Andrew Douglas Pitonyak wrote:
A reference implementation is used as a definitive interpretation for
that specification. This helps to discover errors and ambiguities in
the specification. It is very difficult to fully specify anything, and
at least with a reference implementation, you know who is correct.
I frequently see web pages rendered differently depending on the web
browser used. Which one is correct and accurate? A reference
implementation would answer that question immediately.
It is clearly important to have a fully and correctly specified
system, but that does not remove the helpfulness of a reference
implementation.
I only vaguely remember the precise behavior that you use to claim
that a destructive edit exists. Did I consider it a bug? Yes! Now,
which behavior is correct?
In this context, however, the first issue of discussion seems to be
the ODF file format and how it should be rendered. We only care that
it is always and consistently rendered exactly the same. If I save a
file, reopen the file, and the rendering has changed, well, where is
the bug? Is the data properly rendered? Was the data changed on write?
A reference implementation would help sort that out.
I feel that my point has been missed. With regards to two web pages that
are rendered differently, assuming that the markup is valid and to spec,
the answer to the question regarding which is correct is both of them.
My point is that the ODF spec should be like proper use of XHTML strict;
a rigid and well-defined set of semantic rules with a 1:1 mapping to
document elements. Were that the case, it wouldn't matter at all if the
documents were *rendered* differently, as editing them in different apps
would not result in any semantic irregularities. If I were working on a
document in OOo and my colleague was working on the same document in
Abiword, if the ODF specification was properly abstracted, neither of us
would be any the wiser that the document was rendered differently to the
other.
- Naz.
--
السلام عليكم
Web: www.mrnaz.com
Ph: +61 400 460 662
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
