[rfc-i] format and interoperability

[Andrew Sullivan]

On Mon, Mar 26, 2012 at 01:50:10PM +0200, Dave Crocker wrote:
> 
> ^L is for line printers.  No, we don't use those anymore.  But we do
> use printers.  That is, there is still utility in have page-oriented
> constructs. (But notice, please, how abstract the word "construct"
> is)

Right.  My point was simply that (upthread) someone was arguing that
the poor printability is a failing of the printers.  That is, bluntly,
preposterous.  If we believed that, we would be raising an accidental
property of a particular, historical printing system implementation
into something we expect to be a standard for our contemporary
printing systems.  Printing systems didn't (couldn't) make that
promise.
 
> The other ASCII-based characteristics were/are important not (just)
> for printers but for all rendering.

I agree, and I agree that it's a separate problem.  What is a very
nice feature about the historic line printer format is that it is not
extremely hard -- it's not no work, but not extremely hard -- to
extract unformatted ASCII text from it.  If _that's_ the feature we
want, I have no objection.  But let's be clear: the format as it is
doesn't do that without work; for instance, we have embedded white
space and line ends in the existing format.  (Indeed, that's part of
the problem in cases like epub.)  

To be a little more constructive: one key feature we ought to be
looking for is a way of separating presentation format from content,
while still having a canonical (presentation?) format so that things
like references to particular text can still be made stably.  A second
key feature should be a lowst-common-denominator content format, so
that it is possible to reuse some of that content trivially.

> The primary point that has been fundamentally irritating and
> embarrassing for perhaps 20 years is the lack of support for
> non-ASCII characters.

That is not a primary point for me.  It's a serious irritation, and I
would dearly like to fix it.  

The fixed-page issue, however, has been problematic for me almost from
the moment I started paying any attention to the IETF (I had a
PalmPilot before I knew how to spell "IETF", for instance).

Best,

A

-- 
Andrew Sullivan
ajs at anvilwalrusden.com