Re: lynx-dev HTML4.0 and default charset

[Leonid Pauzner]

1-Mar-99 08:28 David Woolley wrote:
>> The most logical way to have servers configured properly :-)
>> IMO, charset parameter should be always specified when document
>> have characters >128. Unfortunately, http/html specs keep compatibility

> Unfortunately this is unrealistic because:

> 1) Many web server operators don't understand such things (the web
>    server is a way to make money/retain customers, not something to
>    understand); ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
     ^^^^^^^^^^
Yes, this is a generic problem :-)

> 2) Most content providers don't understand such things, even when they
>    do have access to meta data (after all, it works on their GUI
>    browser);

> 3) Many, if not most pages, by volume, are on some sort of "free" space
>    where users have no access to meta data;

> 4) Although META headers are supposed to control the server, I have
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^???
no, they are for clients.
>    seen little evidence of this, and a lot of evidence that they
>    are controlling the browser instead - in any case a variation of
>    item 2 applies, as these are normally set to the Windows Western
                              ^^^^^^^^^^^^^^^^^^^
>    European proprietory character set, by the authoring tools, even
                                         ^^^^^^^^^^^^^^^^^^^^^^
The authoring tools SHOULD write a charset header automatically
(by transferring info to the server via FrontPage upload protocol
or directly in HTML code via META header), but they are always broken
to have this behaviour as default...  So read 1) and 2) in brackets.

>    if ASCII or ISO 8859/1 would have been sufficient (I presume
>    Japanese versions of Front Page use a different default).

>> with very old browsers which does not understand charset parameter
>> and even can not ignore it (and such browsers are obviously broken

> Such browsers, apart from HTTP 0.9 browsers, have always been broken.

>> for any modern HTML in any case). That is the root of the problem.

> You are shooting yourself in the foot here; most content providers would
> probably define modern browsers as NS 4.x or IE 4/5 with JavaScript,
> Java and probably Win32 ActiveX turned on.  They certainly wouldn't
                                              ^^^^^^^^^^^^^^^^^^^^^^^
sure, but since we talk about charsets no problem with lynx:
its charset handling just superiour (2.7.2 and up).
> include Lynx; Lynx survives at all because some people follow the spirit
> of HTML and write universal HTML.