Re: LYNX-DEV did something happen to <img alt="xxx"> ??

[Klaus Weide]

On Tue, 7 Oct 1997, Nelson Henry Eric wrote:

> I still don't know what is what, but three types of encoding are now
> available in "http://163.51.110.11/lynxdev/SJISvISO2022JP.html"; and
> "http://163.51.110.11/lynxdev/WITHMETA.html";.  I created what I THINK
> is SJIS this time by editing with my terminal emulator set to write
> "SJIS", and tried to encode EUC by setting my terminal emulator to
> write "EUC".  

You seem to have got it right - otherwise the three strings wouldn't
appear as the same characters at least in some situations.

> EUC turns out to be what I called SJIS yesterday, so
> you were right!
> 
> The META tag seems to interact with what is labeled SJIS when it is an
> alt="", but not when it is plain text.  The plain text, which is a
> mixture of the encodings each separated by a single space, looks the
> same and is rendered into the correct Japanese on either page.  Without
> a META tag, all three encodings look the same and are the intended
> Japanese characters in the alt strings.  With a META tag (WITHMETA.html),
> the SJIS appears as four ascii letters: "Ew_w", not as Japanese, in the
> alt="", whereas the other two encodings are correctly rendered.

However: do you have any older version of Lynx which behaves different for
this exact same WITHMETA.html?  It seems to me Lynx behaves in this case
like it did before.

Now for the fun part:  your "dreaded SJIS" ALT text is the very first
thing in the document that gets displayed.  Make it so that this isn't the
case.  For example, put an ASCII character before the "<img".  You'll be
surprised...

Also try to make your non-ALT SJIS text the very first text data in the
file.  Also see what happens when you omit the spaces in your plain text
mixture.

Again please check whether any previous incarnation of Lynx behaves
differently.

The reason for what you'll observe seems to be that the
Japanese-encoding-guessing forgets what it has figured out each time when
a ' ' (SPACE character) is encountered (not in the input document, but in
the rendered text).  Before the very first ' ' the charset parameter has
an effect, after that it gets forgotten.

I don't know what is the best way to change this - does it need fixing?
If yes, should Lynx (for Japanese) pay more attention to what character
encoding is implied by the META tag, or should it completely ignore it?

   Klaus


;
; To UNSUBSCRIBE:  Send a mail message to address@hidden
;                  with "unsubscribe lynx-dev" (without the
;                  quotation marks) on a line by itself.
;