The date of mathematical works is hardly relevant. Are you aware of any publications in the last 10 years that are crucial for people to know about, or that otherwise supersede material here?
The defacto standard workflow in CS academia is still to produce a dvi file from latex sources and then generate postscript from it. I think you will remember that unlike postscript, pdf became open only in 2008. You can of course generate pdf directly using pdflatex, unfortunately it doesnt work with another defacto standard for vector graphics "eps".
What I find unusual is that acrobat-reader and some other pdf viewers do not attempt to display postscript. A big part of pdf is really postscript underneath, a pdf viewer already knows how to display a large subset of postscript.
Proof: This claim is obvious from the algorithm. Every time it makes a mistake, the chosen block will be demoted.
Proof by inspection, my favourite.