[Top] [All Lists]

Re: mime formats and versions in format specifications

1992-04-02 01:46:45
What it seems is going on is that people disagree in this forum about
the proper role of versions and format management, 
   Actually, I don't think very much, but it is possible...
but, in the
interest of getting something out the door, the disagreement is being
swept aside. 
   Not my reading at all.

I guess that's OK with me too, as long as the issue is
actually resolved before this thing actually becomes an internet
standard -- resolved because the spec is ambiguous and the issue needs
to be addressed, independent of whether initial testers of the spec
think it is a problem.
   I think the disagreement, if there is one, is right here.  Let me
suggest that all complex specs are ambiguous (except those that are
written in code s.t. the "spec" is "anything that works like this is
correct"--those have other problems, but, if one believes that all side
effects are intentional, ambiguity isn't one of them).  The key question
isn't "ambiguous" versus "not-ambiguous", it is whether the level of
ambiguity is tolerable, e.g., whether multiple implementations to the
same spec, written within the variance range permitted by the putative
ambiguity will interoperate to a satisfactory degree.  
  Now that sentence is filled with fuzzy language in its use of words
like "tolerable" and "variance range" and "satisfactory".  They can be
pinned down only by experiment and operational experience: otherwise, we
just have a hypothesis by one group (including you) that the answer is
"not a chance without additional clarification", a hypothesis by another
group (including Ned) that the answer is "within the bounds of what can
practically be done either internal or external to Postscript, what is
there will permit satisfactory interoperability", and an opportunity for
a small-scale religious war and a lot of wasted bandwidth.
  So I think the issue isn't one of "sweeping aside the disagreement" as
much as it is of trying to deploy some implementations to see where we
get into trouble and have to narrow things down.

  Note that my "it would be a good idea to tighten the text up... later"
hypothesis is mostly orthogonal to the above, since one can entertain
that belief regardless of whether one things the present text is
actually "ambiguous" or just a small trap for the unwary and/or future
would-be extenders and/or IANA, etc.

<Prev in Thread] Current Thread [Next in Thread>