Re: lynx-dev Lynx does not open some http sites, forces download

[Jacob Poon]

On Wed, 3 Feb 1999, Bela Lubkin wrote:

> Mykola Sereda wrote:
> 
> >         Why label the contents? Why should a user leaf through some
> > labels which have no interest to him? Such as, for me, some "Taiwanese"
> > or a pile of other nuisance, including "NeXT" and other choices? And
> > what will happen, when all the other deserving world cultures and
> > CharSets will be registered? The choices will go into some hundreds ... 
> > 
> >         Instead, the 8 bit texts should be transmitted or displayed
> > automatically without any bargaining. If I hit some Taiwanese site,
> > and the display would be incompehensible, what difference it makes
> > whether it is incomprehesensible because I do not know Taiwanese,
> > or whether on top of it, it is incompehensible to someone who unlike
> > me understands Taiwanese, the text being improperly displayed? If
> > someone wishes to read Taiwanese, or Ukrainian, or whatever, let him
> > take care of displaying Taiwanese, or Ukrainian - but let him receive
> > it integrally. 
> 
> You are ignoring the possibility of a user who reads *both* Taiwanese
> and Ukranian -- or any other two languages which require different
> character sets.  Such a user benefits tremendously from the automatic
> notification, HTTP server -> browser, which allows the browser to
> display everything in sane character sets.

And don't forget, when you are dealing with CJK documents, the same
language may (and does) have 2 or more common, and incompatible encoding
standards.  Chinese uses Big-5 and GB, Japanese uses JIS and EUC-JP, and
Korean uses KS and and GB (yes, GB, due to historical reasons).  And of
course, all 3 languages are part of Unicode (and its various derivatives). 

What does that mean?  It means there may be only a few people (at least it
seems to be the case in this ML anyway) who can read two '8-bit'
languages, but there are a lot more people who can perfectly understand
the same language encoded in different standards.  Therefore the need for
content-type negotiations is very important, for serving readers in any
language (even Klingon too :).

But Mykola has a point.  If content-type is somehow mis-negotiated, user
should have a choice for using raw 8-bit transfers (eg: force download,
'raw text' browsing).  However, pre-negotiated transfers and displays
should not be turned on by default, at least not until the users know what
they are doing.