[rfc-i] On backwards compatibility for v2

Nico Williams nico at cryptonector.com
Mon Feb 10 14:29:02 PST 2014

On Mon, Feb 10, 2014 at 4:15 PM, Ted Lemon <mellon at fugue.com> wrote:
> One important reason why v2 compatibility is better achieved through translation is that it allows workarounds to be translated into idioms, rather than requiring that workarounds be preserved intact.


We're using XML, which means that XSLT is *the* language of choice for
writing converters between similar schemas.  Many are already using
XSLT likely without knowing it (e.g., via Greenbytes' converters).
Anyone who has written even a modicum of XSL will agree that a
converter between the v2 we know and the v3 we're sketching is utterly
doable (and in all likelihood simple).

One effort (backwards-compat w/o a converter) means tossing lots of
special cases into xml2rfc.  The other means one special case: if v2,
convert first.  The converter can be developed in parallel or later,
and it can be crowd-sourced (e.g., I might volunteer one).

If we trust that we're using XML for its power then we should opt for
conversion over backwards-compatibility.  Else one has to wonder why
bother with XML.

I'll grant one argument: ease the user learning curve by letting them
mix v2 and v3 content.  That might yet be a winner of an argument,
though I doubt it.


More information about the rfc-interest mailing list