ietf
[Top] [All Lists]

Re: Why XML is perferable

2001-02-23 03:10:02

In message 
<451D45016C2CD2119DA50000F8FE7F0705D8EF5E(_at_)mlngcbnt01(_dot_)hc(_dot_)bt(_dot_)com>,
 gra
ham(_dot_)travers(_at_)bt(_dot_)com typed:

Let's consider a few basic principles.
 
ok - lots of good points below - a few responses...

1.  Neither ASCII nor XML are ever displayed.  They are CODES for
representing characters in a computer. It is the CHARACTERS ( glyphs ) that
are displayed ( presented / rendered ). There is a mapping between the codes
and the glyphs.

but the glyphs are in HARDWARE in many devices(e.g. printed on
keycaps, in printer wheels, in crt display chips etc)...

2.  ASCII has a strictly limited set of characters and glyphs ( even the
"international" version ), which can not represent many languages in the
world, and does a poor job of rendering diagrams, pictures, etc.


yes, this point has been made a lot - however, the discipline of
getting a diagram into ascii art has OFTEN caused people in the ietf
to udnerstand the problem better (e.g. by choosing the most
parsimonious topology to explain a partiocular routing problem)

3.  As some people have emphasised, the importance of ASCII lies in the (
American Standard Code for Information ) INTERCHANGE.  Interchange implies
the ability to transfer in a manner which can be understood by both parties
to the transfer. The MOST COMMON global method of transferring will be the
most effective.

yes, yes, and yes......but also, collating, indexing, and searching -
manmy of the search engines are optimised to the roman alhpabet, the
english dictionary, and the english freqeuncy distribution of
words....

4.  Interchange does not guarantee understanding - either of presentation
format or content.  I wouldn't like to have to deal with Einstein's Theory
of Relativity ( content ), especially in Chinese ( format ).  ASCII does not
interchange Chinese characters, so it's presentation format is NOT readily
understandable by "most people".  

5.  A more comprehensive coding scheme, such as the Universal Character Set
( ISO 10646 ) would allow many more characters and glyphs to be used.

6.  The key to usage of encoding schemes is how widely they can be
interpreted by character presentation ( or rendering ) applications ( word
processors, etc. ), in mapping the internal codes to the glyphs rendered on
the screen or on paper.  Applications which can render more characters would
allow the use of larger code ranges and more characters.  

Until something replaces ASCII as the most commonly available global
interchange format ( and could it be HTML / XML ? ), it will remain the
default.  That doesn't mean that we should just accept it for evermore.  If
that principle were followed, we would still be drawing on cave walls and
large red rock formations ( Ayres in Australia ! ), which are not very
transportable !  

One of the things that the IETF could, and in my opinion SHOULD, do it to
make its documents available in several presentation formats, not to say
languages.  Yes, we would still need a master copy and format, which could
be ASCII, but other, more presentable formats, would make life easier for
many people.  The ITU-T ( I'm sorry to mention it, but they have been doing
this for decades ) publishes its documents in three languages. If the IETF
is really working for the world, it should take a more global view and
consider a similar sort of policy. Don't we have a work stream on
internationalisation ?

Of course, this sort of effort costs money - lots of it.  That's why the
ITU-T charges for documents.  If you want it free, you take the IETF
approach and get the inexpensive, ASCII, American language version.
 
thats why the ITU claims it charges. i think you overstate the
contrast. btw, as someone who has written documents in english english
for 20 years using ascii codes, i dont see your point about American
_language_ - coding for alhpabet doesnt necessarily code for language
(ever used greeklish?:-)

anyhow, the point about cost is good - basically, do people want to
think about a funding model for multi-lingual internet standards...?
worth a brief discussion (there are alternates to the ITU charging
model, clearly)

j.



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