[Top] [All Lists]

Re: Q+A: Mnemonic to Proposed Standard

1992-04-03 01:24:47
(If the Informational document is on some other standards track, then
IETF rules of maturity apply.  For proprietary stuff, such as NFS,
the rule does *not* apply.  It is my impression that EUNet is not
a standards-making body, but rather its comparison to UUnet suggests
that it be viewed as a vendor.)

822ext people, I have no idea whether EUnet and UUnet can be
meaningfully compared to each other, but, in my opinion, NFS and
Keld's "charsets" document cannot be compared. NFS was simply a brand
new protocol, and there were few items in the specs that could be
challenged on accuracy grounds (though people may have had other
reasons to complain -- that's a separate issue). [I'm not interested
in replies to this paragraph.]

What bothers me is that Keld's gigantic document contains many tables
of existing character sets, many of which have already been officially
registered with ECMA according to ISO procedures, and some of which
are proprietary standards.

Essentially, Keld's work is a duplication of the information already
contained in other documents. My concern is whether this duplication
is in fact faithful. I have no doubts about Keld's intentions (they're
good), but work of this size is bound to have bugs in it.

If the work simply involved copying each character's name from the
official document to a table in Keld's document, there is some chance
that it would be correct, but it would still have to be checked of

But if I am not mistaken the work involves more than simply copying
character names. For each character added to the tables, you need to
check if that character has already been added to some other table,
and give it the same name (i.e. the two-byte mnemonic).

Now _that_ is where the problem comes in. How do you decide whether
any particular pair of characters are the same? Are they the same if
their names in their respective official documents are similar? Are
they the same if they are graphically similar?

What Keld has done is to unify the characters from the various
character sets, very similar to Unicode's work. A heck of a lot of
work. (Wow, Keld!)

But now tell me -- is Keld's document the sort of document that the
822ext WG can put its stamp of approval on?

I'd say "No". Many of the people in this WG are relatively clueless as
far as character sets are concerned. (I am one of them.) This WG would
be slapping the true experts in the face if it approves the document
for publication as an informational RFC. It would be presumptuous to
say the least.


PS  I find it amazing that this situation can even arise in the IETF.
Procedures are extremely loose. But that's why it's fun too!

PPS  If this WG "recommends" Keld's document, and the RFC editor
publishes it as an informational RFC, you can bet on it that Keld will
make that fact known to various other working groups, and use it as a
lever to get his work approved elsewhere.

PPPS  If "informational RFCs" are really not such a big deal, and the
RFC editor doesn't care about the accuracy of the information, and not
many other people care either, then I take it all back, and you can
ignore this message. (!)

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