[rfc-i] On backwards compatibility for v2
nico at cryptonector.com
Mon Feb 10 14:11:25 PST 2014
On Mon, Feb 10, 2014 at 4:03 PM, Heather Flanagan (RFC Series Editor)
<rse at rfc-editor.org> wrote:
> On 2/10/14 1:53 PM, Nico Williams wrote:
>> If v3 will not be backwards compatible even if we intend it to be (for
>> all the reasons you state, with which I agree) then we might as well
>> disclaim compatibility now and lose the shackles of v2's schema.
> Well, there is a difference between bugs representing a break in
> backwards compatibility and a wholesale toss of the old vocabulary.
I drew no equivalence between the two. I'm saying that if the cost of
backwards compatibility is significantly higher than developing a
converter (or crowd-sourcing it) then better drop the backwards-compat
requirement. If experience (per-Ted's comments in this case)
indicates that meeting such a requirement would be difficult, might as
well not impose it.
>> It may be easier to work on the v3 processor and a converter in
>> parallel than it is to work on a backwards-compatible v3 processor.
>> Either way, leave it to the developers.
> The developers will need the requirements from the authors and editors.
> That's why we're having this conversation now, so we can hash out the
> detailed requirements and get the tool written.
I'm OK with (and proposing) us deciding to delegate this particular
decision (backwards-compat or not) to the developers. Surely it must
be OK to choose to explicitly not set some requirements in particular
cases. I'm not saying (nor would I) that we must always do that.
More information about the rfc-interest