Quoting John Hughes (johh.hughes@???):

> The major advantage of PDF over PostScript is that PDF is not Turing
> complete.

Now that you mention it, I _vaguely_ recall hearing something about
that. I also recall that the Turing-complete nature of PS is why the
GhostScript utilities have a -safer option.

Thus, I sit corrected: PS and PDF are _almost_ completely bidirectionally
convertable without lossage.

> (Of course PDF has now been fucked up by the inclusion of JavaScript,
> but that's an other story...)

ISTR that this is not actually a problem internal to PDF, but rather is
fallout from Adobe Acrobat Reader (aka 'acroread', aka 'acrocrud') being
willing to process, unless you uncheck a checkbox in Preferences, a
Javascript wrapper sent along with a PDF file. I might be wrong. The
method of embedding or sending the JS is unclear to me.

In any event, this threat model can be defeated by never, never, never
allowing acrocrud to process public data -- and I would suggest a
root-and-branch approach, to the effect that 'Friends don't let friends
install or use acrocrud'. (There is an arguable niche role for PDF
forms with JS, e.g., fill-in PDF forms that enforce type and length
restrictions on entered data for each field. OTOH, even on benighted
OS platforms such as MS-Windows, there are non-PS-supporting and safer
PDF-reading utilities such as the open source utility MuPDF.)

> By the way, when will anyone point out that  printer-driver-hpcups
> might be enough for printing to non-ps HP printers and it doesn't need
> d-bus?

I _did_ already respond to someone suggesting that a printer supporting
either PS or PCL would be suitable, so the latter AFAIK is what you are
speaking of.