[pdftex] Another request for an enhancement to pdf(La)TeX
jjq at galcit.caltech.edu
Wed Oct 27 17:39:11 CEST 2010
On Wed, 27 Oct 2010, Martin Schröder wrote:
> 2010/10/27 James Quirk <jjq at galcit.caltech.edu>:
> > It's worth noting that while the PDF spec stipulates the xref table appear
> > at the EOF, the PDF Reference suggests PDF viewers should be able to cope
> > with the table appearing anywhere in the last KByte of the file.
> I find this in the PDF Ref (H.18), but not in the norm.
> It makes sense, though: http://en.wikipedia.org/wiki/Robustness_principle
Yes. It also allows some practical applications. For instance, you can
use GPG to produce a detached binary signature for a PDF and then prepend
the signature to the file, so that's there's only one file to keep track
of. Alternatively, you can craft a PDF, viewable by AR and xpdf, which is
also a valid Perl, Python, Ruby script. Why would you want to do that?
Well, with /RichMedia annotations it's now possible to talk to a localhost
server and in some instances it makes sense to have the server unpack
itself on-the-fly from the PDF. Of course to do it properly, you really
want a wrapper to AR that provides security mechanisms to safeguard
against trojan horses, as well as operator error. And said wrapper can
always stream edit a file, so the present latitude is not really needed.
Nontheless it's worth noting that it's now possible to add an interactive
TeX widget to a TeX-genertaed PDF document. The one I have works much like
Mathtran, but with scalable fonts. It's even possible to add a Bash widget
and run a vi session from inside a document. This opens up the possibiites
of authoring manuals where the interested reader is led through an
actual terminal session, say for R or Octave.
It's also worth noting that Adobe announced Adobe Rome on Monday, see:
http://rome.adobe.com/index.html And if you read between the lines, it's
clear that more and more interactivity will find its way into PDF/AR. For
instance, at present /RichMedia annotations have to be anchored to a page.
But the logical progression will likely see them anchored to the document.
Then it's possible we will see folder-level SWFs, analagous to the current
The reason for bringing all this up, is that when making feature requests
for pdftex, and family, requesters should be looking at the bigger
picture, such as where is PDF heading. They should also, unless it's a
commercial secret, state what it is they want to achieve, conceptually,
and not dwell on the mechanics of their chosen implementation path, which
may actually be off beam. That way third-parties have the opportunity
to point out alternative approaches, which may be either more tractable
or more viable over the long term.
More information about the pdftex