On 11/26/2011 11:51 AM, John R. Levine wrote:
I gather that you consider ECMA-376 and ISO/IEC 29500 formats to be
In this case, we've seen references to /continuing/ interoperability problems
when trying to use docx.
I wouldn't disagree, but if we mean easy to interoperate, let's say so.
Word 97-2003 format is totally proprietary, but is now sufficiently widely
reverse engineered that it interoperates pretty well. On the other hand, ODF
format is well documented and interoperates well among different software that
support it, but since old versions of Microsoft Office don't support it, we get
For a production requirement, such as being discussed here, the requirement
should only call for use of extremely well-established data representations,
where 'extremely well-established' means highly stable and massively widespread
for a significant number of years. We expect our own use to be for many years
and adopting something that is still in the flush of transition is extremely
unwise. Equally, the extent to which we worry about archaic software is
relevant, but can be marginal for specific cases.
The world being imperfect, interoperability will never be perfect. So we have
to look at the degree of it that exists and decide whether it is enough.
For example, do we want to worry about packages that are perhaps more than 15
years old and don't support a particular representation? Is there enough use of
such old software to be a real concern? "Enough" is, of course, the critical word.
Ietf mailing list