Re: lynx-dev Comparing html rendition of lynx and explorer

[David Woolley]

> Front Page generates pages with code page 1252 (Windows ANSI) coding, not
> ISO-8859-1.  Set display character set to 'WinLatin1 (cp1252)' to view
> these characters.

Display character set should always reflect the true local display 
character set.  The server++ should identify the document character set.
So for a valid document encoded in cp1252, there should be no need to
fudge Lynx settings.  If one has to work round broken server
configurations, it should be the assumed document character set, not
the display one that is faked.

However, the real problem with Front Page is that, early versions, at
least, generated numeric entity codes between 127 and 159.  Numeric
entity codes are *ALWAYS* in Unicode, never the local code page, so these
codes are unassigned and the page illegal (e.g. see the mdash thread).

++ The current Microsoft approach is to use META for this purpose and 
have the browser switch character sets on this basis.  However, the
official HTML position is that such use of META is intended purely 
for the server.  Misuse of this feature for setting HTTP headers in 
the HTTP body causes problems for caches, which are content transparent.

One reason for doing this is probably bundled web space, where the
content providers can't control the server configuration.  Another reason
is plain ignorance that HTTP exists.