Re: postscript vs PDF, Call for Community Feedback: Retiring IETF FTP Service

[John C Klensin <[e-mail]>]

--On Saturday, November 28, 2020 17:17 -0500 John Levine
<[e-mail]> wrote:

> Even ASCII has its bitrot aspects. As I'm sure you recall, the
> character ^ that now looks like a circumflex sometimes used to
> be an uparrow, and _ which now looks like an underline was
> often a left arrow.  But we seem to be OK with old ASCII
> documents anyway.

IIR, not quiet, but I don't have time to dig out my copies of
the first and second versions of ASA/.../ANSI X3.4 and we are
already more than far enough off-topic.  

Where I'm confident we can agree is that it is wiser to rely on
a stable page image format (like PDF and ideally with no
external references) than on a print specification format (like
Postscript) to be sure that what we are seeing today and what
the author saw are the same.  I think we also agree that,
between Postscript and PDF, the differences are likely to be
small (you point out the possible font issues with the former
and the possibility of external references as well as font
issues are why the effort that produced the PDF/A specifications
came into being.  I presume we would also both agree that either
one is more likely to reproduce what the author saw than almost
any sort of textual markup language in the absence of external
specifications or assumptions or of a simple stream of ASCII
characters (given that ASCII, like Unicode, associated abstract
characters and not particular glyphs with code points/values).

The question I've been trying to get at, and won't have an
answer to until we hear from Tim, is whether what he meant by
what "the author intended" is actually what you are describing
or whether he was really talking about intent.  I think what you
are talking about would be closer to "what the author saw when
the document was first made available".  Even then there is a
sometimes-significant difference between some original RFCs and
the results of the conversion project to get RFCs into
machine-readable format.  For some of the RFCs covered by that
project, that project lost assorted appendices and illustrations
that the authors certainly intended to be present.

So, again, I'm not sure exactly what we are talking about -- and
even, given Julian's question, whether we are talking about RFCs
in the RFC Editor repository, anything else the RFC Editor may
be mirroring, or I-Ds, RFCs, and other materials in IETF
repositorites.

    john