[rfc-i] On backwards compatibility for v2
dhc at dcrocker.net
Mon Feb 10 16:08:20 PST 2014
On 2/10/2014 3:39 PM, Ted Lemon wrote:
> On Feb 10, 2014, at 5:28 PM, Dave Crocker <dhc at dcrocker.net> wrote:
>> You really don't see the basic flaw in your logic, from sentence 2 to sentence 3? Really?
> What, that we can ensure compatibility in the implementation through hard work and careful testing? Sure, but we can also ensure compatibility in the implementation of the converter in the same way.
And I really suck at formalisms, but still...
1. "My point is simply that keeping backward compatibility in the
grammar does not ensure backward compatibility in the implementation."
2. So the upside to keeping backward compatibility is questionable
The assertion in the first sentence is not that we can't get
compatibility or even that it is difficult. It's that doing it one
place doesn't "ensure" that it will (automatically) occur in a second
This makes no reference to the actual difficulty of getting it in that
second place, or what might cause incompatibilities. It also says
nothing about the value of compatibility.
So the conclusion in the second sentence has almost literally no basis
from the previous sentence.
Unless you can explain the justification for the general form:
"IF doing A does not guarantee B, then there is 'questionable
benefit' in seeking A & B."
which I am seriously not seeing.
More information about the rfc-interest