> > If this growing and distributed burden is more attractive than a 4-byte
> > naming convention that doesn't interfere with existing processing, then I
> > suppose we should drop the suffix, and find out how popular this
> > non-approach proves to be in a couple of years. At that point, it will be
> > very difficult to fix things. I continue to argue that the suffix is a
> > remarkably low-cost solution with significant benefits.
>Well, you've just convinced me. I hereby retract my assertion that
>content sniffing is a "mostly harmless" but partial solution to this
>problem. In light of this it doesn't look like a solution at all.
Ned, I remain unconvinced. "Growing" here means probably about one new type
every week or so.
Speaking as the media type reviewer, the run rate has proved to be quite
a bit higher than this. And I expect it to grow a lot in the future.
This greatly reduces the power of the word "burden".
Remember, this is only of interest to specialized systems, that is, the
ones who want to pass random (that is, unknown format) XML to a generic XML
parser. Yes, that's "distributed"... over a very small number of systems
that wan to do this. Even if it is many systems, the cost of sniffing one
new document type a week is incredibly low, and is only incurred in systems
where there is not a human who updates the type-to-translation table.
But the corresponding costs of the naming convention are incredibly low.
So, even if the utility is limited (and I'm growing convinced that it isn't
as limited as I first thought), the cost-benefit conclusion seems clear.
Ned