[EXT] Re: l3backend: adding new DVI specials?

Zdenek Wagner zdenek.wagner at gmail.com
Fri Feb 14 14:08:26 CET 2020


pá 14. 2. 2020 v 13:47 odesílatel Philip Taylor <P.Taylor at rhul.ac.uk> napsal:
>
> Zdeněk Wagner made three different points :
>
> That a universal special would be equivalent to developing a new language
>
> That different drivers require different specials to be inserted at different points in the source code
>
> That a special that cannot be handled by a particular driver should result in a warning.
>
> I will address each in turn.
>
> I do not agree.  The underlying languages remain unchanged — all I am proposing is that the \specials become polyglot, and add a tag identifying the language in which each component is expressed.
> Is easily addressed by including all necessary specials in the source document at the appropriate places; a driver would then interpret only those elements of each \special that are targetted at that particular driver.

No, It would IMHO lead to even higher confusion. Now we have just one
type of PS specials, with this suggestion we will need two:

a) the one understood by dvips but not xdvipdfmx
b) the one understood by both

Now we have only one PDF literal, after your suggestion we will need three:

a) one understood by pdftex (and luatex) but not xdvipdfmx
b) one understood by xdvipdfmx but not pdftex and luatex
c) one undestood by all of them

In addition, the authors of the macro packages will be forced to learn
all possible \specials for all possible drivers and use them.

Selecting a correct backend when loading a macro package and
documentation which is the default is IMHO the right way.

> Is the crux of the matter, but directly conflicts with David's point about regression testing.  Therefore we need to add one further \special, at the very beginning of the document, to indicate whether or not subsequent \specials should lead to warnings if not understood by any particular driver.
>
Several drivers already have a command line switch for disabling the warnings.

> But perhaps the last is all that actually need to be implemented — a universal \special, understood by all drivers, that indicates whether or not a warning should be issued when an unrecognised \special is encountered ...  Turn on for regression testing, off for production.
>
> ** Phil.


Zdeněk Wagner
http://ttsm.icpf.cas.cz/team/wagner.shtml
http://icebearsoft.euweb.cz



More information about the tex-live mailing list.