[tex-live] Bad .cfg files in TeX source tree?

[David Kastrup]

Heiko Oberdiek <oberdiek at uni-freiburg.de> writes:

> On Fri, Jan 19, 2007 at 03:45:18PM +0100, David Kastrup wrote:
>
>> "Zdenek Wagner" <zdenek.wagner at gmail.com> writes:
>> 
>> > 2007/1/19, David Kastrup <dak at gnu.org>:
>> >> Jonathan Kew <jonathan_kew at sil.org> writes:
>> >> ...
>> >
>> >> I think this is too complicated.  Packages should just check whether
>> >> \pdfpagewidth is defined and set it if it is.
>> >>
>> >> Whether the actual engine is PDFTeX, XeTeX, Aleph or whatever else is
>> >> rather irrelevant.
>> >>
>> > If \pdfoutput=0, \pdfpagewidth is available but has no effect.
>> 
>> Setting it would nevertheless not do any harm.
>
> It does harm:
> * Who knows, how future versions of pdfTeX behave? It is reasonable
>   to issue a warning or error if \pdfpagewidth and \pdfpageheight
>   are used in DVI mode.

No.  I don't see appropriate prior art for "barf on assignment".  That
is no sensible behavior.  There are a few readonly variables in TeX,
but they have sensible valued when read out.  Reading those variables
out in DVI mode could not deliver useful values.

So there are only two remotely useful behaviors: make the control
sequences undefined, or make them just hold a value that is ignored.

> * The package thinks the setting of the page dimensions is
> successful without a chance for doing it right or issuing a
> warning/error at least.

The package does not have a chance for doing it right or issuing a
warning/error itself if TeX decides to barf-on-assignment all by
itself.

-- 
David Kastrup