I can understand this. I have already fought exactly the same battle
in HTML-WG. I proposed using UTF8 and UCS2 as the lingua franca for
the WWW.
I should note that I expect a free {whatever}->Unicode module to
appear at some point in the future (I have plans to write one, as do
others), which would be based on loadable conversion modules. This
means that adding support for a new charcater set and/or encoding
would be just a matter fo getting the module and registering it.
>Can we define the set supported (at least for VRML 1.1)?
Possibly. I am currently looking into this. Note: as I said before, I
think UTF8 +ISO 10646 should be strongly recommended.
>What if we instead tag the beginning of the file with the encoding?
Yes, that possible, but what say you want to put Japanese in SJIS in
one node, Japanese in EUC in another, and Chinese in BIG5 in another?
The only way to achieve comparable functionality would be to use
Unicode.
>This would make it easier for use in editors that produce a specific
>encoding.
Almost all encodings are ASCII compatible...
>Not only would it be easy to enter text strings in the native
>encoding, but node labels and comments could be in it as well.
Sure, and this might be nice. Again, I proposed using something
similar for HTML (called the ERCS, which was designed by Rick
Jellife).
There is a strong argument to be made for allowing native names for
objects, and in general, I sympathise. However, given that most
languges have an ASCII-compatible representation, I do not think it is
an overriding necessity.
Again, I will say that I think using ISO 10646 and UTF8 (in
combination with language tagging), to be a great answer to 95% of the
worlds text-processing needs, but that is just *my* opinion; I
designed the nodes with the opinions of *others* in mind, as well as
my own.