object type in some 
cases? Indeed, that's not a bad idea: A document class with _very_ 
special heading needs can be expected to use its own object type for 
headings (in addition to the basic definition of instances, this would 
require redefining all sectioning commands and defining new templates 
for the new type, but it is perfectly doable). Packages meant to 
configure standard headings will then not work off the self for this 
class, but it is still as configurable as the standard headings.

If we're clear that completely bypassing a subsystem such as the 
heading templates is OK also in LaTeX3 (though this comes at the cost 
of having to reimplement a bunch of stuff yourself), then it's not the 
end of the world if there are things that the standard heading object 
type cannot accomplish. It's better that it does 90% of all things well 
and elegantly, than that it does 99.99% of all things with an API only 
wizards can comprehend.

Lars Hellström